summaryrefslogtreecommitdiffstats
path: root/xlators/debug
diff options
context:
space:
mode:
authorAnand Avati <avati@gluster.com>2010-09-03 13:59:48 +0000
committerVijay Bellur <vijay@dev.gluster.com>2010-09-04 01:45:21 -0700
commit8b9a144414162413a399d59975fe3b7347907f4f (patch)
treeb37a01b2c146b47a772dd3a7f4ce3563577923fa /xlators/debug
parentd838e4496dec6b8889de47dd0ee31c4899ec7407 (diff)
gfid: changes in mkdir() prototype to have params dictionary with uuid in it
Signed-off-by: Anand V. Avati <avati@blackhole.gluster.com> Signed-off-by: Anand V. Avati <avati@amp.gluster.com> Signed-off-by: Vijay Bellur <vijay@dev.gluster.com> BUG: 971 (dynamic volume management) URL: http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=971
Diffstat (limited to 'xlators/debug')
-rw-r--r--xlators/debug/error-gen/src/error-gen.c4
-rw-r--r--xlators/debug/io-stats/src/io-stats.c4
-rw-r--r--xlators/debug/trace/src/trace.c5
3 files changed, 7 insertions, 6 deletions
diff --git a/xlators/debug/error-gen/src/error-gen.c b/xlators/debug/error-gen/src/error-gen.c
index 955b197ed63..e7ed8c6a850 100644
--- a/xlators/debug/error-gen/src/error-gen.c
+++ b/xlators/debug/error-gen/src/error-gen.c
@@ -730,7 +730,7 @@ error_gen_mkdir_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
int
error_gen_mkdir (call_frame_t *frame, xlator_t *this,
- loc_t *loc, mode_t mode)
+ loc_t *loc, mode_t mode, dict_t *params)
{
int op_errno = 0;
eg_t *egp = NULL;
@@ -752,7 +752,7 @@ error_gen_mkdir (call_frame_t *frame, xlator_t *this,
STACK_WIND (frame, error_gen_mkdir_cbk,
FIRST_CHILD(this),
FIRST_CHILD(this)->fops->mkdir,
- loc, mode);
+ loc, mode, params);
return 0;
}
diff --git a/xlators/debug/io-stats/src/io-stats.c b/xlators/debug/io-stats/src/io-stats.c
index ce2bac51ad9..b283bd7b4bc 100644
--- a/xlators/debug/io-stats/src/io-stats.c
+++ b/xlators/debug/io-stats/src/io-stats.c
@@ -884,14 +884,14 @@ io_stats_mknod (call_frame_t *frame, xlator_t *this,
int
io_stats_mkdir (call_frame_t *frame, xlator_t *this,
- loc_t *loc, mode_t mode)
+ loc_t *loc, mode_t mode, dict_t *params)
{
BUMP_FOP (MKDIR);
STACK_WIND (frame, io_stats_mkdir_cbk,
FIRST_CHILD(this),
FIRST_CHILD(this)->fops->mkdir,
- loc, mode);
+ loc, mode, params);
return 0;
}
diff --git a/xlators/debug/trace/src/trace.c b/xlators/debug/trace/src/trace.c
index 339ebeaf57f..7b5164c8439 100644
--- a/xlators/debug/trace/src/trace.c
+++ b/xlators/debug/trace/src/trace.c
@@ -1349,7 +1349,8 @@ trace_mknod (call_frame_t *frame, xlator_t *this, loc_t *loc,
int
-trace_mkdir (call_frame_t *frame, xlator_t *this, loc_t *loc, mode_t mode)
+trace_mkdir (call_frame_t *frame, xlator_t *this, loc_t *loc, mode_t mode,
+ dict_t *params)
{
if (trace_fop_names[GF_FOP_MKDIR].enabled) {
gf_log (this->name, GF_LOG_NORMAL,
@@ -1361,7 +1362,7 @@ trace_mkdir (call_frame_t *frame, xlator_t *this, loc_t *loc, mode_t mode)
STACK_WIND (frame, trace_mkdir_cbk,
FIRST_CHILD(this),
FIRST_CHILD(this)->fops->mkdir,
- loc, mode);
+ loc, mode, params);
return 0;
}
='graph'>
-rw-r--r--cli/src/cli-xml-output.c27
-rw-r--r--cli/src/cli.c12
-rw-r--r--cli/src/cli.h13
-rw-r--r--configure.ac191
-rw-r--r--contrib/fuse-lib/mount.c1
-rw-r--r--contrib/sunrpc/xdr_sizeof.c204
-rw-r--r--doc/developer-guide/fuse-interrupt.md97
-rw-r--r--doc/developer-guide/identifying-resource-leaks.md24
-rw-r--r--doc/developer-guide/thread-naming.md5
-rw-r--r--doc/release-notes/8.0.md368
-rw-r--r--events/src/eventsapiconf.py.in2
-rw-r--r--events/src/glustereventsd.py37
-rw-r--r--events/src/peer_eventsapi.py3
-rw-r--r--events/src/utils.py3
-rwxr-xr-xextras/distributed-testing/distributed-test-runner.py12
-rw-r--r--extras/ganesha/ocf/ganesha_nfsd2
-rw-r--r--extras/ganesha/scripts/ganesha-ha.sh7
-rwxr-xr-xextras/glusterfs-georep-upgrade.py77
-rwxr-xr-xextras/hook-scripts/set/post/S32gluster_enable_shared_storage.sh27
-rwxr-xr-xextras/hook-scripts/start/post/S31ganesha-start.sh2
-rwxr-xr-xextras/quota/quota_fsck.py8
-rwxr-xr-xextras/snap_scheduler/gcron.py7
-rwxr-xr-xextras/snap_scheduler/snap_scheduler.py2
-rwxr-xr-xextras/statedumpparse.rb151
-rw-r--r--extras/systemd/glusterd.service.in5
-rw-r--r--geo-replication/gsyncd.conf.in2
-rw-r--r--geo-replication/syncdaemon/master.py5
-rw-r--r--geo-replication/syncdaemon/syncdutils.py45
-rw-r--r--glusterfs.spec.in23
-rw-r--r--glusterfsd/src/Makefile.am5
-rw-r--r--glusterfsd/src/gf_attach.c45
-rw-r--r--glusterfsd/src/glusterfsd-mgmt.c5
-rw-r--r--glusterfsd/src/glusterfsd.c12
-rw-r--r--heal/src/glfs-heal.c4
-rw-r--r--libglusterfs/src/Makefile.am3
-rw-r--r--libglusterfs/src/client_t.c4
-rw-r--r--libglusterfs/src/common-utils.c115
-rw-r--r--libglusterfs/src/ctx.c8
-rw-r--r--libglusterfs/src/dict.c271
-rw-r--r--libglusterfs/src/event.c3
-rw-r--r--libglusterfs/src/events.c29
-rw-r--r--libglusterfs/src/gidcache.c5
-rw-r--r--libglusterfs/src/glusterfs/common-utils.h118
-rw-r--r--libglusterfs/src/glusterfs/dict.h10
-rw-r--r--libglusterfs/src/glusterfs/gf-event.h4
-rw-r--r--libglusterfs/src/glusterfs/globals.h4
-rw-r--r--libglusterfs/src/glusterfs/glusterfs-acl.h2
-rw-r--r--libglusterfs/src/glusterfs/glusterfs.h10
-rw-r--r--libglusterfs/src/glusterfs/latency.h22
-rw-r--r--libglusterfs/src/glusterfs/mem-pool.h32
-rw-r--r--libglusterfs/src/glusterfs/mem-types.h1
-rw-r--r--libglusterfs/src/glusterfs/stack.h14
-rw-r--r--libglusterfs/src/glusterfs/statedump.h2
-rw-r--r--libglusterfs/src/glusterfs/store.h2
-rw-r--r--libglusterfs/src/glusterfs/syncop.h20
-rw-r--r--libglusterfs/src/glusterfs/syscall.h9
-rw-r--r--libglusterfs/src/glusterfs/xlator.h2
-rw-r--r--libglusterfs/src/graph.c9
-rw-r--r--libglusterfs/src/latency.c93
-rw-r--r--libglusterfs/src/libglusterfs.sym10
-rw-r--r--libglusterfs/src/logging.c131
-rw-r--r--libglusterfs/src/mem-pool.c114
-rw-r--r--libglusterfs/src/monitoring.c8
-rw-r--r--libglusterfs/src/stack.c4
-rw-r--r--libglusterfs/src/statedump.c164
-rw-r--r--libglusterfs/src/store.c12
-rw-r--r--libglusterfs/src/syncop.c48
-rw-r--r--libglusterfs/src/syscall.c24
-rw-r--r--libglusterfs/src/xlator.c46
-rwxr-xr-xrfc.sh120
-rw-r--r--rpc/rpc-lib/src/Makefile.am2
-rw-r--r--rpc/rpc-lib/src/libgfrpc.sym1
-rw-r--r--rpc/rpc-lib/src/protocol-common.h1
-rw-r--r--rpc/rpc-lib/src/rpc-clnt-ping.c2
-rw-r--r--rpc/rpc-lib/src/rpc-clnt.c27
-rw-r--r--rpc/rpc-lib/src/rpc-clnt.h1
-rw-r--r--rpc/rpc-lib/src/rpcsvc.c130
-rw-r--r--rpc/rpc-lib/src/rpcsvc.h7
-rw-r--r--rpc/rpc-transport/socket/src/socket.c36
-rw-r--r--rpc/xdr/src/glusterd1-xdr.x15
-rw-r--r--rpc/xdr/src/libgfxdr.sym2
-rwxr-xr-xrun-tests.sh72
-rw-r--r--tests/00-geo-rep/00-georep-verify-non-root-setup.t27
-rw-r--r--tests/00-geo-rep/00-georep-verify-setup.t3
-rw-r--r--tests/00-geo-rep/georep-upgrade.t79
-rw-r--r--tests/000-flaky/basic_afr_split-brain-favorite-child-policy.t (renamed from tests/basic/afr/split-brain-favorite-child-policy.t)4
-rw-r--r--tests/000-flaky/basic_changelog_changelog-snapshot.t (renamed from tests/basic/changelog/changelog-snapshot.t)4
-rw-r--r--tests/000-flaky/basic_distribute_rebal-all-nodes-migrate.t (renamed from tests/basic/distribute/rebal-all-nodes-migrate.t)8
-rw-r--r--[-rwxr-xr-x]tests/000-flaky/basic_ec_ec-quorum-count-partial-failure.t (renamed from tests/basic/ec/ec-quorum-count-partial-failure.t)4
-rw-r--r--[-rwxr-xr-x]tests/000-flaky/basic_mount-nfs-auth.t (renamed from tests/basic/mount-nfs-auth.t)0
-rw-r--r--tests/000-flaky/bugs_core_multiplex-limit-issue-151.t (renamed from tests/bugs/core/multiplex-limit-issue-151.t)6
-rw-r--r--[-rwxr-xr-x]tests/000-flaky/bugs_distribute_bug-1117851.t (renamed from tests/bugs/distribute/bug-1117851.t)4
-rw-r--r--tests/000-flaky/bugs_distribute_bug-1122443.t (renamed from tests/bugs/distribute/bug-1122443.t)17
-rw-r--r--tests/000-flaky/bugs_glusterd_bug-857330/common.rc (renamed from tests/bugs/glusterd/bug-857330/common.rc)2
-rwxr-xr-xtests/000-flaky/bugs_glusterd_bug-857330/normal.t (renamed from tests/bugs/glusterd/bug-857330/normal.t)4
-rwxr-xr-xtests/000-flaky/bugs_glusterd_bug-857330/xml.t (renamed from tests/bugs/glusterd/bug-857330/xml.t)4
-rw-r--r--[-rwxr-xr-x]tests/000-flaky/bugs_glusterd_quorum-value-check.t (renamed from tests/bugs/glusterd/quorum-value-check.t)4
-rw-r--r--tests/000-flaky/bugs_nfs_bug-1116503.t (renamed from tests/bugs/nfs/bug-1116503.t)6
-rw-r--r--tests/000-flaky/features_lock-migration_lkmigration-set-option.t (renamed from tests/features/lock-migration/lkmigration-set-option.t)4
-rw-r--r--tests/afr.rc6
-rw-r--r--tests/basic/afr/afr-anon-inode-no-quorum.t63
-rw-r--r--tests/basic/afr/afr-anon-inode.t114
-rw-r--r--tests/basic/afr/entry-self-heal-anon-dir-off.t459
-rw-r--r--tests/basic/afr/gfid-mismatch-resolution-with-fav-child-policy.t1
-rw-r--r--tests/basic/afr/rename-data-loss.t72
-rw-r--r--tests/basic/afr/split-brain-favorite-child-policy-client-side-healing.t124
-rw-r--r--tests/basic/distribute/spare_file_rebalance.t51
-rw-r--r--tests/basic/ec/ec-quorum-count.t2
-rw-r--r--tests/basic/fencing/afr-lock-heal-basic.t6
-rw-r--r--tests/basic/fuse/active-io-graph-switch.t65
-rw-r--r--tests/basic/gfapi/gfapi-ssl-load-volfile-test.c127
-rwxr-xr-xtests/basic/gfapi/gfapi-ssl-load-volfile-test.t76
-rw-r--r--tests/basic/gfapi/protocol-client-ssl.vol.in15
-rw-r--r--tests/basic/metadisp/fsyncdir.c29
-rw-r--r--tests/basic/metadisp/ftruncate.c34
-rw-r--r--tests/basic/metadisp/fxattr.c107
-rw-r--r--tests/basic/metadisp/gfs-fsetxattr.c141
-rw-r--r--tests/basic/metadisp/metadisp.t316
-rw-r--r--tests/basic/metadisp/metadisp.vol14
-rw-r--r--tests/basic/quick-read-with-upcall.t15
-rw-r--r--tests/bugs/distribute/bug-1600379.t54
-rwxr-xr-xtests/bugs/distribute/issue-1327.t33
-rwxr-xr-xtests/bugs/fuse/many-groups-for-acl.t13
-rw-r--r--tests/bugs/glusterd/brick-mux-validation-in-cluster.t4
-rw-r--r--tests/bugs/glusterd/brick-mux.t2
-rw-r--r--tests/bugs/glusterd/brick-order-check-add-brick.t61
-rw-r--r--tests/bugs/glusterd/check_elastic_server.t3
-rw-r--r--tests/bugs/glusterd/mgmt-handshake-and-volume-sync-post-glusterd-restart.t18
-rw-r--r--tests/bugs/glusterd/remove-brick-validation.t (renamed from tests/bugs/glusterd/enable-shared-storage-and-remove-brick-validation.t)14
-rw-r--r--tests/bugs/glusterd/validating-options-for-replicated-volume.t13
-rw-r--r--tests/bugs/glusterfs-server/bug-873549.t2
-rwxr-xr-xtests/bugs/logging/bug-823081.t8
-rw-r--r--tests/bugs/replicate/bug-1101647.t2
-rw-r--r--tests/bugs/replicate/bug-1744548-heal-timeout.t6
-rw-r--r--tests/cluster.rc25
-rw-r--r--tests/features/flock_interrupt.t1
-rw-r--r--tests/features/interrupt.t14
-rw-r--r--tests/features/ssl-ciphers.t11
-rwxr-xr-xtests/features/trash.t74
-rw-r--r--tests/geo-rep.rc5
-rw-r--r--tests/line-coverage/cli-peer-and-volume-operations.t40
-rw-r--r--tests/line-coverage/log-and-brick-ops-negative-case.t82
-rw-r--r--tests/volume.rc8
-rw-r--r--tools/gfind_missing_files/gfind_missing_files.sh2
-rw-r--r--tools/glusterfind/src/utils.py6
-rw-r--r--xlators/cluster/afr/src/afr-common.c96
-rw-r--r--xlators/cluster/afr/src/afr-dir-read.c15
-rw-r--r--xlators/cluster/afr/src/afr-open.c8
-rw-r--r--xlators/cluster/afr/src/afr-read-txn.c8
-rw-r--r--xlators/cluster/afr/src/afr-self-heal-common.c187
-rw-r--r--xlators/cluster/afr/src/afr-self-heal-entry.c205
-rw-r--r--xlators/cluster/afr/src/afr-self-heal-name.c35
-rw-r--r--xlators/cluster/afr/src/afr-self-heal.h5
-rw-r--r--xlators/cluster/afr/src/afr-self-heald.c184
-rw-r--r--xlators/cluster/afr/src/afr-self-heald.h2
-rw-r--r--xlators/cluster/afr/src/afr.c47
-rw-r--r--xlators/cluster/afr/src/afr.h15
-rw-r--r--xlators/cluster/dht/src/dht-common.c173
-rw-r--r--xlators/cluster/dht/src/dht-common.h120
-rw-r--r--xlators/cluster/dht/src/dht-diskusage.c12
-rw-r--r--xlators/cluster/dht/src/dht-helper.c12
-rw-r--r--xlators/cluster/dht/src/dht-inode-write.c32
-rw-r--r--xlators/cluster/dht/src/dht-mem-types.h3
-rw-r--r--xlators/cluster/dht/src/dht-messages.h36
-rw-r--r--xlators/cluster/dht/src/dht-rebalance.c668
-rw-r--r--xlators/cluster/dht/src/dht-selfheal.c68
-rw-r--r--xlators/cluster/dht/src/dht-shared.c85
-rw-r--r--xlators/cluster/dht/src/nufa.c1
-rw-r--r--xlators/cluster/ec/src/ec-combine.c19
-rw-r--r--xlators/cluster/ec/src/ec-common.c76
-rw-r--r--xlators/cluster/ec/src/ec-heal.c15
-rw-r--r--xlators/cluster/ec/src/ec-heald.c51
-rw-r--r--xlators/cluster/ec/src/ec-types.h5
-rw-r--r--xlators/cluster/ec/src/ec.c20
-rw-r--r--xlators/debug/error-gen/src/error-gen.c46
-rw-r--r--xlators/debug/io-stats/src/io-stats.c145
-rw-r--r--xlators/debug/trace/src/trace.c20
-rw-r--r--xlators/features/Makefile.am6
-rw-r--r--xlators/features/bit-rot/src/bitd/bit-rot-scrub-status.c12
-rw-r--r--xlators/features/bit-rot/src/bitd/bit-rot-scrub-status.h20
-rw-r--r--xlators/features/bit-rot/src/bitd/bit-rot-scrub.c68
-rw-r--r--xlators/features/bit-rot/src/stub/bit-rot-stub-messages.h5
-rw-r--r--xlators/features/bit-rot/src/stub/bit-rot-stub.c68
-rw-r--r--xlators/features/changelog/src/changelog-helpers.c52
-rw-r--r--xlators/features/changelog/src/changelog-helpers.h21
-rw-r--r--xlators/features/changelog/src/changelog-messages.h8
-rw-r--r--xlators/features/changelog/src/changelog.c51
-rw-r--r--xlators/features/cloudsync/src/Makefile.am4
-rw-r--r--xlators/features/cloudsync/src/cloudsync-plugins/src/cloudsyncs3/src/libcloudsyncs3.c2
-rw-r--r--xlators/features/index/src/index.c4
-rw-r--r--xlators/features/leases/src/leases-internal.c4
-rw-r--r--xlators/features/locks/src/clear.c4
-rw-r--r--xlators/features/locks/src/common.c6
-rw-r--r--xlators/features/locks/src/entrylk.c12
-rw-r--r--xlators/features/locks/src/inodelk.c12
-rw-r--r--xlators/features/locks/src/locks.h17
-rw-r--r--xlators/features/locks/src/posix.c33
-rw-r--r--xlators/features/metadisp/Makefile.am3
-rw-r--r--xlators/features/metadisp/src/Makefile.am38
-rw-r--r--xlators/features/metadisp/src/backend.c45
-rw-r--r--xlators/features/metadisp/src/fops-tmpl.c10
-rw-r--r--xlators/features/metadisp/src/gen-fops.py160
-rw-r--r--xlators/features/metadisp/src/metadisp-create.c101
-rw-r--r--xlators/features/metadisp/src/metadisp-fops.h51
-rw-r--r--xlators/features/metadisp/src/metadisp-fsync.c54
-rw-r--r--xlators/features/metadisp/src/metadisp-lookup.c90
-rw-r--r--xlators/features/metadisp/src/metadisp-open.c70
-rw-r--r--xlators/features/metadisp/src/metadisp-readdir.c65
-rw-r--r--xlators/features/metadisp/src/metadisp-setattr.c90
-rw-r--r--xlators/features/metadisp/src/metadisp-stat.c124
-rw-r--r--xlators/features/metadisp/src/metadisp-unlink.c160
-rw-r--r--xlators/features/metadisp/src/metadisp.c46
-rw-r--r--xlators/features/metadisp/src/metadisp.h45
-rw-r--r--xlators/features/quota/src/quota-enforcer-client.c6
-rw-r--r--xlators/features/quota/src/quota.c59
-rw-r--r--xlators/features/quota/src/quota.h5
-rw-r--r--xlators/features/read-only/src/worm-helper.c15
-rw-r--r--xlators/features/trash/src/trash.c4
-rw-r--r--xlators/features/upcall/src/upcall-internal.c10
-rw-r--r--xlators/meta/src/meta-helpers.c9
-rw-r--r--xlators/mgmt/glusterd/src/Makefile.am8
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-brick-ops.c336
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-conn-mgmt.c19
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-ganesha.c64
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-geo-rep.c55
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-gfproxyd-svc.c11
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-handler.c259
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-handshake.c143
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-hooks.c100
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-log-ops.c10
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-mem-types.h1
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-messages.h152
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-mgmt-handler.c211
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-mgmt.c431
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-mgmt.h10
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-mountbroker.c10
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-op-sm.c274
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-peer-utils.c47
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-pmap.c12
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-quota.c14
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-quotad-svc.c4
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-rebalance.c23
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-replace-brick.c2
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-rpc-ops.c74
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-scrub-svc.c4
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-server-quorum.c13
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-shd-svc.c91
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-sm.c78
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-snapd-svc.c17
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-snapshot-utils.c71
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-snapshot.c158
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-store.c283
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-store.h15
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-svc-helper.c19
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-svc-mgmt.c19
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-syncop.c77
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-utils.c1004
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-utils.h5
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-volgen.c354
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-volume-ops.c424
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-volume-set.c49
-rw-r--r--xlators/mgmt/glusterd/src/glusterd.c66
-rw-r--r--xlators/mgmt/glusterd/src/glusterd.h15
-rw-r--r--xlators/mount/fuse/src/fuse-bridge.c110
-rw-r--r--xlators/mount/fuse/src/fuse-bridge.h25
-rw-r--r--xlators/mount/fuse/src/fuse-helpers.c71
-rwxr-xr-xxlators/mount/fuse/utils/mount_glusterfs.in9
-rw-r--r--xlators/nfs/server/src/acl3.c17
-rw-r--r--xlators/nfs/server/src/acl3.h2
-rw-r--r--xlators/nfs/server/src/auth-cache.c4
-rw-r--r--xlators/nfs/server/src/mount3.c11
-rw-r--r--xlators/nfs/server/src/mount3udp_svc.c2
-rw-r--r--xlators/nfs/server/src/nfs.c14
-rw-r--r--xlators/nfs/server/src/nfs3-helpers.c4
-rw-r--r--xlators/nfs/server/src/nfs3.c2
-rw-r--r--xlators/nfs/server/src/nlm4.c17
-rw-r--r--xlators/nfs/server/src/nlmcbk_svc.c5
-rw-r--r--xlators/performance/io-cache/src/io-cache.c36
-rw-r--r--xlators/performance/io-cache/src/io-cache.h27
-rw-r--r--xlators/performance/io-cache/src/page.c6
-rw-r--r--xlators/performance/io-threads/src/io-threads.c11
-rw-r--r--xlators/performance/md-cache/src/md-cache.c72
-rw-r--r--xlators/performance/nl-cache/src/nl-cache-helper.c8
-rw-r--r--xlators/performance/nl-cache/src/nl-cache.c6
-rw-r--r--xlators/performance/nl-cache/src/nl-cache.h2
-rw-r--r--xlators/performance/open-behind/src/open-behind.c52
-rw-r--r--xlators/performance/quick-read/src/quick-read.c41
-rw-r--r--xlators/performance/quick-read/src/quick-read.h2
-rw-r--r--xlators/performance/write-behind/src/write-behind.c4
-rw-r--r--xlators/protocol/server/src/Makefile.am4
-rw-r--r--xlators/protocol/server/src/server-common.c2
-rw-r--r--xlators/protocol/server/src/server.c2
-rw-r--r--xlators/storage/posix/src/posix-common.c16
-rw-r--r--xlators/storage/posix/src/posix-entry-ops.c39
-rw-r--r--xlators/storage/posix/src/posix-helpers.c73
-rw-r--r--xlators/storage/posix/src/posix-inode-fd-ops.c69
-rw-r--r--xlators/storage/posix/src/posix.h7
-rw-r--r--xlators/system/posix-acl/src/posix-acl.c10
320 files changed, 11522 insertions, 4502 deletions
diff --git a/.github/ISSUE_TEMPLATE b/.github/ISSUE_TEMPLATE
index 329b4aac35d..386ed2d8dd5 100644
--- a/.github/ISSUE_TEMPLATE
+++ b/.github/ISSUE_TEMPLATE
@@ -1,12 +1,13 @@
<!-- Please use this template while reporting an issue, providing as much information as possible. Failure to do so may result in a delayed response. Thank you! -->
-**Description of problem: **
+**Description of problem:**
**The exact command to reproduce the issue**:
-**The full output of the command that failed**:<details>
+**The full output of the command that failed**:
+<details>
@@ -18,11 +19,12 @@
**Additional info:**
-** - The output of the `gluster volume info` command**:<details>
+**- The output of the `gluster volume info` command**:
+<details>
</details>
-** - The operating system / glusterfs version**:
+**- The operating system / glusterfs version**:
diff --git a/.github/RELEASE_TRACKER_TEMPLATE b/.github/RELEASE_TRACKER_TEMPLATE
new file mode 100644
index 00000000000..502bbd5556c
--- /dev/null
+++ b/.github/RELEASE_TRACKER_TEMPLATE
@@ -0,0 +1,12 @@
+<!-- Please use this template while creating a tracker issue -->
+
+**Description of problem:**
+A tracker issue to track the issues that will be fixed as a part of this release
+
+
+**Major or minor release**:
+
+
+**Release version**:
+
+
diff --git a/.github/stale.yml b/.github/stale.yml
new file mode 100644
index 00000000000..460e327c6ea
--- /dev/null
+++ b/.github/stale.yml
@@ -0,0 +1,25 @@
+# Number of days of inactivity before an issue becomes stale
+daysUntilStale: 210
+# Number of days of inactivity before a stale issue is closed
+daysUntilClose: 15
+# Issues with these labels will never be considered stale
+exemptLabels:
+ - pinned
+ - security
+# Label to use when marking an issue as stale
+staleLabel: wontfix
+
+# Comment to post when marking an issue as stale. Set to `false` to disable
+markComment: >
+ Thank you for your contributions.
+
+ Noticed that this issue is not having any activity in last ~6 months! We
+ are marking this issue as stale because it has not had recent activity.
+
+ It will be closed in 2 weeks if no one responds with a comment here.
+
+
+# Comment to post when closing a stale issue. Set to `false` to disable
+closeComment: >
+ Closing this issue as there was no update since my last update on issue.
+ If this is an issue which is still valid, feel free to open it.
diff --git a/.gitignore b/.gitignore
index 849a0f1a8be..fc5ba586f8e 100644
--- a/.gitignore
+++ b/.gitignore
@@ -121,3 +121,5 @@ xlators/features/cloudsync/src/cloudsync-autogen-fops.c
xlators/features/cloudsync/src/cloudsync-autogen-fops.h
xlators/features/utime/src/utime-autogen-fops.c
xlators/features/utime/src/utime-autogen-fops.h
+tests/basic/metadisp/ftruncate
+xlators/features/metadisp/src/fops.c
diff --git a/.testignore b/.testignore
index 190573191e1..fe8f838bf2b 100644
--- a/.testignore
+++ b/.testignore
@@ -1,4 +1,6 @@
.github/ISSUE_TEMPLATE
+.github/PULL_REQUEST_TEMPLATE
+.github/stale.yml
.gitignore
.mailmap
.testignore
@@ -6,7 +8,7 @@
rfc.sh
submit-for-review.sh
AUTHORS
-CONTRIBUTING
+CONTRIBUTING.md
COPYING-GPLV2
COPYING-LGPLV3
ChangeLog
diff --git a/CONTRIBUTING b/CONTRIBUTING
deleted file mode 100644
index eb30d76d1b4..00000000000
--- a/CONTRIBUTING
+++ /dev/null
@@ -1,30 +0,0 @@
-# GlusterFS project Contribution guidelines
-
-
-## By contributing to this project, the contributor would need to agree to below.
-
-### Developer's Certificate of Origin 1.1
-
-By making a contribution to this project, I certify that:
-
-(a) The contribution was created in whole or in part by me and I
- have the right to submit it under the open source license
- indicated in the file; or
-
-(b) The contribution is based upon previous work that, to the best
- of my knowledge, is covered under an appropriate open source
- license and I have the right under that license to submit that
- work with modifications, whether created in whole or in part
- by me, under the same open source license (unless I am
- permitted to submit under a different license), as indicated
- in the file; or
-
-(c) The contribution was provided directly to me by some other
- person who certified (a), (b) or (c) and I have not modified
- it.
-
-(d) I understand and agree that this project and the contribution
- are public and that a record of the contribution (including all
- personal information I submit with it, including my sign-off) is
- maintained indefinitely and may be redistributed consistent with
- this project or the open source license(s) involved.
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
new file mode 100644
index 00000000000..65fc3497104
--- /dev/null
+++ b/CONTRIBUTING.md
@@ -0,0 +1,114 @@
+# GlusterFS project Contribution guidelines
+
+## Development Workflow
+
+We follow most of the details as per the [document here](https://help.github.com/en/github/collaborating-with-issues-and-pull-requests). If you are not aware of the github workflow, it is recommended to go through them before continuing here.
+
+
+#### Get the Repository setup
+
+0. Fork Repository
+ - Fork [GlusterFS repository](https://github.com/gluster/glusterfs/fork).
+
+1. Clone Repository
+ - Clone the glusterfs repo freshly from github using below steps.
+
+```
+ git clone git@github.com:${username}/glusterfs.git
+ cd glusterfs/
+ git remote add upstream git@github.com:gluster/glusterfs.git
+```
+
+About two tasks are one time for the life time. You can continue to use the same repository for all the work in future.
+
+#### Development & Other flows
+
+0. Issue:
+ - Make sure there is an issue filed for the task you are working on.
+ - If it is not filed, open the issue with all the description.
+ - If it is a bug fix, add label "Type:Bug".
+ - If it is an RFC, provide all the documentation, and request for "DocApproved", and "SpecApproved" label.
+
+1. Code:
+ - Start coding
+ - Build and test locally
+ - Make sure clang-format is installed and is run on the patch.
+
+2. Keep up-to-date
+ - GlusterFS is a large project with many developers, so there would be one or the other patch everyday.
+ - It is critical for developer to be up-to-date with `devel` repo to be Conflict-Free when PR is opened.
+ - Git provides many options to keep up-to-date, below is one of them
+```
+ git fetch upstream
+ git rebase upstream/devel
+```
+ - It is recommended you keep pushing to your repo every day, so you don't loose any work.
+ - It can be done by `./rfc.sh` (or `git push origin HEAD:issueNNN`)
+
+2. Commit Message / PR description:
+ - The name of the branch on your personal fork can start with issueNNNN, followed by anything of your choice.
+ - PRs continue to have the title of format "component: \<title\>", like it is practiced now.
+ - When you open a PR, having a reference Issue for the commit is mandatory in GlusterFS.
+ - Commit message can have, either `Fixes: #NNNN` or `Updates: #NNNN` in a separate line in the commit message.
+ - Here, NNNN is the Issue ID in glusterfs repository.
+ - Each commit needs the author to have the "Signed-off-by: Name \<email\>" line.
+ - Can do this by `-s` option for `git commit`.
+ - If the PR is not ready for review, apply the label `work-in-progress`.
+ - Check the availability of "Draft PR" is present for you, if yes, use that instead.
+
+3. Tests:
+ - All the required smoke tests would be auto-triggered.
+ - Developers get a chance to retrigger the smoke tests using **"/recheck smoke"** as comment.
+ - The "regression" tests would be triggered by a comment **"/run regression"** from developers in the [@gluster-maintainers](https://github.com/orgs/gluster/teams/gluster-maintainers) group.
+ - Ask for help as comment in PR if you have any questions about the process!
+
+4. Review Process:
+ - `+2` : is equivalent to "Approve" from the people in the maintainer's group.
+ - `+1` : can be given by a maintainer/reviewer by explicitly stating that in the comment.
+ - `-1` : provide details on required changes and pick "Request Changes" while submitting your review.
+ - `-2` : done by adding the `DO-NOT-MERGE` label.
+
+ - Any further discussions can happen as comments in the PR.
+
+5. Making changes:
+ - There are 2 approaches to submit changes done after addressing review comments.
+ - Commit changes as a new commit on top of the original commits in the branch, and push the changes to same branch (issueNNNN)
+ - Commit changes into the same commit with `--amend` option, and do a push to the same branch with `--force` option.
+
+6. Merging:
+ - GlusterFS project follows 'Squash and Merge' method
+ - This is mainly to preserve the historic Gerrit method of one patch in `git log` for one URL link.
+ - This also makes every merge a complete patch, which has passed all tests.
+ - The merging of the patch is expected to be done by the maintainers.
+ - It can be done when all the tests (smoke and regression) pass.
+ - When the PR has 'Approved' flag from corresponding maintainer.
+ - If you feel there is delay, feel free to add a comment, discuss the same in Slack channel, or send email.
+
+## By contributing to this project, the contributor would need to agree to below.
+
+### Developer's Certificate of Origin 1.1
+
+By making a contribution to this project, I certify that:
+
+(a) The contribution was created in whole or in part by me and I
+ have the right to submit it under the open source license
+ indicated in the file; or
+
+(b) The contribution is based upon previous work that, to the best
+ of my knowledge, is covered under an appropriate open source
+ license and I have the right under that license to submit that
+ work with modifications, whether created in whole or in part
+ by me, under the same open source license (unless I am
+ permitted to submit under a different license), as indicated
+ in the file; or
+
+(c) The contribution was provided directly to me by some other
+ person who certified (a), (b) or (c) and I have not modified
+ it.
+
+(d) I understand and agree that this project and the contribution
+ are public and that a record of the contribution (including all
+ personal information I submit with it, including my sign-off) is
+ maintained indefinitely and may be redistributed consistent with
+ this project or the open source license(s) involved.
+
diff --git a/MAINTAINERS b/MAINTAINERS
index b4d8195e0e9..953e8755fd9 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -53,8 +53,7 @@ General Project Architects
--------------------------
M: Amar Tumballi <amarts@gmail.com>
M: Xavier Hernandez <xhernandez@redhat.com>
-P: Pranith Karampuri <pkarampu@redhat.com>
-P: Raghavendra Gowdappa <rgowdapp@redhat.com>
+P: Pranith Karampuri <pranith.karampuri@phonepe.com>
P: Atin Mukherjee <amukherj@redhat.com>
xlators:
@@ -67,13 +66,14 @@ F: xlators/system/posix-acl/
Arbiter
M: Ravishankar N <ravishankar@redhat.com>
-P: Pranith Karampuri <pkarampu@redhat.com>
+P: Pranith Karampuri <pranith.karampuri@phonepe.com>
S: Maintained
F: xlators/features/arbiter/
Automatic File Replication (AFR)
-M: Pranith Karampuri <pkarampu@redhat.com>
-P: Ravishankar N <ravishankar@redhat.com>
+M: Pranith Karampuri <pranith.karampuri@phonepe.com>
+M: Ravishankar N <ravishankar@redhat.com>
+P: Karthik US <ksubrahm@redhat.com>
S: Maintained
F: xlators/cluster/afr/
@@ -96,13 +96,12 @@ S: Maintained
F: xlators/features/changelog/
Distributed Hashing Table (DHT)
-M: Raghavendra Gowdappa <rgowdapp@redhat.com>
P: Susant Palai <spalai@redhat.com>
S: Maintained
F: xlators/cluster/dht/
Erasure Coding
-M: Pranith Karampuri <pkarampu@redhat.com>
+M: Pranith Karampuri <pranith.karampuri@phonepe.com>
M: Xavier Hernandez <xhernandez@redhat.com>
P: Ashish Pandey <aspandey@redhat.com>
S: Maintained
@@ -115,20 +114,18 @@ F: xlators/debug/error-gen/
FUSE Bridge
M: Csaba Henk <chenk@redhat.com>
-P: Raghavendra Gowdappa <rgowdapp@redhat.com>
P: Niels de Vos <ndevos@redhat.com>
S: Maintained
F: xlators/mount/
Index
-M: Pranith Karampuri <pkarampu@redhat.com>
+M: Pranith Karampuri <pranith.karampuri@phonepe.com>
P: Ravishankar N <ravishankar@redhat.com>
S: Maintained
F: xlators/features/index/
IO Cache
-M: Raghavendra Gowdappa <rgowdapp@redhat.com>
-P: Mohammed Rafi KC <rkavunga@redhat.com>
+P: Mohammed Rafi KC <rafi.kavungal@iternity.com>
S: Maintained
F: xlators/performance/io-cache/
@@ -139,7 +136,7 @@ S: Maintained
F: xlators/debug/io-stats/
IO threads
-M: Pranith Karampuri <pkarampu@redhat.com>
+M: Pranith Karampuri <pranith.karampuri@phonepe.com>
P: Ravishankar N <ravishankar@redhat.com>
S: Maintained
F: xlators/performance/io-threads/
@@ -158,13 +155,12 @@ S: Maintained
F: xlators/features/locks/
Marker
-M: Raghavendra Gowdappa <rgowdapp@redhat.com>
M: Kotresh HR <khiremat@redhat.com>
S: Maintained
F: xlators/features/marker/
Meta
-M: Mohammed Rafi KC <rkavunga@redhat.com>
+M: Mohammed Rafi KC <rafi.kavungal@iternity.com>
S: Maintained
F: xlators/features/meta/
@@ -176,7 +172,7 @@ F: xlators/performance/md-cache/
Negative-lookup Cache
M: Poornima G <pgurusid@redhat.com>
-P: Pranith Karampuri <pkarampu@redhat.com>
+P: Pranith Karampuri <pranith.karampuri@phonepe.com>
S: Maintained
F: xlators/performance/nl-cache/
@@ -188,7 +184,6 @@ S: Odd Fixes
F: xlators/nfs/server/
Open-behind
-M: Raghavendra Gowdappa <rgowdapp@redhat.com>
S: Maintained
F: xlators/performance/open-behind/
@@ -200,25 +195,21 @@ S: Maintained
F: xlators/storage/posix/
Quick-read
-M: Raghavendra Gowdappa <rgowdapp@redhat.com>
S: Maintained
F: xlators/performance/quick-read/
Quota
-M: Raghavendra Gowdappa <rgowdapp@redhat.com>
M: Shyamsundar Ranganathan <srangana@redhat.com>
P: Hari Gowtham <hgowtham@redhat.com>
S: Maintained
F: xlators/features/quota/
Read-ahead
-M: Raghavendra Gowdappa <rgowdapp@redhat.com>
P: Csaba Henk <chenk@redhat.com>
S: Maintained
F: xlators/performance/read-ahead/
Readdir-ahead
-M: Raghavendra Gowdappa <rgowdapp@redhat.com>
S: Maintained
F: xlators/performance/readdir-ahead/
@@ -242,7 +233,6 @@ S: Maintained
F: xlators/features/upcall/
Write-behind
-M: Raghavendra Gowdappa <rgowdapp@redhat.com>
P: Csaba Henk <chenk@redhat.com>
S: Maintained
F: xlators/performance/write-behind/
@@ -292,8 +282,7 @@ M: Xavier Hernandez <xhernandez@redhat.com>
M: Jeff Darcy <jeff@pl.atyp.us>
P: Kaleb Keithley <kkeithle@redhat.com>
P: Niels de Vos <ndevos@redhat.com>
-P: Pranith Karampuri <pkarampu@redhat.com>
-P: Raghavendra Gowdappa <rgowdapp@redhat.com>
+P: Pranith Karampuri <pranith.karampuri@phonepe.com>
P: Shyamsundar Ranganathan <srangana@redhat.com>
S: Maintained
F: libglusterfs/
@@ -316,13 +305,11 @@ F: xlators/mgmt/glusterd/
Protocol
M: Niels de Vos <ndevos@redhat.com>
-M: Raghavendra Gowdappa <rgowdapp@redhat.com>
-P: Mohammed Rafi KC <rkavunga@redhat.com>
+P: Mohammed Rafi KC <rafi.kavungal@iternity.com>
S: Maintained
F: xlators/protocol/
Remote Procedure Call subsystem
-M: Raghavendra Gowdappa <rgowdapp@redhat.com>
P: Mohit Agrawal <moagrawa@redhat.com>
S: Maintained
F: rpc/rpc-lib/
@@ -330,17 +317,16 @@ F: rpc/xdr/
Snapshot
M: Raghavendra Bhat <raghavendra@redhat.com>
-P: Mohammed Rafi KC <rkavunga@redhat.com>
+P: Mohammed Rafi KC <rafi.kavungal@iternity.com>
P: Sunny Kumar <sunkumar@redhat.com>
S: Maintained
F: xlators/mgmt/glusterd/src/glusterd-snap*
F: extras/snap-scheduler.py
Socket subsystem
-M: Raghavendra Gowdappa <rgowdapp@redhat.com>
P: Krutika Dhananjay <kdhananj@redhat.com>
P: Milind Changire <mchangir@redhat.com>
-P: Mohammed Rafi KC <rkavunga@redhat.com>
+P: Mohammed Rafi KC <rafi.kavungal@iternity.com>
P: Mohit Agrawal <moagrawa@redhat.com>
S: Maintained
F: rpc/rpc-transport/socket/
@@ -521,3 +507,4 @@ M: Luis Pabon
M: Ira Cooper
M: Shwetha Panduranga
M: Nithya Balachandran
+M: Raghavendra Gowdappa
diff --git a/Makefile.am b/Makefile.am
index aa0ca5c2a04..98ea5c1038d 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -19,9 +19,13 @@ pkgconfig_DATA = glusterfs-api.pc libgfchangelog.pc
CLEANFILES = glusterfs-api.pc libgfchangelog.pc contrib/umountd/Makefile
+clean-local:
+ find . -name '*.o' -o -name '*.lo' -o -name '.Po' | xargs rm -f
+
gitclean: distclean
find . -name Makefile.in -exec rm -f {} \;
find . -name mount.glusterfs -exec rm -f {} \;
+ find . -name .deps -o -name .libs | xargs rm -rf
rm -fr autom4te.cache
rm -f missing aclocal.m4 config.h.in config.guess config.sub ltmain.sh install-sh configure depcomp
@@ -49,4 +53,3 @@ gen-VERSION:
./build-aux/pkg-version --full \
> $(abs_top_builddir)/$(distdir)/VERSION; \
fi
-
diff --git a/README.md b/README.md
index 92f829e431e..9d68e033782 100644
--- a/README.md
+++ b/README.md
@@ -3,8 +3,7 @@
petabytes. It provides interfaces for object, block and file storage.
## Development
- Contributions to gluster in the form of patches and new feature additions can
- be made by following steps outlined at [Developers Guide](http://docs.gluster.org/en/latest/Developer-guide/Developers-Index/#contributing-to-the-gluster-community).
+ The development workflow is documented in [Contributors guide](CONTRIBUTING.md)
## Documentation
The Gluster documentation can be found at [Gluster Docs](http://docs.gluster.org).
diff --git a/api/src/glfs-fops.c b/api/src/glfs-fops.c
index fc2cfa04309..6aa3c5602d1 100644
--- a/api/src/glfs-fops.c
+++ b/api/src/glfs-fops.c
@@ -1100,6 +1100,11 @@ pub_glfs_read(struct glfs_fd *glfd, void *buf, size_t count, int flags)
};
ssize_t ret = 0;
+ if (glfd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
iov.iov_base = buf;
iov.iov_len = count;
@@ -1151,6 +1156,11 @@ pub_glfs_readv(struct glfs_fd *glfd, const struct iovec *iov, int count,
{
ssize_t ret = 0;
+ if (glfd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
ret = pub_glfs_preadv(glfd, iov, count, glfd->offset, flags);
return ret;
@@ -1387,6 +1397,11 @@ pub_glfs_read_async34(struct glfs_fd *glfd, void *buf, size_t count, int flags,
};
ssize_t ret = 0;
+ if (glfd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
iov.iov_base = buf;
iov.iov_len = count;
@@ -1406,6 +1421,11 @@ pub_glfs_read_async(struct glfs_fd *glfd, void *buf, size_t count, int flags,
};
ssize_t ret = 0;
+ if (glfd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
iov.iov_base = buf;
iov.iov_len = count;
@@ -1460,6 +1480,11 @@ pub_glfs_readv_async34(struct glfs_fd *glfd, const struct iovec *iov, int count,
{
ssize_t ret = 0;
+ if (glfd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
ret = glfs_preadv_async_common(glfd, iov, count, glfd->offset, flags,
_gf_true, (void *)fn, data);
return ret;
@@ -1472,6 +1497,11 @@ pub_glfs_readv_async(struct glfs_fd *glfd, const struct iovec *iov, int count,
{
ssize_t ret = 0;
+ if (glfd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
ret = glfs_preadv_async_common(glfd, iov, count, glfd->offset, flags,
_gf_false, fn, data);
return ret;
@@ -1733,6 +1763,11 @@ pub_glfs_write(struct glfs_fd *glfd, const void *buf, size_t count, int flags)
};
ssize_t ret = 0;
+ if (glfd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
iov.iov_base = (void *)buf;
iov.iov_len = count;
@@ -1748,6 +1783,11 @@ pub_glfs_writev(struct glfs_fd *glfd, const struct iovec *iov, int count,
{
ssize_t ret = 0;
+ if (glfd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
ret = pub_glfs_pwritev(glfd, iov, count, glfd->offset, flags);
return ret;
@@ -1938,6 +1978,11 @@ pub_glfs_write_async34(struct glfs_fd *glfd, const void *buf, size_t count,
};
ssize_t ret = 0;
+ if (glfd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
iov.iov_base = (void *)buf;
iov.iov_len = count;
@@ -1957,6 +2002,11 @@ pub_glfs_write_async(struct glfs_fd *glfd, const void *buf, size_t count,
};
ssize_t ret = 0;
+ if (glfd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
iov.iov_base = (void *)buf;
iov.iov_len = count;
@@ -2011,6 +2061,11 @@ pub_glfs_writev_async34(struct glfs_fd *glfd, const struct iovec *iov,
{
ssize_t ret = 0;
+ if (glfd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
ret = glfs_pwritev_async_common(glfd, iov, count, glfd->offset, flags,
_gf_true, (void *)fn, data);
return ret;
@@ -2023,6 +2078,11 @@ pub_glfs_writev_async(struct glfs_fd *glfd, const struct iovec *iov, int count,
{
ssize_t ret = 0;
+ if (glfd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
ret = glfs_pwritev_async_common(glfd, iov, count, glfd->offset, flags,
_gf_false, fn, data);
return ret;
@@ -3334,6 +3394,11 @@ GFAPI_SYMVER_PUBLIC_DEFAULT(glfs_telldir, 3.4.0)
long
pub_glfs_telldir(struct glfs_fd *fd)
{
+ if (fd == NULL) {
+ errno = EBADF;
+ return -1;
+ }
+
return fd->offset;
}
@@ -3344,6 +3409,11 @@ pub_glfs_seekdir(struct glfs_fd *fd, long offset)
gf_dirent_t *entry = NULL;
gf_dirent_t *tmp = NULL;
+ if (fd == NULL) {
+ errno = EBADF;
+ return;
+ }
+
if (fd->offset == offset)
return;
diff --git a/api/src/glfs-mgmt.c b/api/src/glfs-mgmt.c
index 08df2b1a4b7..7c82b8cd162 100644
--- a/api/src/glfs-mgmt.c
+++ b/api/src/glfs-mgmt.c
@@ -1013,11 +1013,6 @@ glfs_mgmt_init(struct glfs *fs)
if (ret)
goto out;
- if (sys_access(SECURE_ACCESS_FILE, F_OK) == 0) {
- ctx->secure_mgmt = 1;
- ctx->ssl_cert_depth = glusterfs_read_secure_access_file();
- }
-
rpc = rpc_clnt_new(options, THIS, THIS->name, 8);
if (!rpc) {
ret = -1;
diff --git a/api/src/glfs.c b/api/src/glfs.c
index 5259356b4c5..b4bf1423f6d 100644
--- a/api/src/glfs.c
+++ b/api/src/glfs.c
@@ -251,6 +251,11 @@ glfs_volumes_init(struct glfs *fs)
if (!vol_assigned(cmd_args))
return -1;
+ if (sys_access(SECURE_ACCESS_FILE, F_OK) == 0) {
+ fs->ctx->secure_mgmt = 1;
+ fs->ctx->ssl_cert_depth = glusterfs_read_secure_access_file();
+ }
+
if (cmd_args->volfile_server) {
ret = glfs_mgmt_init(fs);
goto out;
@@ -650,6 +655,11 @@ GFAPI_SYMVER_PUBLIC_DEFAULT(glfs_from_glfd, 3.4.0)
struct glfs *
pub_glfs_from_glfd(struct glfs_fd *glfd)
{
+ if (glfd == NULL) {
+ errno = EBADF;
+ return NULL;
+ }
+
return glfd->fs;
}
diff --git a/cli/src/cli-cmd-global.c b/cli/src/cli-cmd-global.c
index ba3c1ea6446..2c9a5f01bb1 100644
--- a/cli/src/cli-cmd-global.c
+++ b/cli/src/cli-cmd-global.c
@@ -27,8 +27,6 @@
#include <glusterfs/syscall.h>
#include <glusterfs/common-utils.h>
-extern rpc_clnt_prog_t *cli_rpc_prog;
-
int
cli_cmd_global_help_cbk(struct cli_state *state, struct cli_cmd_word *in_word,
const char **words, int wordcount);
diff --git a/cli/src/cli-cmd-misc.c b/cli/src/cli-cmd-misc.c
index 658677bbca3..e961d88da86 100644
--- a/cli/src/cli-cmd-misc.c
+++ b/cli/src/cli-cmd-misc.c
@@ -18,10 +18,6 @@
#include "cli-mem-types.h"
#include "protocol-common.h"
-extern struct rpc_clnt *global_rpc;
-
-extern rpc_clnt_prog_t *cli_rpc_prog;
-
extern struct cli_cmd volume_cmds[];
extern struct cli_cmd bitrot_cmds[];
extern struct cli_cmd quota_cmds[];
diff --git a/cli/src/cli-cmd-peer.c b/cli/src/cli-cmd-peer.c
index 342b1d67aaa..084998701d8 100644
--- a/cli/src/cli-cmd-peer.c
+++ b/cli/src/cli-cmd-peer.c
@@ -20,10 +20,6 @@
#include "protocol-common.h"
#include <glusterfs/events.h>
-extern struct rpc_clnt *global_rpc;
-
-extern rpc_clnt_prog_t *cli_rpc_prog;
-
int
cli_cmd_peer_help_cbk(struct cli_state *state, struct cli_cmd_word *in_word,
const char **words, int wordcount);
diff --git a/cli/src/cli-cmd-snapshot.c b/cli/src/cli-cmd-snapshot.c
index 176c6d745b5..859d6b2e40d 100644
--- a/cli/src/cli-cmd-snapshot.c
+++ b/cli/src/cli-cmd-snapshot.c
@@ -17,8 +17,6 @@
#include "cli-cmd.h"
#include "cli-mem-types.h"
-extern rpc_clnt_prog_t *cli_rpc_prog;
-
int
cli_cmd_snapshot_help_cbk(struct cli_state *state, struct cli_cmd_word *in_word,
const char **words, int wordcount);
diff --git a/cli/src/cli-cmd-system.c b/cli/src/cli-cmd-system.c
index d5732630582..801e8f4efed 100644
--- a/cli/src/cli-cmd-system.c
+++ b/cli/src/cli-cmd-system.c
@@ -18,10 +18,6 @@
#include "cli-mem-types.h"
#include "protocol-common.h"
-extern struct rpc_clnt *global_rpc;
-
-extern rpc_clnt_prog_t *cli_rpc_prog;
-
int
cli_cmd_system_help_cbk(struct cli_state *state, struct cli_cmd_word *in_word,
const char **words, int wordcount);
diff --git a/cli/src/cli-cmd-volume.c b/cli/src/cli-cmd-volume.c
index 82b4a7da225..f238851586e 100644
--- a/cli/src/cli-cmd-volume.c
+++ b/cli/src/cli-cmd-volume.c
@@ -28,10 +28,6 @@
#include <glusterfs/common-utils.h>
#include <glusterfs/events.h>
-extern struct rpc_clnt *global_rpc;
-extern struct rpc_clnt *global_quotad_rpc;
-
-extern rpc_clnt_prog_t *cli_rpc_prog;
extern rpc_clnt_prog_t cli_quotad_clnt;
static int
diff --git a/cli/src/cli-cmd.c b/cli/src/cli-cmd.c
index af009757e93..2d458b16a56 100644
--- a/cli/src/cli-cmd.c
+++ b/cli/src/cli-cmd.c
@@ -28,7 +28,7 @@ static pthread_cond_t conn = PTHREAD_COND_INITIALIZER;
static pthread_mutex_t conn_mutex = PTHREAD_MUTEX_INITIALIZER;
int cli_op_ret = 0;
-int connected = 0;
+static gf_boolean_t connected = _gf_false;
static unsigned
cli_cmd_needs_connection(struct cli_cmd_word *word)
@@ -328,19 +328,32 @@ cli_cmd_await_connected(unsigned conn_timo)
}
int32_t
-cli_cmd_broadcast_connected()
+cli_cmd_broadcast_connected(gf_boolean_t status)
{
pthread_mutex_lock(&conn_mutex);
{
- connected = 1;
+ connected = status;
pthread_cond_broadcast(&conn);
}
-
pthread_mutex_unlock(&conn_mutex);
return 0;
}
+gf_boolean_t
+cli_cmd_connected(void)
+{
+ gf_boolean_t status;
+
+ pthread_mutex_lock(&conn_mutex);
+ {
+ status = connected;
+ }
+ pthread_mutex_unlock(&conn_mutex);
+
+ return status;
+}
+
int
cli_cmd_submit(struct rpc_clnt *rpc, void *req, call_frame_t *frame,
rpc_clnt_prog_t *prog, int procnum, struct iobref *iobref,
diff --git a/cli/src/cli-quotad-client.c b/cli/src/cli-quotad-client.c
index 52ab97ee815..772b8f75bd9 100644
--- a/cli/src/cli-quotad-client.c
+++ b/cli/src/cli-quotad-client.c
@@ -10,9 +10,6 @@
#include "cli-quotad-client.h"
-extern struct rpc_clnt global_quotad_rpc;
-extern struct rpc_clnt_program cli_quotad_clnt;
-
int
cli_quotad_submit_request(void *req, call_frame_t *frame, rpc_clnt_prog_t *prog,
int procnum, struct iobref *iobref, xlator_t *this,
@@ -60,7 +57,7 @@ cli_quotad_submit_request(void *req, call_frame_t *frame, rpc_clnt_prog_t *prog,
}
/* Send the msg */
- ret = rpc_clnt_submit(&global_quotad_rpc, prog, procnum, cbkfn, &iov, count,
+ ret = rpc_clnt_submit(global_quotad_rpc, prog, procnum, cbkfn, &iov, count,
NULL, 0, iobref, frame, NULL, 0, NULL, 0, NULL);
ret = 0;
diff --git a/cli/src/cli-rpc-ops.c b/cli/src/cli-rpc-ops.c
index 7c7151568ef..9b6b0c7fa50 100644
--- a/cli/src/cli-rpc-ops.c
+++ b/cli/src/cli-rpc-ops.c
@@ -47,10 +47,7 @@
enum gf_task_types { GF_TASK_TYPE_REBALANCE, GF_TASK_TYPE_REMOVE_BRICK };
-extern struct rpc_clnt *global_quotad_rpc;
rpc_clnt_prog_t cli_quotad_clnt;
-extern rpc_clnt_prog_t *cli_rpc_prog;
-extern int connected;
static int32_t
gf_cli_remove_brick(call_frame_t *frame, xlator_t *this, void *data);
@@ -3406,7 +3403,7 @@ gf_cli_quota_list(cli_local_t *local, char *volname, dict_t *dict,
char *default_sl, int count, int op_ret, int op_errno,
char *op_errstr)
{
- if (!connected)
+ if (!cli_cmd_connected())
goto out;
if (count > 0) {
@@ -4875,14 +4872,6 @@ gf_cli_gsync_status_output(dict_t *dict, gf_boolean_t is_detail)
ret = -1;
goto out;
}
- for (i = 0; i < gsync_count; i++) {
- sts_vals[i] = GF_CALLOC(1, sizeof(gf_gsync_status_t),
- gf_common_mt_char);
- if (!sts_vals[i]) {
- ret = -1;
- goto out;
- }
- }
ret = gf_cli_read_status_data(dict, sts_vals, spacing, gsync_count,
num_of_fields);
@@ -5352,7 +5341,8 @@ cmd_profile_volume_brick_out(dict_t *dict, int count, int interval)
double total_percentage_latency = 0;
for (i = 0; i < 32; i++) {
- snprintf(key, sizeof(key), "%d-%d-read-%d", count, interval, (1 << i));
+ snprintf(key, sizeof(key), "%d-%d-read-%" PRIu32, count, interval,
+ (1U << i));
ret = dict_get_uint64(dict, key, &rb_counts[i]);
if (ret) {
gf_log("cli", GF_LOG_DEBUG, "failed to get %s from dict", key);
@@ -5360,7 +5350,8 @@ cmd_profile_volume_brick_out(dict_t *dict, int count, int interval)
}
for (i = 0; i < 32; i++) {
- snprintf(key, sizeof(key), "%d-%d-write-%d", count, interval, (1 << i));
+ snprintf(key, sizeof(key), "%d-%d-write-%" PRIu32, count, interval,
+ (1U << i));
ret = dict_get_uint64(dict, key, &wb_counts[i]);
if (ret) {
gf_log("cli", GF_LOG_DEBUG, "failed to get %s from dict", key);
@@ -5446,7 +5437,8 @@ cmd_profile_volume_brick_out(dict_t *dict, int count, int interval)
if ((rb_counts[i] == 0) && (wb_counts[i] == 0))
continue;
per_line++;
- snprintf(output + index, sizeof(output) - index, "%19db+ ", (1 << i));
+ snprintf(output + index, sizeof(output) - index, "%19" PRIu32 "b+ ",
+ (1U << i));
if (rb_counts[i]) {
snprintf(read_blocks + index, sizeof(read_blocks) - index,
"%21" PRId64 " ", rb_counts[i]);
@@ -5733,7 +5725,7 @@ gf_cli_top_volume_cbk(struct rpc_req *req, struct iovec *iov, int count,
double time = 0;
int32_t time_sec = 0;
long int time_usec = 0;
- char timestr[256] = {
+ char timestr[GF_TIMESTR_SIZE] = {
0,
};
char *openfd_str = NULL;
@@ -7823,7 +7815,7 @@ cmd_heal_volume_brick_out(dict_t *dict, int brick)
char *status = NULL;
uint64_t i = 0;
uint32_t time = 0;
- char timestr[32] = {0};
+ char timestr[GF_TIMESTR_SIZE] = {0};
char *shd_status = NULL;
snprintf(key, sizeof key, "%d-hostname", brick);
diff --git a/cli/src/cli-xml-output.c b/cli/src/cli-xml-output.c
index 5f33aa9e0a3..069de75801c 100644
--- a/cli/src/cli-xml-output.c
+++ b/cli/src/cli-xml-output.c
@@ -1661,15 +1661,15 @@ cli_xml_output_vol_top_rw_perf(xmlTextWriterPtr writer, dict_t *dict,
int ret = -1;
char *filename = NULL;
uint64_t throughput = 0;
- long int time_sec = 0;
- long int time_usec = 0;
- char timestr[256] = {
+ struct timeval tv = {
+ 0,
+ };
+ char timestr[GF_TIMESTR_SIZE] = {
0,
};
char key[1024] = {
0,
};
- int len;
/* <file> */
ret = xmlTextWriterStartElement(writer, (xmlChar *)"file");
@@ -1692,19 +1692,16 @@ cli_xml_output_vol_top_rw_perf(xmlTextWriterPtr writer, dict_t *dict,
XML_RET_CHECK_AND_GOTO(ret, out);
snprintf(key, sizeof(key), "%d-time-sec-%d", brick_index, member_index);
- ret = dict_get_int32(dict, key, (int32_t *)&time_sec);
+ ret = dict_get_int32(dict, key, (int32_t *)&tv.tv_sec);
if (ret)
goto out;
snprintf(key, sizeof(key), "%d-time-usec-%d", brick_index, member_index);
- ret = dict_get_int32(dict, key, (int32_t *)&time_usec);
+ ret = dict_get_int32(dict, key, (int32_t *)&tv.tv_usec);
if (ret)
goto out;
- gf_time_fmt(timestr, sizeof timestr, time_sec, gf_timefmt_FT);
- len = strlen(timestr);
- snprintf(timestr + len, sizeof(timestr) - len, ".%" GF_PRI_SUSECONDS,
- time_usec);
+ gf_time_fmt_tv(timestr, sizeof timestr, &tv, gf_timefmt_FT);
ret = xmlTextWriterWriteFormatElement(writer, (xmlChar *)"time", "%s",
timestr);
XML_RET_CHECK_AND_GOTO(ret, out);
@@ -1953,11 +1950,11 @@ cli_xml_output_vol_profile_stats(xmlTextWriterPtr writer, dict_t *dict,
XML_RET_CHECK_AND_GOTO(ret, out);
ret = xmlTextWriterWriteFormatElement(writer, (xmlChar *)"size",
- "%" PRIu32, (1 << i));
+ "%" PRIu32, (1U << i));
XML_RET_CHECK_AND_GOTO(ret, out);
- snprintf(key, sizeof(key), "%d-%d-read-%d", brick_index, interval,
- (1 << i));
+ snprintf(key, sizeof(key), "%d-%d-read-%" PRIu32, brick_index, interval,
+ (1U << i));
ret = dict_get_uint64(dict, key, &read_count);
if (ret)
read_count = 0;
@@ -1965,8 +1962,8 @@ cli_xml_output_vol_profile_stats(xmlTextWriterPtr writer, dict_t *dict,
"%" PRIu64, read_count);
XML_RET_CHECK_AND_GOTO(ret, out);
- snprintf(key, sizeof(key), "%d-%d-write-%d", brick_index, interval,
- (1 << i));
+ snprintf(key, sizeof(key), "%d-%d-write-%" PRIu32, brick_index,
+ interval, (1U << i));
ret = dict_get_uint64(dict, key, &write_count);
if (ret)
write_count = 0;
diff --git a/cli/src/cli.c b/cli/src/cli.c
index 2c96b7ae6e6..a52b39c5fb8 100644
--- a/cli/src/cli.c
+++ b/cli/src/cli.c
@@ -33,12 +33,6 @@
#include <malloc.h>
#endif
-#ifdef HAVE_MALLOC_STATS
-#ifdef DEBUG
-#include <mcheck.h>
-#endif
-#endif
-
#include "cli.h"
#include "cli-quotad-client.h"
#include "cli-cmd.h"
@@ -61,7 +55,6 @@
#include "xdr-generic.h"
-extern int connected;
/* using argp for command line parsing */
const char *argp_program_version =
@@ -78,6 +71,7 @@ const char *argp_program_version =
const char *argp_program_bug_address = "<" PACKAGE_BUGREPORT ">";
struct rpc_clnt *global_quotad_rpc;
+
struct rpc_clnt *global_rpc;
rpc_clnt_prog_t *cli_rpc_prog;
@@ -309,14 +303,14 @@ cli_rpc_notify(struct rpc_clnt *rpc, void *mydata, rpc_clnt_event_t event,
switch (event) {
case RPC_CLNT_CONNECT: {
- cli_cmd_broadcast_connected();
+ cli_cmd_broadcast_connected(_gf_true);
gf_log(this->name, GF_LOG_TRACE, "got RPC_CLNT_CONNECT");
break;
}
case RPC_CLNT_DISCONNECT: {
+ cli_cmd_broadcast_connected(_gf_false);
gf_log(this->name, GF_LOG_TRACE, "got RPC_CLNT_DISCONNECT");
- connected = 0;
if (!global_state->prompt && global_state->await_connected) {
ret = 1;
cli_out(
diff --git a/cli/src/cli.h b/cli/src/cli.h
index cd4db3dac2c..c0d933e8f8a 100644
--- a/cli/src/cli.h
+++ b/cli/src/cli.h
@@ -189,6 +189,12 @@ typedef ssize_t (*cli_serialize_t)(struct iovec outmsg, void *args);
extern struct cli_state *global_state; /* use only in readline callback */
+extern struct rpc_clnt *global_quotad_rpc;
+
+extern struct rpc_clnt *global_rpc;
+
+extern rpc_clnt_prog_t *cli_rpc_prog;
+
typedef const char *(*cli_selector_t)(void *wcon);
char *
@@ -329,11 +335,14 @@ cli_local_get();
void
cli_local_wipe(cli_local_t *local);
+gf_boolean_t
+cli_cmd_connected();
+
int32_t
-cli_cmd_await_connected();
+cli_cmd_await_connected(unsigned timeout);
int32_t
-cli_cmd_broadcast_connected();
+cli_cmd_broadcast_connected(gf_boolean_t status);
int
cli_rpc_notify(struct rpc_clnt *rpc, void *mydata, rpc_clnt_event_t event,
diff --git a/configure.ac b/configure.ac
index 0d236f6c359..e2d6fd66cec 100644
--- a/configure.ac
+++ b/configure.ac
@@ -164,6 +164,8 @@ AC_CONFIG_FILES([Makefile
xlators/features/cloudsync/src/cloudsync-plugins/src/cloudsyncs3/src/Makefile
xlators/features/cloudsync/src/cloudsync-plugins/src/cvlt/Makefile
xlators/features/cloudsync/src/cloudsync-plugins/src/cvlt/src/Makefile
+ xlators/features/metadisp/Makefile
+ xlators/features/metadisp/src/Makefile
xlators/playground/Makefile
xlators/playground/template/Makefile
xlators/playground/template/src/Makefile
@@ -273,36 +275,65 @@ AC_ARG_ENABLE([debug],
[Enable debug build options.]))
if test "x$enable_debug" = "xyes"; then
BUILD_DEBUG=yes
- GF_CFLAGS="${GF_CFLAGS} -g -rdynamic -O0 -DDEBUG"
+ GF_CFLAGS="${GF_CFLAGS} -g -O0 -DDEBUG"
else
BUILD_DEBUG=no
fi
+SANITIZER=none
+
AC_ARG_ENABLE([asan],
AC_HELP_STRING([--enable-asan],
[Enable Address Sanitizer support]))
if test "x$enable_asan" = "xyes"; then
- BUILD_ASAN=yes
- AC_CHECK_LIB([asan], [__asan_report_error], ,
- [AC_MSG_ERROR([libasan.so not found, this is required for --enable-asan])])
- GF_CFLAGS="${GF_CFLAGS} -O1 -g -fsanitize=address -fno-omit-frame-pointer"
- dnl -lasan always need to be the first library, otherwise libxml complains
- GF_LDFLAGS="-lasan ${GF_LDFLAGS}"
-else
- BUILD_ASAN=no
+ SANITIZER=asan
+ AC_CHECK_LIB([asan], [__asan_init], ,
+ [AC_MSG_ERROR([--enable-asan requires libasan.so, exiting])])
+ GF_CFLAGS="${GF_CFLAGS} -O2 -g -fsanitize=address -fno-omit-frame-pointer"
+ GF_LDFLAGS="${GF_LDFLAGS} -lasan"
fi
-AC_ARG_ENABLE([atan],
+AC_ARG_ENABLE([tsan],
AC_HELP_STRING([--enable-tsan],
- [Enable ThreadSanitizer support]))
+ [Enable Thread Sanitizer support]))
if test "x$enable_tsan" = "xyes"; then
- BUILD_TSAN=yes
+ if test "x$SANITIZER" != "xnone"; then
+ AC_MSG_ERROR([only one sanitizer can be enabled at once])
+ fi
+ SANITIZER=tsan
AC_CHECK_LIB([tsan], [__tsan_init], ,
- [AC_MSG_ERROR([libtsan.so not found, this is required for --enable-tsan])])
- GF_CFLAGS="${GF_CFLAGS} -O2 -g -fsanitize=thread"
+ [AC_MSG_ERROR([--enable-tsan requires libtsan.so, exiting])])
+ if test "x$ac_cv_lib_tsan___tsan_init" = xyes; then
+ AC_MSG_CHECKING([whether tsan API can be used])
+ saved_CFLAGS=${CFLAGS}
+ CFLAGS="${CFLAGS} -fsanitize=thread"
+ AC_COMPILE_IFELSE(
+ [AC_LANG_PROGRAM([
+ [#include <sanitizer/tsan_interface.h>]],
+ [[__tsan_create_fiber(0)]])],
+ [TSAN_API=yes], [TSAN_API=no])
+ AC_MSG_RESULT([$TSAN_API])
+ if test x$TSAN_API = "xyes"; then
+ AC_DEFINE(HAVE_TSAN_API, 1, [Define if tsan API can be used.])
+ fi
+ CFLAGS=${saved_CFLAGS}
+ fi
+ GF_CFLAGS="${GF_CFLAGS} -O2 -g -fsanitize=thread -fno-omit-frame-pointer"
GF_LDFLAGS="${GF_LDFLAGS} -ltsan"
-else
- BUILD_TSAN=no
+fi
+
+AC_ARG_ENABLE([ubsan],
+ AC_HELP_STRING([--enable-ubsan],
+ [Enable Undefined Behavior Sanitizer support]))
+if test "x$enable_ubsan" = "xyes"; then
+ if test "x$SANITIZER" != "xnone"; then
+ AC_MSG_ERROR([only one sanitizer can be enabled at once])
+ fi
+ SANITIZER=ubsan
+ AC_CHECK_LIB([ubsan], [__ubsan_default_options], ,
+ [AC_MSG_ERROR([--enable-ubsan requires libubsan.so, exiting])])
+ GF_CFLAGS="${GF_CFLAGS} -O2 -g -fsanitize=undefined -fno-omit-frame-pointer"
+ GF_LDFLAGS="${GF_LDFLAGS} -lubsan"
fi
# Initialize CFLAGS before usage
@@ -378,12 +409,27 @@ esac
# --enable-valgrind prevents calling dlclose(), this leaks memory
AC_ARG_ENABLE([valgrind],
- AC_HELP_STRING([--enable-valgrind],
- [Enable valgrind for resource leak debugging.]))
-if test "x$enable_valgrind" = "xyes"; then
- AC_DEFINE(RUN_WITH_VALGRIND, 1, [define if all processes should run under valgrind])
-fi
-
+ AC_HELP_STRING([--enable-valgrind@<:@=memcheck,drd@:>@],
+ [Enable valgrind for resource leak (memcheck, which is
+ the default) or thread synchronization (drd) debugging.]))
+case x$enable_valgrind in
+ xmemcheck|xyes)
+ AC_DEFINE(RUN_WITH_MEMCHECK, 1,
+ [Define if all processes should run under 'valgrind --tool=memcheck'.])
+ VALGRIND_TOOL=memcheck
+ ;;
+ xdrd)
+ AC_DEFINE(RUN_WITH_DRD, 1,
+ [Define if all processes should run under 'valgrind --tool=drd'.])
+ VALGRIND_TOOL=drd
+ ;;
+ x|xno)
+ VALGRIND_TOOL=no
+ ;;
+ *)
+ AC_MSG_ERROR([Please specify --enable-valgrind@<:@=memcheck,drd@:>@])
+ ;;
+esac
AC_ARG_WITH([previous-options],
[AS_HELP_STRING([--with-previous-options],
@@ -684,6 +730,14 @@ if test "x$enable_fuse_client" != "xno"; then
fi
AC_SUBST(FUSE_CLIENT_SUBDIR)
+
+AC_ARG_ENABLE([fuse-notifications],
+ AS_HELP_STRING([--disable-fuse-notifications], [Disable FUSE notifications]))
+
+AS_IF([test "x$enable_fuse_notifications" != "xno"], [
+ AC_DEFINE([HAVE_FUSE_NOTIFICATIONS], [1], [Use FUSE notifications])
+])
+
# end FUSE section
@@ -731,6 +785,9 @@ AC_ARG_ENABLE([georeplication],
BUILD_SYNCDAEMON=no
case $host_os in
+ freebsd*)
+#do nothing
+ ;;
linux*)
#do nothing
;;
@@ -770,6 +827,17 @@ fi
AC_SUBST(GEOREP_EXTRAS_SUBDIR)
AM_CONDITIONAL(USE_GEOREP, test "x$enable_georeplication" != "xno")
+# METADISP section
+AC_ARG_ENABLE([metadisp],
+ AC_HELP_STRING([--enable-metadisp],
+ [Enable the metadata dispersal xlator]))
+BUILD_METADISP=no
+if test "x${enable_metadisp}" = "xyes"; then
+ BUILD_METADISP=yes
+fi
+AM_CONDITIONAL([BUILD_METADISP], [test "x$BUILD_METADISP" = "xyes"])
+# end METADISP section
+
# Events section
AC_ARG_ENABLE([events],
AC_HELP_STRING([--disable-events],
@@ -908,6 +976,25 @@ if test "x${have_backtrace}" = "xyes"; then
fi
AC_SUBST(HAVE_BACKTRACE)
+dnl Old (before C11) compiler can compile (but not link) this:
+dnl
+dnl int main () {
+dnl _Static_assert(1, "True");
+dnl return 0;
+dnl }
+dnl
+dnl assuming that _Static_assert is an implicitly declared function. So
+dnl we're trying to link just to make sure that this is not the case.
+
+AC_MSG_CHECKING([whether $CC supports C11 _Static_assert])
+AC_TRY_LINK([], [_Static_assert(1, "True");],
+ [STATIC_ASSERT=yes], [STATIC_ASSERT=no])
+
+AC_MSG_RESULT([$STATIC_ASSERT])
+if test x$STATIC_ASSERT = "xyes"; then
+ AC_DEFINE(HAVE_STATIC_ASSERT, 1, [Define if C11 _Static_assert is supported.])
+fi
+
if test "x${have_backtrace}" != "xyes"; then
AC_TRY_COMPILE([#include <math.h>], [double x=0.0; x=ceil(0.0);],
[],
@@ -915,11 +1002,11 @@ AC_TRY_COMPILE([#include <math.h>], [double x=0.0; x=ceil(0.0);],
fi
dnl glusterfs prints memory usage to stderr by sending it SIGUSR1
-AC_CHECK_FUNC([malloc_stats], [have_malloc_stats=yes])
-if test "x${have_malloc_stats}" = "xyes"; then
- AC_DEFINE(HAVE_MALLOC_STATS, 1, [define if found malloc_stats])
+AC_CHECK_FUNC([mallinfo], [have_mallinfo=yes])
+if test "x${have_mallinfo}" = "xyes"; then
+ AC_DEFINE(HAVE_MALLINFO, 1, [define if found mallinfo])
fi
-AC_SUBST(HAVE_MALLOC_STATS)
+AC_SUBST(HAVE_MALLINFO)
dnl Linux, Solaris, Cygwin
AC_CHECK_MEMBERS([struct stat.st_atim.tv_nsec])
@@ -946,6 +1033,9 @@ case $host_os in
CFLAGS="${CFLAGS} -isystem /usr/local/include"
ARGP_LDADD=-largp
;;
+ *netbsd*)
+ ARGP_LDADD=-largp
+ ;;
esac
dnl argp-standalone does not provide a pkg-config file
AC_CHECK_HEADER([argp.h], AC_DEFINE(HAVE_ARGP, 1, [have argp]))
@@ -1015,6 +1105,19 @@ else
CFLAGS=${OLD_CFLAGS}
fi
+AC_CHECK_FUNC([syncfs], [have_syncfs=yes])
+if test "x${have_syncfs}" = "xyes"; then
+ AC_DEFINE(HAVE_SYNCFS, 1, [define if syncfs exists])
+else
+ OLD_CFLAGS=${CFLAGS}
+ CFLAGS="-D_GNU_SOURCE"
+ AC_CHECK_DECL([SYS_syncfs], , , [#include <sys/syscall.h>])
+ if test "x${ac_cv_have_decl_SYS_syncfs}" = "xyes"; then
+ AC_DEFINE(HAVE_SYNCFS_SYS, 1, [define if SYS_syncfs is available])
+ fi
+ CFLAGS=${OLD_CFLAGS}
+fi
+
BUILD_NANOSECOND_TIMESTAMPS=no
AC_CHECK_FUNC([utimensat], [have_utimensat=yes])
if test "x${have_utimensat}" = "xyes"; then
@@ -1207,10 +1310,6 @@ case $host_os in
;;
*freebsd*)
GF_HOST_OS="GF_BSD_HOST_OS"
- GF_CFLAGS="${GF_CFLAGS} -O0"
- GF_CFLAGS="${GF_CFLAGS} -DTHREAD_UNSAFE_BASENAME"
- GF_CFLAGS="${GF_CFLAGS} -DTHREAD_UNSAFE_DIRNAME"
- GF_CFLAGS="${GF_CFLAGS} -D_LIBGEN_H_"
GF_CFLAGS="${GF_CFLAGS} -DO_DSYNC=0"
GF_CFLAGS="${GF_CFLAGS} -Dxdr_quad_t=xdr_longlong_t"
GF_CFLAGS="${GF_CFLAGS} -Dxdr_u_quad_t=xdr_u_longlong_t"
@@ -1491,9 +1590,9 @@ case $host_os in
;;
esac
dnl GF_XLATOR_DEFAULT_LDFLAGS is for most xlators that expose a common set of symbols
-GF_XLATOR_DEFAULT_LDFLAGS='-avoid-version -export-symbols $(top_srcdir)/xlators/xlator.sym $(UUID_LIBS) $(GF_NO_UNDEFINED)'
+GF_XLATOR_DEFAULT_LDFLAGS='-avoid-version -export-symbols $(top_srcdir)/xlators/xlator.sym $(UUID_LIBS) $(GF_NO_UNDEFINED) $(TIRPC_LIBS)'
dnl GF_XLATOR_LDFLAGS is for xlators that expose extra symbols, e.g. dht
-GF_XLATOR_LDFLAGS='-avoid-version $(UUID_LIBS) $(GF_NO_UNDEFINED)'
+GF_XLATOR_LDFLAGS='-avoid-version $(UUID_LIBS) $(GF_NO_UNDEFINED) $(TIRPC_LIBS)'
AC_SUBST(GF_HOST_OS)
AC_SUBST(GF_CFLAGS)
@@ -1508,6 +1607,13 @@ AC_SUBST(AM_LIBTOOLFLAGS)
AC_SUBST(GF_NO_UNDEFINED)
AC_SUBST(GF_XLATOR_DEFAULT_LDFLAGS)
AC_SUBST(GF_XLATOR_LDFLAGS)
+AC_SUBST(GF_XLATOR_MGNT_LIBADD)
+
+case $host_os in
+ *freebsd*)
+ GF_XLATOR_MGNT_LIBADD="-lutil -lprocstat"
+ ;;
+esac
CONTRIBDIR='$(top_srcdir)/contrib'
AC_SUBST(CONTRIBDIR)
@@ -1579,8 +1685,8 @@ echo "readline : $BUILD_READLINE"
echo "georeplication : $BUILD_SYNCDAEMON"
echo "Linux-AIO : $BUILD_LIBAIO"
echo "Enable Debug : $BUILD_DEBUG"
-echo "Enable ASAN : $BUILD_ASAN"
-echo "Enable TSAN : $BUILD_TSAN"
+echo "Run with Valgrind : $VALGRIND_TOOL"
+echo "Sanitizer enabled : $SANITIZER"
echo "Use syslog : $USE_SYSLOG"
echo "XML output : $BUILD_XML_OUTPUT"
echo "Unit Tests : $BUILD_UNITTEST"
@@ -1598,17 +1704,14 @@ echo "IPV6 default : $with_ipv6_default"
echo "Use TIRPC : $with_libtirpc"
echo "With Python : ${PYTHON_VERSION}"
echo "Cloudsync : $BUILD_CLOUDSYNC"
+echo "Metadata dispersal : $BUILD_METADISP"
echo "Link with TCMALLOC : $BUILD_TCMALLOC"
echo
-if test "x$BUILD_ASAN" = "xyes"; then
- echo "### Run below command before executing your tests if your system"
- echo "### has 'gcc --version' above 7.x (works on Fedora 27 and Above)"
- echo "export ASAN_OPTIONS=log_path=/var/log/glusterfs/asan-output.log"
- echo ""
- echo "### Above is required to get details of asan run, as glusterfs"
- echo "### processes are daemon processes. Further details and more"
- echo "### options can be found under 'Run-time flags' at"
- echo "### https://github.com/google/sanitizers/wiki/AddressSanitizerFlags"
- echo
+# dnl Note: ${X^^} capitalization assumes bash >= 4.x
+if test "x$SANITIZER" != "xnone"; then
+ echo "Note: since glusterfs processes are daemon processes, use"
+ echo "'export ${SANITIZER^^}_OPTIONS=log_path=/path/to/xxx.log' to collect"
+ echo "sanitizer output. Further details and more options can be"
+ echo "found at https://github.com/google/sanitizers."
fi
diff --git a/contrib/fuse-lib/mount.c b/contrib/fuse-lib/mount.c
index 351972ce3cf..06ff191f542 100644
--- a/contrib/fuse-lib/mount.c
+++ b/contrib/fuse-lib/mount.c
@@ -390,6 +390,7 @@ fuse_mount_sys (const char *mountpoint, char *fsname,
build_iovec (&iov, &iovlen, "from", "/dev/fuse", -1);
build_iovec (&iov, &iovlen, "volname", source, -1);
build_iovec (&iov, &iovlen, "fd", fdstr, -1);
+ build_iovec (&iov, &iovlen, "allow_other", NULL, -1);
ret = nmount (iov, iovlen, mountflags);
#else
ret = mount (source, mountpoint, fstype, mountflags,
diff --git a/contrib/sunrpc/xdr_sizeof.c b/contrib/sunrpc/xdr_sizeof.c
deleted file mode 100644
index ca1f7bf0a5e..00000000000
--- a/contrib/sunrpc/xdr_sizeof.c
+++ /dev/null
@@ -1,204 +0,0 @@
-/*
- * Copyright (c) 1999 Apple Computer, Inc. All rights reserved.
- *
- * @APPLE_LICENSE_HEADER_START@
- *
- * Portions Copyright (c) 1999 Apple Computer, Inc. All Rights
- * Reserved. This file contains Original Code and/or Modifications of
- * Original Code as defined in and that are subject to the Apple Public
- * Source License Version 1.1 (the "License"). You may not use this file
- * except in compliance with the License. Please obtain a copy of the
- * License at http://www.apple.com/publicsource and read it before using
- * this file.
- *
- * The Original Code and all software distributed under the License are
- * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
- * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
- * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE OR NON- INFRINGEMENT. Please see the
- * License for the specific language governing rights and limitations
- * under the License.
- *
- * @APPLE_LICENSE_HEADER_END@
- */
-
-/*
- * Sun RPC is a product of Sun Microsystems, Inc. and is provided for
- * unrestricted use provided that this legend is included on all tape
- * media and as a part of the software program in whole or part. Users
- * may copy or modify Sun RPC without charge, but are not authorized
- * to license or distribute it to anyone else except as part of a product or
- * program developed by the user.
- *
- * SUN RPC IS PROVIDED AS IS WITH NO WARRANTIES OF ANY KIND INCLUDING THE
- * WARRANTIES OF DESIGN, MERCHANTIBILITY AND FITNESS FOR A PARTICULAR
- * PURPOSE, OR ARISING FROM A COURSE OF DEALING, USAGE OR TRADE PRACTICE.
- *
- * Sun RPC is provided with no support and without any obligation on the
- * part of Sun Microsystems, Inc. to assist in its use, correction,
- * modification or enhancement.
- *
- * SUN MICROSYSTEMS, INC. SHALL HAVE NO LIABILITY WITH RESPECT TO THE
- * INFRINGEMENT OF COPYRIGHTS, TRADE SECRETS OR ANY PATENTS BY SUN RPC
- * OR ANY PART THEREOF.
- *
- * In no event will Sun Microsystems, Inc. be liable for any lost revenue
- * or profits or other special, indirect and consequential damages, even if
- * Sun has been advised of the possibility of such damages.
- *
- * Sun Microsystems, Inc.
- * 2550 Garcia Avenue
- * Mountain View, California 94043
- */
-
-/*
- * xdr_sizeof.c
- *
- * Copyright 1990 Sun Microsystems, Inc.
- *
- * General purpose routine to see how much space something will use
- * when serialized using XDR.
- */
-
-#ifdef GF_DARWIN_HOST_OS
-
-#include <rpc/types.h>
-#include <rpc/xdr.h>
-#include <sys/types.h>
-#include <sys/cdefs.h>
-
-#include <stdlib.h>
-
-/* ARGSUSED */
-#ifdef GF_DARWIN_HOST_OS
-static bool_t
-x_putlong (XDR *xdrs, const int *longp)
-{
- xdrs->x_handy += BYTES_PER_XDR_UNIT;
- return TRUE;
-}
-
-#else
-static bool_t
-x_putlong (XDR *xdrs, const long *longp)
-{
- xdrs->x_handy += BYTES_PER_XDR_UNIT;
- return TRUE;
-}
-#endif
-
-/* ARGSUSED */
-static bool_t
-x_putbytes (XDR *xdrs, const char *bp, u_int len)
-{
- xdrs->x_handy += len;
- return TRUE;
-}
-
-#ifdef GF_DARWIN_HOST_OS
-static u_int
-x_getpostn (XDR *xdrs)
-{
- return xdrs->x_handy;
-}
-#else
-static u_int
-x_getpostn (const XDR *xdrs)
-{
- return xdrs->x_handy;
-}
-#endif
-
-/* ARGSUSED */
-static bool_t
-x_setpostn (XDR *xdrs, u_int len)
-{
- /* This is not allowed */
- return FALSE;
-}
-
-static int32_t *
-x_inline (XDR *xdrs, u_int len)
-{
- if (len == 0)
- return NULL;
- if (xdrs->x_op != XDR_ENCODE)
- return NULL;
- if (len < (u_int) (long int) xdrs->x_base)
- {
- /* x_private was already allocated */
- xdrs->x_handy += len;
- return (int32_t *) xdrs->x_private;
- }
- else
- {
- /* Free the earlier space and allocate new area */
- free (xdrs->x_private);
- if ((xdrs->x_private = (caddr_t) malloc (len)) == NULL)
- {
- xdrs->x_base = 0;
- return NULL;
- }
- xdrs->x_base = (void *) (long) len;
- xdrs->x_handy += len;
- return (int32_t *) xdrs->x_private;
- }
-}
-
-static int
-harmless (void)
-{
- /* Always return FALSE/NULL, as the case may be */
- return 0;
-}
-
-static void
-x_destroy (XDR *xdrs)
-{
- xdrs->x_handy = 0;
- xdrs->x_base = 0;
- if (xdrs->x_private)
- {
- free (xdrs->x_private);
- xdrs->x_private = NULL;
- }
- return;
-}
-
-unsigned long
-xdr_sizeof (xdrproc_t func, void *data)
-{
- XDR x;
- struct xdr_ops ops;
- bool_t stat;
-
-#ifdef GF_DARWIN_HOST_OS
- typedef bool_t (*dummyfunc1) (XDR *, int *);
-#else
- typedef bool_t (*dummyfunc1) (XDR *, long *);
-#endif
- typedef bool_t (*dummyfunc2) (XDR *, caddr_t, u_int);
-
- ops.x_putlong = x_putlong;
- ops.x_putbytes = x_putbytes;
- ops.x_inline = x_inline;
- ops.x_getpostn = x_getpostn;
- ops.x_setpostn = x_setpostn;
- ops.x_destroy = x_destroy;
-
- /* the other harmless ones */
- ops.x_getlong = (dummyfunc1) harmless;
- ops.x_getbytes = (dummyfunc2) harmless;
-
- x.x_op = XDR_ENCODE;
- x.x_ops = &ops;
- x.x_handy = 0;
- x.x_private = (caddr_t) NULL;
- x.x_base = (caddr_t) 0;
-
- stat = func (&x, data, 0);
- if (x.x_private)
- free (x.x_private);
- return (stat == TRUE ? (unsigned) x.x_handy : 0);
-}
-#endif /* GF_DARWIN_HOST_OS */
diff --git a/doc/developer-guide/fuse-interrupt.md b/doc/developer-guide/fuse-interrupt.md
index f92b5532eaf..ec991b81ec5 100644
--- a/doc/developer-guide/fuse-interrupt.md
+++ b/doc/developer-guide/fuse-interrupt.md
@@ -23,9 +23,10 @@ not exported to a header file).
```
enum fuse_interrupt_state {
- INTERRUPT_NONE,
+ /* ... */
INTERRUPT_SQUELCHED,
INTERRUPT_HANDLED,
+ /* ... */
};
typedef enum fuse_interrupt_state fuse_interrupt_state_t;
struct fuse_interrupt_record;
@@ -62,8 +63,58 @@ dummy implementation only for demonstration purposes.) Flush is chosen
because a `FLUSH` interrupt is easy to trigger (see
*tests/features/interrupt.t*). Interrupt handling for flush is switched on
by `--fuse-flush-handle-interrupt` (a hidden glusterfs command line flag).
-The flush interrupt handling code is guarded by the
-`flush_handle_interrupt` Boolean member of `fuse_private_t`.
+The implementation of flush interrupt is contained in the
+`fuse_flush_interrupt_handler()` function and blocks guarded by the
+
+```
+if (priv->flush_handle_interrupt) { ...
+```
+
+conditional (where `priv` is a `*fuse_private_t`).
+
+### Overview
+
+"Regular" fuse fops and interrupt handlers interact via a list containing
+interrupt records.
+
+If a fop wishes to have its interrupts handled, it needs to set up an
+interrupt record and insert it into the list; also when it's to finish
+(ie. in its "cbk" stage) it needs to delete the record from the list.
+
+If no interrupt happens, basically that's all to it - a list insertion
+and deletion.
+
+However, if an interrupt comes for the fop, the interrupt FUSE request
+will carry the data identifying an ongoing fop (that is, its `unique`),
+and based on that, the interrupt record will be looked up in the list, and
+the specific interrupt handler (a member of the interrupt record) will be
+called.
+
+Usually the fop needs to share some data with the interrupt handler to
+enable it to perform its task (also shared via the interrupt record).
+The interrupt API offers two approaches to manage shared data:
+- _Async or reference-counting strategy_: from the point on when the interrupt
+ record is inserted to the list, it's owned jointly by the regular fop and
+ the prospective interrupt handler. Both of them need to check before they
+ return if the other is still holding a reference; if not, then they are
+ responsible for reclaiming the shared data.
+- _Sync or borrow strategy_: the interrupt handler is considered a borrower
+ of the shared data. The interrupt handler should not reclaim the shared
+ data. The fop will wait for the interrupt handler to finish (ie., the borrow
+ to be returned), then it has to reclaim the shared data.
+
+The user of the interrupt API need to call the following functions to
+instrument this control flow:
+- `fuse_interrupt_record_insert()` in the fop to insert the interrupt record to
+ the list;
+- `fuse_interrupt_finish_fop()`in the fop (cbk) and
+- `fuse_interrupt_finish_interrupt()`in the interrupt handler
+
+to perform needed synchronization at the end their tenure. The data management
+strategies are implemented by the `fuse_interrupt_finish_*()` functions (which
+have an argument to specify which strategy to use); these routines take care
+of freeing the interrupt record itself, while the reclamation of the shared data
+is left to the API user.
### Usage
@@ -75,12 +126,15 @@ steps:
call (directly or as async callback) `fuse_interrupt_finish_interrupt()`.
The `intstat` argument to `fuse_interrupt_finish_interrupt` should be
either `INTERRUPT_SQUELCHED` or `INTERRUPT_HANDLED`.
- - `INTERRUPT_SQUELCHED` means that we choose not to handle the interrupt
+ - `INTERRUPT_SQUELCHED` means that the interrupt could not be delivered
and the fop is going on uninterrupted.
- `INTERRUPT_HANDLED` means that the interrupt was actually handled. In
this case the fop will be answered from interrupt context with errno
`EINTR` (that is, the fop should not send a response to the kernel).
+ (the enum `fuse_interrupt_state` includes further members, which are reserved
+ for internal use).
+
We return to the `sync` and `datap` arguments later.
- In the `fuse_<FOP>` function create an interrupt record using
`fuse_interrupt_record_new()`, passing the incoming `fuse_in_header` and
@@ -92,10 +146,10 @@ steps:
`fuse_interrupt_record_insert()`.
- In `fuse_<FOP>_cbk` call `fuse_interrupt_finish_fop()`.
- `fuse_interrupt_finish_fop()` returns a Boolean according to whether the
- interrupt was handled. If it was, then the fuse request is already
+ interrupt was handled. If it was, then the FUSE request is already
answered and the stack gets destroyed in `fuse_interrupt_finish_fop` so
- `fuse_<FOP>_cbk` can just return (zero). Otherwise follow the standard
- cbk logic (answer the fuse request and destroy the stack -- these are
+ `fuse_<FOP>_cbk()` can just return (zero). Otherwise follow the standard
+ cbk logic (answer the FUSE request and destroy the stack -- these are
typically accomplished by `fuse_err_cbk()`).
- The last two argument of `fuse_interrupt_finish_fop()` and
`fuse_interrupt_finish_interrupt()` are `gf_boolean_t sync` and
@@ -124,7 +178,34 @@ steps:
then that pointer will be directed to the `data` member of the interrupt
record and it's up to the caller what it's doing with it.
- If `sync` is true, interrupt handler can use `datap = NULL`, and
- fop handler will have `datap` set.
+ fop handler will have `datap` point to a valid pointer.
- If `sync` is false, and handlers pass a pointer to a pointer for
`datap`, they should check if the pointed pointer is NULL before
attempting to deal with the data.
+
+### FUSE answer for the interrupted fop
+
+The kernel acknowledges a successful interruption for a given FUSE request
+if the filesystem daemon answers it with errno EINTR; upon that, the syscall
+which induced the request will be abruptly terminated with an interrupt, rather
+than returning a value.
+
+In glusterfs, this can be arranged in two ways.
+
+- If the interrupt handler wins the race for the interrupt record, ie.
+ `fuse_interrupt_finish_fop()` returns true to `fuse_<FOP>_cbk()`, then, as
+ said above, `fuse_<FOP>_cbk()` does not need to answer the FUSE request.
+ That's because then the interrupt handler will take care about answering
+ it (with errno EINTR).
+- If `fuse_interrupt_finish_fop()` returns false to `fuse_<FOP>_cbk()`, then
+ this return value does not inform the fop handler whether there was an interrupt
+ or not. This return value occurs both when fop handler won the race for the
+ interrupt record against the interrupt handler, and when there was no interrupt
+ at all.
+
+ However, the internal logic of the fop handler might detect from other
+ circumstances that an interrupt was delivered. For example, the fop handler
+ might be sleeping, waiting for some data to arrive, so that a premature
+ wakeup (with no data present) occurs if the interrupt handler intervenes. In
+ such cases it's the responsibility of the fop handler to reply the FUSE
+ request with errro EINTR.
diff --git a/doc/developer-guide/identifying-resource-leaks.md b/doc/developer-guide/identifying-resource-leaks.md
index 851fc4424bc..950cae79b0a 100644
--- a/doc/developer-guide/identifying-resource-leaks.md
+++ b/doc/developer-guide/identifying-resource-leaks.md
@@ -174,3 +174,27 @@ In this case, the resource leak can be addressed by adding a single line to the
Running the same Valgrind command and comparing the output will show that the
memory leak in `xlators/meta/src/meta.c:init` is not reported anymore.
+
+### Running DRD, the Valgrind thread error detector
+
+When configuring GlusterFS with:
+
+```shell
+./configure --enable-valgrind
+```
+
+the default Valgrind tool (Memcheck) is enabled. But it's also possble to select
+one of Memcheck or DRD by using:
+
+```shell
+./configure --enable-valgrind=memcheck
+```
+
+or:
+
+```shell
+./configure --enable-valgrind=drd
+```
+
+respectively. When using DRD, it's recommended to consult
+https://valgrind.org/docs/manual/drd-manual.html before running.
diff --git a/doc/developer-guide/thread-naming.md b/doc/developer-guide/thread-naming.md
index 74efba28e45..513140d4437 100644
--- a/doc/developer-guide/thread-naming.md
+++ b/doc/developer-guide/thread-naming.md
@@ -29,10 +29,10 @@ gf_thread_create_detached (pthread_t *thread,
As max name length for a thread in POSIX is only 16 characters including the
'\0' character, you have to be a little creative with naming. Also, it is
important that all Gluster threads have common prefix. Considering these
-conditions, we have "gluster" as prefix for all the threads created by these
+conditions, we have "glfs_" as prefix for all the threads created by these
wrapper functions. It is responsibility of the owner of thread to provide the
suffix part of the name. It does not have to be a descriptive name, as it has
-only 8 letters to work with. However, it should be unique enough such that it
+only 10 letters to work with. However, it should be unique enough such that it
can be matched with a table which describes it.
If n number of threads are spwaned to perform same function, it is must that the
@@ -87,6 +87,7 @@ such that it can be matched with a table below without ambiguity.
- posixfsy - posix fsync
- posixhc - posix heal
- posixjan - posix janitor
+- posixrsv - posix reserve
- quiesce - quiesce dequeue
- rdmaAsyn - rdma async event handler
- rdmaehan - rdma completion handler
diff --git a/doc/release-notes/8.0.md b/doc/release-notes/8.0.md
deleted file mode 100644
index fb2385681cb..00000000000
--- a/doc/release-notes/8.0.md
+++ /dev/null
@@ -1,368 +0,0 @@
-# Release notes for Gluster 8.0
-
-This is a major release that includes a range of features, code improvements and stability fixes as noted below.
-
-A selection of the key features and changes are documented in this page.
-A full list of bugs that have been addressed is included further below.
-
-- [Announcements](#announcements)
-- [Major changes and features](#major-changes-and-features)
-- [Major issues](#major-issues)
-- [Bugs addressed in the release](#bugs-addressed)
-
-## Announcements
-
-1. Releases that receive maintenance updates post release 8 are, 7 and 8
-([reference](https://www.gluster.org/release-schedule/))
-
-2. Release 8 will receive maintenance updates around the 10th of every month
-for the first 3 months post release (i.e Jun'20, Jul'20, Aug'20). Post the
-initial 3 months, it will receive maintenance updates every 2 months till EOL.
-
-
-
-## Major changes and features
-
-### Highlights
-
-- Several stability fixes addressing
- - coverity, clang-scan, address sanitizer and valgrind reported issues
- - removal of unused and hence, deprecated code and features
-- Performance Improvements
-- CentOS 8 and RHEL 8 is supported
-
-
-
-### Features
-
-- Implemented seek file operation for open-behind
-- Now storage.reserve option will take size of disk as input instead of percentage
-- Added Functionality to enable log rotation for user serviceable snapshot's logs
-- Mandatory locks enhancements in replicate subvolumes
-- To validate other memory allocation implementations instead of libc's malloc added an option to build with tcmalloc library
-- Integrated Thin-arbiter with GD1
-- Client Handling of Elastic Clusters
-
-## Major issues
-
-**None**
-
-
-## Bugs addressed
-
-Bugs addressed since release-7 are listed below.
-
-- [#789278](https://bugzilla.redhat.com/789278): Issues reported by Coverity static analysis tool
-- [#1158130](https://bugzilla.redhat.com/1158130): Not possible to disable fopen-keeo-cache when mounting
-- [#1183054](https://bugzilla.redhat.com/1183054): rpmlint throws couple of errors for RPM spec file
-- [#1193929](https://bugzilla.redhat.com/1193929): GlusterFS can be improved
-- [#1387404](https://bugzilla.redhat.com/1387404): geo-rep: gsync-sync-gfid binary installed in /usr/share/...
-- [#1410439](https://bugzilla.redhat.com/1410439): glusterfind pre output file is empty
-- [#1423442](https://bugzilla.redhat.com/1423442): group files to set volume options should have comments
-- [#1430623](https://bugzilla.redhat.com/1430623): pthread mutexes and condition variables are not destroyed
-- [#1489610](https://bugzilla.redhat.com/1489610): glusterfind saves var data under $prefix instead of localstatedir
-- [#1507896](https://bugzilla.redhat.com/1507896): glfs_init returns incorrect errno on faliure
-- [#1514683](https://bugzilla.redhat.com/1514683): Removal of bricks in volume isn't prevented if remaining brick doesn't contain all the files
-- [#1538900](https://bugzilla.redhat.com/1538900): Found a missing unref in rpc_clnt_reconnect
-- [#1554286](https://bugzilla.redhat.com/1554286): Xattr not updated if increasing the retention of a WORM/Retained file
-- [#1593542](https://bugzilla.redhat.com/1593542): ctime: Upgrade/Enabling ctime feature wrongly updates older files with latest {a|m|c}time
-- [#1620580](https://bugzilla.redhat.com/1620580): Deleted a volume and created a new volume with similar but not the same name. The kubernetes pod still keeps on running and doesn't crash. Still possible to write to gluster mount
-- [#1622665](https://bugzilla.redhat.com/1622665): clang-scan report: glusterfs issues
-- [#1626543](https://bugzilla.redhat.com/1626543): dht/tests: Create a .t to test all possible combinations for file rename
-- [#1635688](https://bugzilla.redhat.com/1635688): Keep only the valid (maintained/supported) components in the build
-- [#1636297](https://bugzilla.redhat.com/1636297): Make it easy to build / host a project which just builds glusterfs translator
-- [#1644322](https://bugzilla.redhat.com/1644322): flooding log with "glusterfs-fuse: read from /dev/fuse returned -1 (Operation not permitted)"
-- [#1651445](https://bugzilla.redhat.com/1651445): [RFE] storage.reserve option should take size of disk as input instead of percentage
-- [#1664335](https://bugzilla.redhat.com/1664335): [geo-rep]: Transport endpoint not connected with arbiter volumes
-- [#1665358](https://bugzilla.redhat.com/1665358): allow regression to not run tests with nfs, if nfs is disabled.
-- [#1668239](https://bugzilla.redhat.com/1668239): [man page] Gluster(8) - Missing disperse-data parameter Gluster Console Manager man page
-- [#1668286](https://bugzilla.redhat.com/1668286): READDIRP incorrectly updates posix-acl inode ctx
-- [#1676479](https://bugzilla.redhat.com/1676479): read-ahead and io-cache degrading performance on sequential read
-- [#1688115](https://bugzilla.redhat.com/1688115): Data heal not checking for locks on source & sink(s) before healing
-- [#1689097](https://bugzilla.redhat.com/1689097): gfapi: provide an option for changing statedump path in glfs-api.
-- [#1690454](https://bugzilla.redhat.com/1690454): mount-shared-storage.sh does not implement mount options
-- [#1693692](https://bugzilla.redhat.com/1693692): Increase code coverage from regression tests
-- [#1694920](https://bugzilla.redhat.com/1694920): Inconsistent locking in presence of disconnects
-- [#1697293](https://bugzilla.redhat.com/1697293): DHT: print hash and layout values in hexadecimal format in the logs
-- [#1698042](https://bugzilla.redhat.com/1698042): quick-read cache invalidation feature has the same key of md-cache
-- [#1707731](https://bugzilla.redhat.com/1707731): [Upgrade] Config files are not upgraded to new version
-- [#1708603](https://bugzilla.redhat.com/1708603): [geo-rep]: Note section in document is required for ignore_deletes true config option where it might delete a file
-- [#1708929](https://bugzilla.redhat.com/1708929): Add more test coverage for shd mux
-- [#1716695](https://bugzilla.redhat.com/1716695): Fix memory leaks that are present even after an xlator fini [client side xlator]
-- [#1716979](https://bugzilla.redhat.com/1716979): Multiple disconnect events being propagated for the same child
-- [#1717754](https://bugzilla.redhat.com/1717754): Enable features.locks-notify-contention by default
-- [#1717824](https://bugzilla.redhat.com/1717824): Fencing: Added the tcmu-runner ALUA feature support but after one of node is rebooted the glfs_file_lock() get stucked
-- [#1717827](https://bugzilla.redhat.com/1717827): tests/geo-rep: Add test case to validate non-root geo-replication setup
-- [#1719290](https://bugzilla.redhat.com/1719290): Glusterfs mount helper script not working with IPv6 because of regular expression or man is wrong
-- [#1720463](https://bugzilla.redhat.com/1720463): [Thin-arbiter] : Wait for connection with TA node before sending lookup/create of ta-replica id file
-- [#1720566](https://bugzilla.redhat.com/1720566): Can't rebalance GlusterFS volume because unix socket's path name is too long
-- [#1721590](https://bugzilla.redhat.com/1721590): tests/bugs/posix/bug-1040275-brick-uid-reset-on-volume-restart.t is failing
-- [#1721686](https://bugzilla.redhat.com/1721686): Remove usage of obsolete function usleep()
-- [#1722507](https://bugzilla.redhat.com/1722507): Incorrect reporting of type/gfid mismatch
-- [#1722541](https://bugzilla.redhat.com/1722541): stale shd process files leading to heal timing out and heal deamon not coming up for all volumes
-- [#1722546](https://bugzilla.redhat.com/1722546): do not assert in inode_unref if the inode table cleanup has started
-- [#1722598](https://bugzilla.redhat.com/1722598): dump the min and max latency of each xlator in statedump
-- [#1722698](https://bugzilla.redhat.com/1722698): DHT: severe memory leak in dht rename
-- [#1722740](https://bugzilla.redhat.com/1722740): [GSS] geo-replication sessions going faulty
-- [#1722802](https://bugzilla.redhat.com/1722802): Incorrect power of two calculation in mem_pool_get_fn
-- [#1722977](https://bugzilla.redhat.com/1722977): ESTALE change in fuse breaks get_real_filename implementation
-- [#1723280](https://bugzilla.redhat.com/1723280): windows cannot access mountpoint exportd from a disperse volume
-- [#1723402](https://bugzilla.redhat.com/1723402): Brick multiplexing is not working.
-- [#1723455](https://bugzilla.redhat.com/1723455): volume set <VOLNAME> group <GROUP> description missing space leading to words being merged in help output
-- [#1723658](https://bugzilla.redhat.com/1723658): [In-service] Post upgrade glusterd is crashing with a backtrace on the upgraded node while issuing gluster volume status from non-upgraded nodes
-- [#1723761](https://bugzilla.redhat.com/1723761): [Ganesha]: truncate operation not updating the ctime
-- [#1723890](https://bugzilla.redhat.com/1723890): Crash in glusterd when running test script bug-1699339.t
-- [#1724024](https://bugzilla.redhat.com/1724024): use more secure mode for mkdir operations
-- [#1724184](https://bugzilla.redhat.com/1724184): Thin-arbiter: SHD takes lock and inspects the state on TA during every index crawl
-- [#1725034](https://bugzilla.redhat.com/1725034): gluster volume help showing multiple commands for top instead of one.
-- [#1725211](https://bugzilla.redhat.com/1725211): User serviceable snapshots (USS) are not accessible after changing transport.socket.bind-address of glusterd
-- [#1726205](https://bugzilla.redhat.com/1726205): Windows client fails to copy large file to GlusterFS volume share with fruit and streams_xattr VFS modules via Samba
-- [#1726783](https://bugzilla.redhat.com/1726783): snapd crashes sometimes
-- [#1726906](https://bugzilla.redhat.com/1726906): get-state does not show correct brick status
-- [#1727068](https://bugzilla.redhat.com/1727068): Deadlock when generating statedumps
-- [#1727081](https://bugzilla.redhat.com/1727081): Disperse volume : data corruption with ftruncate data in 4+2 config
-- [#1727107](https://bugzilla.redhat.com/1727107): geo-replication/setup.py missing license details in setup()
-- [#1727248](https://bugzilla.redhat.com/1727248): [GNFS] showmout -a <IP> cause gnfs crash
-- [#1727256](https://bugzilla.redhat.com/1727256): Directory pending heal in heal info output
-- [#1727329](https://bugzilla.redhat.com/1727329): glustershd dumped core with seg fault at afr_has_quorum
-- [#1727852](https://bugzilla.redhat.com/1727852): gluster-block: improvements to volume group profile options list
-- [#1728047](https://bugzilla.redhat.com/1728047): interrupts leak memory
-- [#1728417](https://bugzilla.redhat.com/1728417): Cleanup references to Hadoop in code base
-- [#1728554](https://bugzilla.redhat.com/1728554): Spelling errors
-- [#1728683](https://bugzilla.redhat.com/1728683): [geo-rep] gluster-mountbroker missing a brief description of what the argument does in # gluster-mountbroker (add|remove|setup) --help
-- [#1728766](https://bugzilla.redhat.com/1728766): Volume start failed when shd is down in one of the node in cluster
-- [#1728770](https://bugzilla.redhat.com/1728770): Failures in remove-brick due to [Input/output error] errors
-- [#1729085](https://bugzilla.redhat.com/1729085): [EC] shd crashed while heal failed due to out of memory error.
-- [#1729107](https://bugzilla.redhat.com/1729107): Memory leak in glusterfsd process
-- [#1729463](https://bugzilla.redhat.com/1729463): gluster v geo-rep status command timing out
-- [#1729772](https://bugzilla.redhat.com/1729772): Disperse volume : Ganesha crash with IO in 4+2 config when one glusterfsd restart every 600s
-- [#1729847](https://bugzilla.redhat.com/1729847): Fix spurious failure of tests/bugs/replicate/bug-1717819-metadata-split-brain-detection.t
-- [#1730175](https://bugzilla.redhat.com/1730175): Seeing failure due to "getxattr err for dir [No data available]" in rebalance
-- [#1730409](https://bugzilla.redhat.com/1730409): core file generated - when EC volume stop and start is executed for 10 loops on a EC+Brickmux setup
-- [#1730715](https://bugzilla.redhat.com/1730715): An Input/Output error happens on a disperse volume when doing unaligned writes to a sparse file
-- [#1730953](https://bugzilla.redhat.com/1730953): mount generates errors after umount
-- [#1731920](https://bugzilla.redhat.com/1731920): [geo-rep]: gluster command not found while setting up a non-root session
-- [#1732496](https://bugzilla.redhat.com/1732496): [Coverity] RETURN_LOCAL in __nlc_inode_ctx_get()
-- [#1732717](https://bugzilla.redhat.com/1732717): fuse: Limit the number of inode invalidation requests in the queue
-- [#1733042](https://bugzilla.redhat.com/1733042): cluster.rc Create separate logdirs for each host instance
-- [#1733166](https://bugzilla.redhat.com/1733166): potential deadlock while processing callbacks in gfapi
-- [#1733425](https://bugzilla.redhat.com/1733425): Setting volume option when one of the glusterd is stopped in the cluster, post glusterd restart seeing couldn't find vol info in glusterd logs and shd, brick process offline
-- [#1733935](https://bugzilla.redhat.com/1733935): Open fd heal should filter O_APPEND/O_EXCL
-- [#1734026](https://bugzilla.redhat.com/1734026): Cannot see the "trusted.glusterfs.mdata" xattr for directory on a new brick after rebalance
-- [#1734252](https://bugzilla.redhat.com/1734252): Heal not completing after geo-rep session is stopped on EC volumes.
-- [#1734299](https://bugzilla.redhat.com/1734299): ctime: When healing ctime xattr for legacy files, if multiple clients access and modify the same file, the ctime might be updated incorrectly.
-- [#1734370](https://bugzilla.redhat.com/1734370): atime/mtime is not restored after healing for entry self heals
-- [#1734738](https://bugzilla.redhat.com/1734738): Unable to create geo-rep session on a non-root setup.
-- [#1736482](https://bugzilla.redhat.com/1736482): capture stat failure error while setting the gfid
-- [#1737288](https://bugzilla.redhat.com/1737288): nfs client gets bad ctime for copied file which is on glusterfs disperse volume with ctime on
-- [#1737291](https://bugzilla.redhat.com/1737291): features/locks: avoid use after freed of frame for blocked lock
-- [#1737484](https://bugzilla.redhat.com/1737484): geo-rep syncing significantly behind and also only one of the directories are synced with tracebacks seen
-- [#1737676](https://bugzilla.redhat.com/1737676): Upgrading a Gluster node fails when user edited glusterd.vol file exists
-- [#1737778](https://bugzilla.redhat.com/1737778): ocf resource agent for volumes don't work in non-standard environment
-- [#1738419](https://bugzilla.redhat.com/1738419): read() returns more than file size when using direct I/O
-- [#1738763](https://bugzilla.redhat.com/1738763): [EC] : fix coverity issue
-- [#1738786](https://bugzilla.redhat.com/1738786): ctime: If atime is updated via utimensat syscall ctime is not getting updated
-- [#1739360](https://bugzilla.redhat.com/1739360): [GNFS] gluster crash with nfs.nlm off
-- [#1740017](https://bugzilla.redhat.com/1740017): tests/bugs/replicate/bug-880898.t created a core file.
-- [#1741734](https://bugzilla.redhat.com/1741734): gluster-smb:glusto-test access gluster by cifs test write report Device or resource busy
-- [#1741779](https://bugzilla.redhat.com/1741779): Fix spelling errors
-- [#1741890](https://bugzilla.redhat.com/1741890): geo-rep: Changelog archive file format is incorrect
-- [#1743020](https://bugzilla.redhat.com/1743020): glusterd start is failed and throwing an error Address already in use
-- [#1743069](https://bugzilla.redhat.com/1743069): bug-1482023-snpashot-issue-with-other-processes-accessing-mounted-path.t fails in brick mux regression spuriously
-- [#1743094](https://bugzilla.redhat.com/1743094): glusterfs build fails on centos7
-- [#1743200](https://bugzilla.redhat.com/1743200): ./tests/bugs/glusterd/bug-1595320.t is failing
-- [#1743573](https://bugzilla.redhat.com/1743573): fuse client hung when issued a lookup "ls" on an ec volume
-- [#1743652](https://bugzilla.redhat.com/1743652): CentOs 6 GlusterFS client creates files with time 01/01/1970
-- [#1744519](https://bugzilla.redhat.com/1744519): log aio_error return codes in posix_fs_health_check
-- [#1744548](https://bugzilla.redhat.com/1744548): Setting cluster.heal-timeout requires volume restart
-- [#1745965](https://bugzilla.redhat.com/1745965): glusterd fails to start due to SIGABRT dumping core
-- [#1745967](https://bugzilla.redhat.com/1745967): File size was not truncated for all files when tried with rebalance in progress.
-- [#1746228](https://bugzilla.redhat.com/1746228): systemctl start glusterd is getting timed out on the scaled setup with 2000 volumes
-- [#1746320](https://bugzilla.redhat.com/1746320): SHORT-WRITE error leads to crash
-- [#1746810](https://bugzilla.redhat.com/1746810): markdown files containing 404 links
-- [#1747746](https://bugzilla.redhat.com/1747746): The result (hostname) of getnameinfo for all bricks (ipv6 addresses) are the same, while they are not.
-- [#1748448](https://bugzilla.redhat.com/1748448): syncop: Bail out if frame creation fails
-- [#1748744](https://bugzilla.redhat.com/1748744): bug-1402841.t-mt-dir-scan-race.t fails spuriously
-- [#1748836](https://bugzilla.redhat.com/1748836): Application should know when update size/version went bad
-- [#1749322](https://bugzilla.redhat.com/1749322): glustershd can not decide heald_sinks, and skip repair, so some entries lingering in volume heal info
-- [#1750387](https://bugzilla.redhat.com/1750387): Deprecated log rotate command still present in "# gluster v help"
-- [#1750618](https://bugzilla.redhat.com/1750618): Cleanup of executable in tests/bugs/gfapi/bug-1447266/bug-1447266.t not done
-- [#1751134](https://bugzilla.redhat.com/1751134): Spurious failure tests/bugs/replicate/bug-1734370-entry-heal-restore-time.t
-- [#1751907](https://bugzilla.redhat.com/1751907): bricks gone down unexpectedly
-- [#1752330](https://bugzilla.redhat.com/1752330): seeing error message in glustershd.log on volume start(or may be as part of shd graph regeneration) inet_pton failed with return code 0 [Invalid argument]
-- [#1752331](https://bugzilla.redhat.com/1752331): Test tests/basic/volume-scale-shd-mux.t is failing on upstream CI
-- [#1753569](https://bugzilla.redhat.com/1753569): git clone fails on gluster volumes exported via nfs-ganesha
-- [#1753592](https://bugzilla.redhat.com/1753592): Segmentation fault occurs while truncate file
-- [#1753843](https://bugzilla.redhat.com/1753843): [Disperse volume]: Regression in IO performance seen in sequential read for large file
-- [#1753857](https://bugzilla.redhat.com/1753857): geo-rep: performance improvement while syncing heavy renames with existing destination
-- [#1753859](https://bugzilla.redhat.com/1753859): Typos in glusterd log messages
-- [#1753880](https://bugzilla.redhat.com/1753880): Set the default lru-limit in fuse to a smaller number
-- [#1753928](https://bugzilla.redhat.com/1753928): geo-rep: non-root session going fault due improper sub-command
-- [#1754448](https://bugzilla.redhat.com/1754448): Re-alignment of Structure attributes
-- [#1754477](https://bugzilla.redhat.com/1754477): Thin-arbiter: Raise error in CLI if replica-count is not 2
-- [#1755344](https://bugzilla.redhat.com/1755344): glustershd.log getting flooded with "W [inode.c:1017:inode_find] (-->/usr/lib64/glusterfs/6.0/xlator/cluster/disperse.so(+0xe3f9) [0x7fd09b0543f9] -->/usr/lib64/glusterfs/6.0/xlator/cluster/disperse.so(+0xe19c) [0x7fd09b05419 TABLE NOT FOUND"
-- [#1755900](https://bugzilla.redhat.com/1755900): heketidbstorage bricks go down during PVC creation
-- [#1756211](https://bugzilla.redhat.com/1756211): tests/bugs/shard/bug-1272986.t fails
-- [#1756900](https://bugzilla.redhat.com/1756900): tests are failing in RHEL8 regression
-- [#1756938](https://bugzilla.redhat.com/1756938): afr: support split-brain CLI for replica 3
-- [#1757399](https://bugzilla.redhat.com/1757399): Rebalance is causing glusterfs crash on client node
-- [#1758579](https://bugzilla.redhat.com/1758579): Rebalance causing IO Error - File descriptor in bad state
-- [#1758878](https://bugzilla.redhat.com/1758878): # gluster v info --xml is always returning <distCount>3</distCount> for all Nx3 volumes
-- [#1758984](https://bugzilla.redhat.com/1758984): Enable direct-io options in group virt
-- [#1759002](https://bugzilla.redhat.com/1759002): Spurious failure tests/bugs/replicate/bug-1744548-heal-timeout.t
-- [#1759081](https://bugzilla.redhat.com/1759081): Spurious failure of /tests/bugs/replicate/bug-1134691-afr-lookup-metadata-heal.t
-- [#1760187](https://bugzilla.redhat.com/1760187): Implement seek fop
-- [#1760189](https://bugzilla.redhat.com/1760189): Use replica aware seek fop
-- [#1760467](https://bugzilla.redhat.com/1760467): rebalance start is succeeding when quorum is not met
-- [#1761759](https://bugzilla.redhat.com/1761759): Failure in ./tests/basic/posix/shared-statfs.t
-- [#1761769](https://bugzilla.redhat.com/1761769): On some distros bug-1272986.t takes more than 2 minutes to run
-- [#1762220](https://bugzilla.redhat.com/1762220): [geo-rep] sync_method showing rsync instead of tarssh post in-service upgrade
-- [#1762438](https://bugzilla.redhat.com/1762438): DHT- gluster rebalance status shows wrong data size after rebalance is completed successfully
-- [#1763036](https://bugzilla.redhat.com/1763036): glusterfsd crashed with "'MemoryError' Cannot access memory at address"
-- [#1763439](https://bugzilla.redhat.com/1763439): [GSS] geo-rep entering into faulty state with OSError: [Errno 13] Permission denied
-- [#1764110](https://bugzilla.redhat.com/1764110): tests/bugs/shard/unlinks-and-renames.t fails on RHEL8
-- [#1764119](https://bugzilla.redhat.com/1764119): gluster rebalance status doesn't show detailed information when a node is rebooted
-- [#1764129](https://bugzilla.redhat.com/1764129): quota_fsck script KeyError: 'contri_size'
-- [#1764208](https://bugzilla.redhat.com/1764208): cgroup control-cpu-load.sh script not working
-- [#1764418](https://bugzilla.redhat.com/1764418): Add Mohit & Sanju as glusterd/cli maintainers
-- [#1765017](https://bugzilla.redhat.com/1765017): gf_event doesn't work for glfsheal process
-- [#1765155](https://bugzilla.redhat.com/1765155): replication shouldn't modify xattr-req coming from parent
-- [#1765186](https://bugzilla.redhat.com/1765186): Problematic coding practices at logger
-- [#1765421](https://bugzilla.redhat.com/1765421): DHT: Add comments to the code
-- [#1765426](https://bugzilla.redhat.com/1765426): test: fix non-root geo-rep test case
-- [#1765542](https://bugzilla.redhat.com/1765542): Add Sunny Kumar as co-maintainer of Geo-replication component
-- [#1768407](https://bugzilla.redhat.com/1768407): glusterfsd memory leak observed after enable tls
-- [#1768896](https://bugzilla.redhat.com/1768896): Long method in glusterfsd - set_fuse_mount_options(...)
-- [#1769712](https://bugzilla.redhat.com/1769712): check if grapj is ready beforce process cli command
-- [#1769754](https://bugzilla.redhat.com/1769754): dht_readdirp_cbk: Do not strip out entries with invalid stats
-- [#1771365](https://bugzilla.redhat.com/1771365): libglusterfs/dict.c : memory leaks
-- [#1771577](https://bugzilla.redhat.com/1771577): [RHEL 6] Geo-replication session not starting after creation
-- [#1771895](https://bugzilla.redhat.com/1771895): geo-rep: Improve debugging in log_raise_exception
-- [#1772006](https://bugzilla.redhat.com/1772006): NULL dict messages flooding fuse mount log
-- [#1773530](https://bugzilla.redhat.com/1773530): ctime value is different from atime/mtime on a create of file
-- [#1773856](https://bugzilla.redhat.com/1773856): Set volume option when one of the node is powered off, After powering the node brick processes are offline
-- [#1774011](https://bugzilla.redhat.com/1774011): Heal Info is hung when I/O is in progress on a gluster block volume
-- [#1774866](https://bugzilla.redhat.com/1774866): man page update needed for gluster volume top command
-- [#1775612](https://bugzilla.redhat.com/1775612): Remove guest access by default for GlusterFS volume SMB shares added by hook scripts
-- [#1776264](https://bugzilla.redhat.com/1776264): RFE: systemd should restart glusterd on crash
-- [#1776757](https://bugzilla.redhat.com/1776757): DHT - Reduce methods scope
-- [#1776784](https://bugzilla.redhat.com/1776784): glfsheal crash on unexpected volume name
-- [#1776801](https://bugzilla.redhat.com/1776801): Bricks are not available when volume create fails
-- [#1776892](https://bugzilla.redhat.com/1776892): [patch] .dirstamp should be in ignored
-- [#1778457](https://bugzilla.redhat.com/1778457): Missing error logs(afr/self-heald )
-- [#1779055](https://bugzilla.redhat.com/1779055): glusterfs process memory leak in ior test
-- [#1779089](https://bugzilla.redhat.com/1779089): glusterfsd do not release posix lock when multiple glusterfs client do flock -xo to the same file paralleled
-- [#1779742](https://bugzilla.redhat.com/1779742): tests/00-geo-rep/00-georep-verify-non-root-setup.t fail on freshly installed builder
-- [#1779760](https://bugzilla.redhat.com/1779760): Improve logging in EC, client and lock xlator
-- [#1780190](https://bugzilla.redhat.com/1780190): glfsheal should be installed and invoked as architecture-dependent binary helper
-- [#1780260](https://bugzilla.redhat.com/1780260): v7 fails to build on Debian 9 [patch?]
-- [#1781440](https://bugzilla.redhat.com/1781440): event_slot_alloc not able to return index after reach slot_used count to 1024
-- [#1782200](https://bugzilla.redhat.com/1782200): glusterd restart failing to start.
-- [#1782495](https://bugzilla.redhat.com/1782495): GlusterFS brick process crash
-- [#1784375](https://bugzilla.redhat.com/1784375): 'gluster volume set <VOLUME> disable.nfs' accidentally killed unexpected process, and forced a data brick offline.
-- [#1785143](https://bugzilla.redhat.com/1785143): Multiple glusterfsd process spawn when glusterd restart during a volume start.
-- [#1785208](https://bugzilla.redhat.com/1785208): glusterfs client process coredump
-- [#1785611](https://bugzilla.redhat.com/1785611): glusterfsd cashes after a few seconds
-- [#1785998](https://bugzilla.redhat.com/1785998): change the error message for heal statistics to reflect its supportability for disperse volume
-- [#1786276](https://bugzilla.redhat.com/1786276): [geo-rep] Help for positional argument SLAVE in schedule_georep.py.in isn't clear.
-- [#1786459](https://bugzilla.redhat.com/1786459): unable to enable brick-multiplex feature
-- [#1786478](https://bugzilla.redhat.com/1786478): default option is disappeared in volume info after volume reset
-- [#1786679](https://bugzilla.redhat.com/1786679): Duplicate entries in 'ls' output after a volume expansion
-- [#1786722](https://bugzilla.redhat.com/1786722): Functionality to enable log rotation for user serviceable snapshot's logs.
-- [#1787122](https://bugzilla.redhat.com/1787122): glusterd allowing to set server.statedump-path to file, non-existent file and non-existent paths
-- [#1787274](https://bugzilla.redhat.com/1787274): heal not actually healing metadata of a regular file when only time stamps are changed(data heal not required)
-- [#1787554](https://bugzilla.redhat.com/1787554): Unable to set/modify optimistic-change-log for replicate volumes
-- [#1789439](https://bugzilla.redhat.com/1789439): Glusterfind pre command fails
-- [#1789478](https://bugzilla.redhat.com/1789478): S57glusterfind-delete-post.py not python3 ready (does not decode bytestring)
-- [#1790748](https://bugzilla.redhat.com/1790748): Remove extra argument
-- [#1790870](https://bugzilla.redhat.com/1790870): Memory corruption when sending events to an IPv6 host
-- [#1791682](https://bugzilla.redhat.com/1791682): fail to build on recent Fedora
-- [#1792276](https://bugzilla.redhat.com/1792276): config ssh-port can accept negative and outside allowed port range value
-- [#1792707](https://bugzilla.redhat.com/1792707): xlators/features/quota/src/quota.c:quota_log_usage
-- [#1793378](https://bugzilla.redhat.com/1793378): dht_hash_compute() crashes when it receives a zero length name
-- [#1793852](https://bugzilla.redhat.com/1793852): Mounts fails after reboot of 1/3 gluster nodes
-- [#1793995](https://bugzilla.redhat.com/1793995): gluster crash when built without gNFS support
-- [#1797869](https://bugzilla.redhat.com/1797869): bitrot: Number of signing process threads should be configurable.
-- [#1797882](https://bugzilla.redhat.com/1797882): Segmentation fault occurs while truncate file
-- [#1797934](https://bugzilla.redhat.com/1797934): Client should propagate ping event from brick
-- [#1800583](https://bugzilla.redhat.com/1800583): Halo replication is not working
-- [#1800956](https://bugzilla.redhat.com/1800956): Rebalance : Status lists failures on stopping rebalance while it is in progress
-- [#1801623](https://bugzilla.redhat.com/1801623): spurious self-heald.t failure
-- [#1801624](https://bugzilla.redhat.com/1801624): Heal pending on volume, even after all the bricks are up
-- [#1801684](https://bugzilla.redhat.com/1801684): Memory corruption when glfs_init() is called after glfs_fini()
-- [#1804786](https://bugzilla.redhat.com/1804786): mount.glusterfs strips off "/" from subdir-mounts
-- [#1808421](https://bugzilla.redhat.com/1808421): WORM: If autocommit-period 0 file will be WORMed with 0 Byte during initial write
-- [#1808875](https://bugzilla.redhat.com/1808875): [brickmux]: glustershd crashed when rebooting 1/3 nodes at regular intervals
-- [#1810042](https://bugzilla.redhat.com/1810042): Changes to gluster peer probe in nightly build breaks ansible:gluster_volume call
-- [#1810842](https://bugzilla.redhat.com/1810842): frequent heal observed when file opened during one brick is down
-- [#1810934](https://bugzilla.redhat.com/1810934): Segfault in FUSE process, potential use after free
-- [#1811631](https://bugzilla.redhat.com/1811631): brick crashed when creating and deleting volumes over time (with brick mux enabled only)
-- [#1812144](https://bugzilla.redhat.com/1812144): Add a warning message during volume expansion or resize on volume with snapshots
-- [#1812353](https://bugzilla.redhat.com/1812353): create-export-ganesha script: mention labelled nfs parameter
-- [#154](https://github.com/gluster/glusterfs/issues/154) Optimized CHANGELOG
-- [#237](https://github.com/gluster/glusterfs/issues/237) Validate other memory allocation implementations instead of l...
-- [#475](https://github.com/gluster/glusterfs/issues/475) Reduce the number or threads used in the brick process
-- [#613](https://github.com/gluster/glusterfs/issues/613) Mandatory locks enhancements in replicate subvolumes
-- [#657](https://github.com/gluster/glusterfs/issues/657) Structured logging format support
-- [#663](https://github.com/gluster/glusterfs/issues/663) Add Ganesha HA bits back to glusterfs code repo
-- [#687](https://github.com/gluster/glusterfs/issues/687) Thin-arbiter integration with GD1
-- [#699](https://github.com/gluster/glusterfs/issues/699) executable program will crash if linked libgfapi together wit...
-- [#703](https://github.com/gluster/glusterfs/issues/703) provide mechanism to test individual xlators
-- [#721](https://github.com/gluster/glusterfs/issues/721) Introduce quorum-count option in disperse volumes as well
-- [#723](https://github.com/gluster/glusterfs/issues/723) Provide scripts to reset xattrs of the entries which could be...
-- [#725](https://github.com/gluster/glusterfs/issues/725) Disperse: A way to read from specific bricks
-- [#741](https://github.com/gluster/glusterfs/issues/741) Client Handling of Elastic Clusters
-- [#745](https://github.com/gluster/glusterfs/issues/745) storage.reserve enhancement for posix_write
-- [#748](https://github.com/gluster/glusterfs/issues/748) Improve MAKE_HANDLE_GFID_PATH macro and posix_handle_gfid_path()
-- [#753](https://github.com/gluster/glusterfs/issues/753) Remove fetching items in gf_cli_replace_brick(), gf_cli_reset...
-- [#755](https://github.com/gluster/glusterfs/issues/755) [RFE] Geo-replication code improvements
-- [#761](https://github.com/gluster/glusterfs/issues/761) Improve MAKE_HANDLE_PATH macro
-- [#763](https://github.com/gluster/glusterfs/issues/763) thin-arbiter: Testing report
-- [#765](https://github.com/gluster/glusterfs/issues/765) nfs.rpc-auth-allow gluster7 + gnfs
-- [#788](https://github.com/gluster/glusterfs/issues/788) run-with-valgrind option causes gnfs and quota to fail to start
-- [#824](https://github.com/gluster/glusterfs/issues/824) Migrate bugzilla workflow to github issues workflow
-- [#832](https://github.com/gluster/glusterfs/issues/832) Permission Denied in logs
-- [#884](https://github.com/gluster/glusterfs/issues/884) [bug:1808688] Data corruption with asynchronous writes (pleas...
-- [#891](https://github.com/gluster/glusterfs/issues/891) [bug:1802451] Optimize posix code to improve file creation
-- [#977](https://github.com/gluster/glusterfs/issues/977) [bug:1811631] brick crashed when creating and deleting volume...
-- [#999](https://github.com/gluster/glusterfs/issues/999) [bug:1791285] Changing permissions on root directory(director...
-- [#1000](https://github.com/gluster/glusterfs/issues/1000) [bug:1193929] GlusterFS can be improved
-- [#1038](https://github.com/gluster/glusterfs/issues/1038) [bug:1787138] Crash on rpcsvc_drc_client_unref() - fails on G...
-- [#1042](https://github.com/gluster/glusterfs/issues/1042) [bug:1806499] afr-lock-heal-basic.t and /afr-lock-heal-advanc...
-- [#1044](https://github.com/gluster/glusterfs/issues/1044) [bug:1790730] Add a basic test file to glusterfind
-- [#1052](https://github.com/gluster/glusterfs/issues/1052) [bug:1693692] Increase code coverage from regression tests
-- [#1060](https://github.com/gluster/glusterfs/issues/1060) [bug:789278] Issues reported by Coverity static analysis tool
-- [#1067](https://github.com/gluster/glusterfs/issues/1067) [bug:1661889] Metadata heal picks different brick each time a...
-- [#1097](https://github.com/gluster/glusterfs/issues/1097) [bug:1635688] Keep only the valid (maintained/supported) comp...
-- [#1102](https://github.com/gluster/glusterfs/issues/1102) dht: gf_defrag_process_dir is called even if gf_defrag_fix_la...
-- [#1104](https://github.com/gluster/glusterfs/issues/1104) geo-replication: descriptive message when worker crashes due ...
-- [#1105](https://github.com/gluster/glusterfs/issues/1105) [bug:1794263] Multiple imports from the same library in the ....
-- [#1127](https://github.com/gluster/glusterfs/issues/1127) Mount crash during background shard cleanup
-- [#1134](https://github.com/gluster/glusterfs/issues/1134) snap_scheduler.py init failing with "TypeError: Can't mix str...
-- [#1140](https://github.com/gluster/glusterfs/issues/1140) getfattr returns ENOATTR for system.posix_acl_access on dispe...
-- [#1141](https://github.com/gluster/glusterfs/issues/1141) Make SSL connection messages useful
-- [#1142](https://github.com/gluster/glusterfs/issues/1142) log the ENOENT error in posix_pstat
-- [#1144](https://github.com/gluster/glusterfs/issues/1144) [Disperse] Add test for reset-brick for disperse volume
-- [#1146](https://github.com/gluster/glusterfs/issues/1146) gfapi/Upcall: Potential deadlock in synctask threads processi...
-- [#1149](https://github.com/gluster/glusterfs/issues/1149) Add error logs to debug failures in ./tests/bugs/protocol/bug...
-- [#1150](https://github.com/gluster/glusterfs/issues/1150) Avoid dict_del logs in posix_is_layout_stale while key is NULL
-- [#1152](https://github.com/gluster/glusterfs/issues/1152) Spurious failure of tests/bugs/protocol/bug-1433815-auth-allow.t
-- [#1153](https://github.com/gluster/glusterfs/issues/1153) Spurious failure of ./tests/bugs/snapshot/bug-1111041.t
-- [#1154](https://github.com/gluster/glusterfs/issues/1154) failing test cases
-- [#1156](https://github.com/gluster/glusterfs/issues/1156) Spurious failure of tests/features/worm.t
-- [#1158](https://github.com/gluster/glusterfs/issues/1158) spurious failure of tests/bugs/glusterd/serialize-shd-manager...
-- [#1160](https://github.com/gluster/glusterfs/issues/1160) sys_stat should be used instead of stat
-- [#1161](https://github.com/gluster/glusterfs/issues/1161) tests: file offsets and sizes shouldn't be truncated to 32-bi...
-- [#1162](https://github.com/gluster/glusterfs/issues/1162) spurious failure of test case tests/bugs/glusterd/removing-mu...
-- [#1169](https://github.com/gluster/glusterfs/issues/1169) common-ha: cluster status shows "FAILOVER" even when all node...
-- [#1180](https://github.com/gluster/glusterfs/issues/1180) (glusterfs-8.0) - GlusterFS 8.0 tracker
-- [#1179](https://github.com/gluster/glusterfs/issues/1179) gnfs split brain when 1 server in 3x1 down (high load)
-- [#1220](https://github.com/gluster/glusterfs/issues/1220) cluster/ec: return correct error code and log the message in case of BADFD
-- [#1223](https://github.com/gluster/glusterfs/issues/1223) Failure of tests/basic/gfapi/gfapi-copy-file-range.t
-- [#1116](https://github.com/gluster/glusterfs/issues/1116) [bug:1790736] gluster volume list returning wrong volume list / volume list time out
-- [#990](https://github.com/gluster/glusterfs/issues/990) [bug:1578405] EIO errors when updating and deleting entries co...
-- [#1126](https://github.com/gluster/glusterfs/issues/1126) packaging: overhaul glusterfs.spec(.in) to align with SUSE and...
-- [#1225](https://github.com/gluster/glusterfs/issues/1225) fuse causes glusterd to dump core
-- [#1243](https://github.com/gluster/glusterfs/issues/1243) Modify and return iatt (especially size and block-count) in sh...
-- [#1254](https://github.com/gluster/glusterfs/issues/1254) Prioritize ENOSPC over other lesser priority errors
-- [#1303](https://github.com/gluster/glusterfs/issues/1303) Failures in rebalance due to [Input/output error]
diff --git a/events/src/eventsapiconf.py.in b/events/src/eventsapiconf.py.in
index 76b5954d325..700093bee60 100644
--- a/events/src/eventsapiconf.py.in
+++ b/events/src/eventsapiconf.py.in
@@ -28,6 +28,8 @@ def get_glusterd_workdir():
return glusterd_workdir
SERVER_ADDRESS = "0.0.0.0"
+SERVER_ADDRESSv4 = "0.0.0.0"
+SERVER_ADDRESSv6 = "::1"
DEFAULT_CONFIG_FILE = "@SYSCONF_DIR@/glusterfs/eventsconfig.json"
CUSTOM_CONFIG_FILE_TO_SYNC = "/events/config.json"
CUSTOM_CONFIG_FILE = get_glusterd_workdir() + CUSTOM_CONFIG_FILE_TO_SYNC
diff --git a/events/src/glustereventsd.py b/events/src/glustereventsd.py
index c4c7b65e332..341a3b60947 100644
--- a/events/src/glustereventsd.py
+++ b/events/src/glustereventsd.py
@@ -13,6 +13,7 @@
from __future__ import print_function
import sys
import signal
+import threading
try:
import socketserver
except ImportError:
@@ -23,10 +24,17 @@ from argparse import ArgumentParser, RawDescriptionHelpFormatter
from eventtypes import all_events
import handlers
import utils
-from eventsapiconf import SERVER_ADDRESS, PID_FILE
+from eventsapiconf import SERVER_ADDRESSv4, SERVER_ADDRESSv6, PID_FILE
from eventsapiconf import AUTO_BOOL_ATTRIBUTES, AUTO_INT_ATTRIBUTES
from utils import logger, PidFile, PidFileLockFailed, boolify
+# Subclass so that specifically IPv4 packets are captured
+class UDPServerv4(socketserver.ThreadingUDPServer):
+ address_family = socket.AF_INET
+
+# Subclass so that specifically IPv6 packets are captured
+class UDPServerv6(socketserver.ThreadingUDPServer):
+ address_family = socket.AF_INET6
class GlusterEventsRequestHandler(socketserver.BaseRequestHandler):
@@ -89,6 +97,10 @@ def signal_handler_sigusr2(sig, frame):
utils.restart_webhook_pool()
+def UDP_server_thread(sock):
+ sock.serve_forever()
+
+
def init_event_server():
utils.setup_logger()
utils.load_all()
@@ -99,15 +111,26 @@ def init_event_server():
sys.stderr.write("Unable to get Port details from Config\n")
sys.exit(1)
- # Start the Eventing Server, UDP Server
+ # Creating the Eventing Server, UDP Server for IPv4 packets
+ try:
+ serverv4 = UDPServerv4((SERVER_ADDRESSv4, port),
+ GlusterEventsRequestHandler)
+ except socket.error as e:
+ sys.stderr.write("Failed to start Eventsd for IPv4: {0}\n".format(e))
+ sys.exit(1)
+ # Creating the Eventing Server, UDP Server for IPv6 packets
try:
- server = socketserver.ThreadingUDPServer(
- (SERVER_ADDRESS, port),
- GlusterEventsRequestHandler)
+ serverv6 = UDPServerv6((SERVER_ADDRESSv6, port),
+ GlusterEventsRequestHandler)
except socket.error as e:
- sys.stderr.write("Failed to start Eventsd: {0}\n".format(e))
+ sys.stderr.write("Failed to start Eventsd for IPv6: {0}\n".format(e))
sys.exit(1)
- server.serve_forever()
+ server_thread1 = threading.Thread(target=UDP_server_thread,
+ args=(serverv4,))
+ server_thread2 = threading.Thread(target=UDP_server_thread,
+ args=(serverv6,))
+ server_thread1.start()
+ server_thread2.start()
def get_args():
diff --git a/events/src/peer_eventsapi.py b/events/src/peer_eventsapi.py
index 26b77a09179..4d2e5f35b1c 100644
--- a/events/src/peer_eventsapi.py
+++ b/events/src/peer_eventsapi.py
@@ -353,8 +353,7 @@ class WebhookModCmd(Cmd):
errcode=ERROR_WEBHOOK_NOT_EXISTS,
json_output=args.json)
- if isinstance(data[args.url], str) or \
- isinstance(data[args.url], unicode):
+ if isinstance(data[args.url], str):
data[args.url]["token"] = data[args.url]
if args.bearer_token != "":
diff --git a/events/src/utils.py b/events/src/utils.py
index 38b707a1b28..6d4e0791a2b 100644
--- a/events/src/utils.py
+++ b/events/src/utils.py
@@ -13,6 +13,7 @@ import sys
import json
import os
import logging
+import logging.handlers
import fcntl
from errno import EBADF
from threading import Thread
@@ -98,7 +99,7 @@ def setup_logger():
logger.setLevel(logging.INFO)
# create the logging file handler
- fh = logging.FileHandler(LOG_FILE)
+ fh = logging.handlers.WatchedFileHandler(LOG_FILE)
formatter = logging.Formatter("[%(asctime)s] %(levelname)s "
"[%(module)s - %(lineno)s:%(funcName)s] "
diff --git a/extras/distributed-testing/distributed-test-runner.py b/extras/distributed-testing/distributed-test-runner.py
index 7bfb6c9652a..5a07e2feab1 100755
--- a/extras/distributed-testing/distributed-test-runner.py
+++ b/extras/distributed-testing/distributed-test-runner.py
@@ -383,14 +383,17 @@ class Handlers:
return self.shell.call("make install") == 0
@synchronized
- def prove(self, id, test, timeout, valgrind=False, asan_noleaks=True):
+ def prove(self, id, test, timeout, valgrind="no", asan_noleaks=True):
assert id == self.client_id
self.shell.cd(self.gluster_root)
env = "DEBUG=1 "
- if valgrind:
+ if valgrind == "memcheck" or valgrind == "yes":
cmd = "valgrind"
cmd += " --tool=memcheck --leak-check=full --track-origins=yes"
cmd += " --show-leak-kinds=all -v prove -v"
+ elif valgrind == "drd":
+ cmd = "valgrind"
+ cmd += " --tool=drd -v prove -v"
elif asan_noleaks:
cmd = "prove -v"
env += "ASAN_OPTIONS=detect_leaks=0 "
@@ -827,8 +830,9 @@ parser.add_argument("--port", help="server port to listen",
type=int, default=DEFAULT_PORT)
# test role
parser.add_argument("--tester", help="start tester", action="store_true")
-parser.add_argument("--valgrind", help="run tests under valgrind",
- action="store_true")
+parser.add_argument("--valgrind[=memcheck,drd]",
+ help="run tests with valgrind tool 'memcheck' or 'drd'",
+ default="no")
parser.add_argument("--asan", help="test with asan enabled",
action="store_true")
parser.add_argument("--asan-noleaks", help="test with asan but no mem leaks",
diff --git a/extras/ganesha/ocf/ganesha_nfsd b/extras/ganesha/ocf/ganesha_nfsd
index 93fc8be5136..f91e8b6b8f7 100644
--- a/extras/ganesha/ocf/ganesha_nfsd
+++ b/extras/ganesha/ocf/ganesha_nfsd
@@ -36,7 +36,7 @@ else
. ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs
fi
-OCF_RESKEY_ha_vol_mnt_default="/var/run/gluster/shared_storage"
+OCF_RESKEY_ha_vol_mnt_default="/run/gluster/shared_storage"
: ${OCF_RESKEY_ha_vol_mnt=${OCF_RESKEY_ha_vol_mnt_default}}
ganesha_meta_data() {
diff --git a/extras/ganesha/scripts/ganesha-ha.sh b/extras/ganesha/scripts/ganesha-ha.sh
index c562bea4779..9790a719e10 100644
--- a/extras/ganesha/scripts/ganesha-ha.sh
+++ b/extras/ganesha/scripts/ganesha-ha.sh
@@ -24,7 +24,7 @@ GANESHA_HA_SH=$(realpath $0)
HA_NUM_SERVERS=0
HA_SERVERS=""
HA_VOL_NAME="gluster_shared_storage"
-HA_VOL_MNT="/var/run/gluster/shared_storage"
+HA_VOL_MNT="/run/gluster/shared_storage"
HA_CONFDIR=$HA_VOL_MNT"/nfs-ganesha"
SERVICE_MAN="DISTRO_NOT_FOUND"
@@ -919,8 +919,9 @@ status()
local index=1
local nodes
- # change tabs to spaces, strip leading spaces
- pcs status | sed -e "s/\t/ /g" -e "s/^[ ]*//" > ${scratch}
+ # change tabs to spaces, strip leading spaces, including any
+ # new '*' at the beginning of a line introduced in pcs-0.10.x
+ pcs status | sed -e "s/\t/ /g" -e "s/^[ ]*\*//" -e "s/^[ ]*//" > ${scratch}
nodes[0]=${1}; shift
diff --git a/extras/glusterfs-georep-upgrade.py b/extras/glusterfs-georep-upgrade.py
new file mode 100755
index 00000000000..634576058d6
--- /dev/null
+++ b/extras/glusterfs-georep-upgrade.py
@@ -0,0 +1,77 @@
+#!/usr/bin/python3
+"""
+
+Copyright (c) 2020 Red Hat, Inc. <http://www.redhat.com>
+This file is part of GlusterFS.
+
+This file is licensed to you under your choice of the GNU Lesser
+General Public License, version 3 or any later version (LGPLv3 or
+later), or the GNU General Public License, version 2 (GPLv2), in all
+cases as published by the Free Software Foundation.
+
+"""
+
+import argparse
+import errno
+import os, sys
+import shutil
+from datetime import datetime
+
+def find_htime_path(brick_path):
+ dirs = []
+ htime_dir = os.path.join(brick_path, '.glusterfs/changelogs/htime')
+ for file in os.listdir(htime_dir):
+ if os.path.isfile(os.path.join(htime_dir,file)) and file.startswith("HTIME"):
+ dirs.append(os.path.join(htime_dir, file))
+ else:
+ raise FileNotFoundError("%s unavailable" % (os.path.join(htime_dir, file)))
+ return dirs
+
+def modify_htime_file(brick_path):
+ htime_file_path_list = find_htime_path(brick_path)
+
+ for htime_file_path in htime_file_path_list:
+ changelog_path = os.path.join(brick_path, '.glusterfs/changelogs')
+ temp_htime_path = os.path.join(changelog_path, 'htime/temp_htime_file')
+ with open(htime_file_path, 'r') as htime_file, open(temp_htime_path, 'w') as temp_htime_file:
+ #extract epoch times from htime file
+ paths = htime_file.read().split("\x00")
+
+ for pth in paths:
+ epoch_no = pth.split(".")[-1]
+ changelog = os.path.basename(pth)
+ #convert epoch time to year, month and day
+ if epoch_no != '':
+ date=(datetime.fromtimestamp(float(int(epoch_no))).strftime("%Y/%m/%d"))
+ #update paths in temp htime file
+ temp_htime_file.write("%s/%s/%s\x00" % (changelog_path, date, changelog))
+ #create directory in the format year/month/days
+ path = os.path.join(changelog_path, date)
+
+ if changelog.startswith("CHANGELOG."):
+ try:
+ os.makedirs(path, mode = 0o600);
+ except OSError as exc:
+ if exc.errno == errno.EEXIST:
+ pass
+ else:
+ raise
+
+ #copy existing changelogs to new directory structure, delete old changelog files
+ shutil.copyfile(pth, os.path.join(path, changelog))
+ os.remove(pth)
+
+ #rename temp_htime_file with htime file
+ os.rename(htime_file_path, os.path.join('%s.bak'%htime_file_path))
+ os.rename(temp_htime_path, htime_file_path)
+
+if __name__ == "__main__":
+ parser = argparse.ArgumentParser()
+ parser.add_argument('brick_path', help="This upgrade script, which is to be run on\
+ server side, takes brick path as the argument, \
+ updates paths inside htime file and alters the directory structure \
+ above the changelog files inorder to support new optimised format \
+ of the directory structure as per \
+ https://review.gluster.org/#/c/glusterfs/+/23733/")
+ args = parser.parse_args()
+ modify_htime_file(args.brick_path)
diff --git a/extras/hook-scripts/set/post/S32gluster_enable_shared_storage.sh b/extras/hook-scripts/set/post/S32gluster_enable_shared_storage.sh
index 885ed03ad5b..1f2564b44ff 100755
--- a/extras/hook-scripts/set/post/S32gluster_enable_shared_storage.sh
+++ b/extras/hook-scripts/set/post/S32gluster_enable_shared_storage.sh
@@ -79,9 +79,9 @@ done
if [ "$option" == "disable" ]; then
# Unmount the volume on all the nodes
- umount /var/run/gluster/shared_storage
- cat /etc/fstab | grep -v "gluster_shared_storage /var/run/gluster/shared_storage/" > /var/run/gluster/fstab.tmp
- mv /var/run/gluster/fstab.tmp /etc/fstab
+ umount /run/gluster/shared_storage
+ cat /etc/fstab | grep -v "gluster_shared_storage /run/gluster/shared_storage/" > /run/gluster/fstab.tmp
+ mv /run/gluster/fstab.tmp /etc/fstab
fi
if [ "$is_originator" == 1 ]; then
@@ -104,8 +104,15 @@ function check_volume_status()
echo $status
}
-mount_cmd="mount -t glusterfs $local_node_hostname:/gluster_shared_storage \
- /var/run/gluster/shared_storage"
+key=`echo $5 | cut -d '=' -f 1`
+val=`echo $5 | cut -d '=' -f 2`
+if [ "$key" == "transport.address-family" ]; then
+ mount_cmd="mount -t glusterfs -o xlator-option=transport.address-family=inet6 \
+ $local_node_hostname:/gluster_shared_storage /run/gluster/shared_storage"
+else
+ mount_cmd="mount -t glusterfs $local_node_hostname:/gluster_shared_storage \
+ /run/gluster/shared_storage"
+fi
if [ "$option" == "enable" ]; then
retry=0;
@@ -120,10 +127,10 @@ if [ "$option" == "enable" ]; then
status=$(check_volume_status)
done
# Mount the volume on all the nodes
- umount /var/run/gluster/shared_storage
- mkdir -p /var/run/gluster/shared_storage
+ umount /run/gluster/shared_storage
+ mkdir -p /run/gluster/shared_storage
$mount_cmd
- cp /etc/fstab /var/run/gluster/fstab.tmp
- echo "$local_node_hostname:/gluster_shared_storage /var/run/gluster/shared_storage/ glusterfs defaults 0 0" >> /var/run/gluster/fstab.tmp
- mv /var/run/gluster/fstab.tmp /etc/fstab
+ cp /etc/fstab /run/gluster/fstab.tmp
+ echo "$local_node_hostname:/gluster_shared_storage /run/gluster/shared_storage/ glusterfs defaults 0 0" >> /run/gluster/fstab.tmp
+ mv /run/gluster/fstab.tmp /etc/fstab
fi
diff --git a/extras/hook-scripts/start/post/S31ganesha-start.sh b/extras/hook-scripts/start/post/S31ganesha-start.sh
index 90ba6bc73a5..7ad6f23ad06 100755
--- a/extras/hook-scripts/start/post/S31ganesha-start.sh
+++ b/extras/hook-scripts/start/post/S31ganesha-start.sh
@@ -4,7 +4,7 @@ OPTSPEC="volname:,gd-workdir:"
VOL=
declare -i EXPORT_ID
ganesha_key="ganesha.enable"
-GANESHA_DIR="/var/run/gluster/shared_storage/nfs-ganesha"
+GANESHA_DIR="/run/gluster/shared_storage/nfs-ganesha"
CONF1="$GANESHA_DIR/ganesha.conf"
GLUSTERD_WORKDIR=
diff --git a/extras/quota/quota_fsck.py b/extras/quota/quota_fsck.py
index 174f2a2a44c..e62f7fc52a3 100755
--- a/extras/quota/quota_fsck.py
+++ b/extras/quota/quota_fsck.py
@@ -156,12 +156,10 @@ def get_quota_xattr_brick(dpath):
xattr_dict = {}
xattr_dict['parents'] = {}
- for xattr in pairs:
+ for xattr in pairs[1:]:
+ xattr = xattr.decode("utf-8")
xattr_key = xattr.split("=")[0]
- if re.search("# file:", xattr_key):
- # skip the file comment
- continue
- elif xattr_key is "":
+ if xattr_key == "":
# skip any empty lines
continue
elif not re.search("quota", xattr_key):
diff --git a/extras/snap_scheduler/gcron.py b/extras/snap_scheduler/gcron.py
index 1127be0e976..0e4df77d481 100755
--- a/extras/snap_scheduler/gcron.py
+++ b/extras/snap_scheduler/gcron.py
@@ -19,10 +19,10 @@ import logging.handlers
import fcntl
-GCRON_TASKS = "/var/run/gluster/shared_storage/snaps/glusterfs_snap_cron_tasks"
+GCRON_TASKS = "/run/gluster/shared_storage/snaps/glusterfs_snap_cron_tasks"
GCRON_CROND_TASK = "/etc/cron.d/glusterfs_snap_cron_tasks"
GCRON_RELOAD_FLAG = "/var/run/gluster/crond_task_reload_flag"
-LOCK_FILE_DIR = "/var/run/gluster/shared_storage/snaps/lock_files/"
+LOCK_FILE_DIR = "/run/gluster/shared_storage/snaps/lock_files/"
log = logging.getLogger("gcron-logger")
start_time = 0.0
@@ -38,7 +38,8 @@ def initLogger(script_name):
sh.setFormatter(formatter)
process = subprocess.Popen(["gluster", "--print-logdir"],
- stdout=subprocess.PIPE)
+ stdout=subprocess.PIPE,
+ universal_newlines=True)
out, err = process.communicate()
if process.returncode == 0:
logfile = os.path.join(out.strip(), script_name[:-3]+".log")
diff --git a/extras/snap_scheduler/snap_scheduler.py b/extras/snap_scheduler/snap_scheduler.py
index 5a29d41c8f8..e8fcc449a9b 100755
--- a/extras/snap_scheduler/snap_scheduler.py
+++ b/extras/snap_scheduler/snap_scheduler.py
@@ -67,7 +67,7 @@ except ImportError:
SCRIPT_NAME = "snap_scheduler"
scheduler_enabled = False
log = logging.getLogger(SCRIPT_NAME)
-SHARED_STORAGE_DIR="/var/run/gluster/shared_storage"
+SHARED_STORAGE_DIR="/run/gluster/shared_storage"
GCRON_DISABLED = SHARED_STORAGE_DIR+"/snaps/gcron_disabled"
GCRON_ENABLED = SHARED_STORAGE_DIR+"/snaps/gcron_enabled"
GCRON_TASKS = SHARED_STORAGE_DIR+"/snaps/glusterfs_snap_cron_tasks"
diff --git a/extras/statedumpparse.rb b/extras/statedumpparse.rb
index 872f284fcb8..1aff43377db 100755
--- a/extras/statedumpparse.rb
+++ b/extras/statedumpparse.rb
@@ -17,25 +17,35 @@ end
# client.c:client_fd_lk_ctx_dump uses a six-dash record separator.
ARRSEP = /^(-{5,6}=-{5,6})?$/
HEAD = /^\[(.*)\]$/
+INPUT_FORMATS = %w[statedump json]
format = 'json'
+input_format = 'statedump'
tz = '+0000'
memstat_select,memstat_reject = //,/\Z./
-human = false
OptionParser.new do |op|
op.banner << " [<] <STATEDUMP>"
- op.on("-f", "--format=F", "json/yaml/memstat") { |s| format = s }
+ op.on("-f", "--format=F", "json/yaml/memstat(-[plain|human|json])") { |s| format = s }
+ op.on("--input-format=F", INPUT_FORMATS.join(?/)) { |s| input_format = s }
op.on("--timezone=T",
"time zone to apply to zoneless timestamps [default UTC]") { |s| tz = s }
- op.on("--memstat-select=RX", "memstat: select memory types maxtching RX") { |s|
+ op.on("--memstat-select=RX", "memstat: select memory types matching RX") { |s|
memstat_select = Regexp.new s
}
- op.on("--memstat-reject=RX", "memstat: reject memory types maxtching RX") { |s|
+ op.on("--memstat-reject=RX", "memstat: reject memory types matching RX") { |s|
memstat_reject = Regexp.new s
}
- op.on("--memstat-human", "memstat: human readable") { human = true }
end.parse!
+
+if format =~ /\Amemstat(?:-(.*))?/
+ memstat_type = $1 || 'plain'
+ unless %w[plain human json].include? memstat_type
+ raise "unknown memstat type #{memstat_type.dump}"
+ end
+ format = 'memstat'
+end
+
repr, logsep = case format
when 'yaml'
require 'yaml'
@@ -50,8 +60,11 @@ else
end
formatter = proc { |e| puts repr.call(e) }
-d1 = {}
+INPUT_FORMATS.include? input_format or raise "unkwown input format '#{input_format}'"
+
+dumpinfo = {}
+# parse a statedump entry
elem_cbk = proc { |s,&cbk|
arraylike = false
s.grep(/\S/).empty? and next
@@ -74,15 +87,8 @@ elem_cbk = proc { |s,&cbk|
body.reject(&:empty?).map { |e|
ea = e.map { |l|
k,v = l.split("=",2)
- [k, begin
- Integer v
- rescue ArgumentError
- begin
- Float v
- rescue ArgumentError
- v
- end
- end]
+ m = /\A(0|-?[1-9]\d*)(\.\d+)?\Z/.match v
+ [k, m ? (m[2] ? Float(v) : Integer(v)) : v]
}
begin
ea.to_h
@@ -93,83 +99,110 @@ elem_cbk = proc { |s,&cbk|
}
if body
- cbk.call [head, arraylike ? body : body[0]]
+ cbk.call [head, arraylike ? body : (body.empty? ? {} : body[0])]
else
STDERR.puts ["WARNING: failed to parse record:", repr.call(s)].join(logsep)
end
}
+# aggregator routine
aggr = case format
when 'memstat'
- mh = {}
+ meminfo = {}
+ # commit memory-related entries to meminfo
proc { |k,r|
case k
when /memusage/
- (mh["GF_MALLOC"]||={})[k] ||= r["size"] if k =~ memstat_select and k !~ memstat_reject
+ (meminfo["GF_MALLOC"]||={})[k] ||= r["size"] if k =~ memstat_select and k !~ memstat_reject
when "mempool"
r.each {|e|
kk = "mempool:#{e['pool-name']}"
- (mh["mempool"]||={})[kk] ||= e["size"] if kk =~ memstat_select and kk !~ memstat_reject
+ (meminfo["mempool"]||={})[kk] ||= e["size"] if kk =~ memstat_select and kk !~ memstat_reject
}
end
}
else
+ # just format data, don't actually aggregate anything
proc { |pair| formatter.call pair }
end
-acc = []
-$<.each { |l|
- l = l.strip
- if l =~ /^(DUMP-(?:START|END)-TIME):\s+(.*)/
- d1["_meta"]||={}
- (d1["_meta"]["date"]||={})[$1] = Time.parse([$2, tz].join " ")
- next
- end
+# processing the data
+case input_format
+when 'statedump'
+ acc = []
+ $<.each { |l|
+ l = l.strip
+ if l =~ /^(DUMP-(?:START|END)-TIME):\s+(.*)/
+ dumpinfo["_meta"]||={}
+ (dumpinfo["_meta"]["date"]||={})[$1] = Time.parse([$2, tz].join " ")
+ next
+ end
- if l =~ HEAD
- elem_cbk.call(acc, &aggr)
- acc = [l]
- next
- end
+ if l =~ HEAD
+ elem_cbk.call(acc, &aggr)
+ acc = [l]
+ next
+ end
- acc << l
-}
-elem_cbk.call(acc, &aggr)
+ acc << l
+ }
+ elem_cbk.call(acc, &aggr)
+when 'json'
+ $<.each { |l|
+ r = JSON.load l
+ case r
+ when Array
+ aggr[r]
+ when Hash
+ dumpinfo.merge! r
+ end
+ }
+end
+# final actions: output aggregated data
case format
when 'memstat'
- hr = proc { |n|
- qa = %w[B kB MB GB]
- q = ((1...qa.size).find {|i| n < (1 << i*10)} || qa.size) - 1
- [n*100 / (1 << q*10) / 100.0, qa[q]].join
- }
-
- ma = mh.values.map(&:to_a).inject(:+)
- totals = mh.map { |coll,h| [coll, h.values.inject(:+)] }.to_h
+ ma = meminfo.values.map(&:to_a).inject(:+)
+ totals = meminfo.map { |coll,h| [coll, h.values.inject(:+)] }.to_h
tt = ma.transpose[1].inject(:+)
- templ = "%{val} %{key}"
- tft = proc { |t| t }
- nft = if human
- nw = [ma.transpose[1], totals.values, tt].flatten.map{|n| hr[n].size}.max
- proc { |n|
- hn = hr[n]
- " " * (nw - hn.size) + hn
+ summary_sep,showm = case memstat_type
+ when 'json'
+ ["", proc { |k,v| puts({type: k, value: v}.to_json) }]
+ when 'plain', 'human'
+ # human-friendly number representation
+ hr = proc { |n|
+ qa = %w[B kB MB GB]
+ q = ((1...qa.size).find {|i| n < (1 << i*10)} || qa.size) - 1
+ "%.2f%s" % [n.to_f / (1 << q*10), qa[q]]
}
+
+ templ = "%{val} %{key}"
+ tft = proc { |t| t }
+ nft = if memstat_type == 'human'
+ nw = [ma.transpose[1], totals.values, tt].flatten.map{|n| hr[n].size}.max
+ proc { |n|
+ hn = hr[n]
+ " " * (nw - hn.size) + hn
+ }
+ else
+ nw = tt.to_s.size
+ proc { |n| "%#{nw}d" % n }
+ end
+ ## Alternative template, key first:
+ # templ = "%{key} %{val}"
+ # tw = ma.transpose[0].map(&:size).max
+ # tft = proc { |t| t + " " * [tw - t.size, 0].max }
+ # nft = (memstat_type == 'human') ? hr : proc { |n| n }
+ ["\n", proc { |k,v| puts templ % {key: tft[k], val: nft[v]} }]
else
- nw = tt.to_s.size
- proc { |n| "%#{nw}d" % n }
+ raise 'this should be impossible'
end
- # templ = "%{key} %{val}"
- # tw = ma.transpose[0].map(&:size).max
- # tft = proc { |t| t + " " * [tw - t.size, 0].max }
- # nft = human ? hr : proc { |n| n }
- showm = proc { |k,v| puts templ % {key: tft[k], val: nft[v]} }
ma.sort_by { |k,v| v }.each(&showm)
- puts
+ print summary_sep
totals.each { |coll,t| showm.call "Total #{coll}", t }
showm.call "TOTAL", tt
else
- formatter.call d1
+ formatter.call dumpinfo
end
diff --git a/extras/systemd/glusterd.service.in b/extras/systemd/glusterd.service.in
index 32542d9f78b..abb0d82911f 100644
--- a/extras/systemd/glusterd.service.in
+++ b/extras/systemd/glusterd.service.in
@@ -1,6 +1,8 @@
[Unit]
Description=GlusterFS, a clustered file-system server
Documentation=man:glusterd(8)
+StartLimitBurst=6
+StartLimitIntervalSec=3600
Requires=@RPCBIND_SERVICE@
After=network.target @RPCBIND_SERVICE@
Before=network-online.target
@@ -10,7 +12,7 @@ Type=forking
PIDFile=@localstatedir@/run/glusterd.pid
LimitNOFILE=65536
Environment="LOG_LEVEL=INFO"
-EnvironmentFile=-@sysconfdir@/sysconfig/glusterd
+EnvironmentFile=-@SYSCONF_DIR@/sysconfig/glusterd
ExecStart=@prefix@/sbin/glusterd -p @localstatedir@/run/glusterd.pid --log-level $LOG_LEVEL $GLUSTERD_OPTIONS
KillMode=process
TimeoutSec=300
@@ -18,7 +20,6 @@ SuccessExitStatus=15
Restart=on-abnormal
RestartSec=60
StartLimitBurst=6
-StartLimitIntervalSec=3600
StartLimitInterval=3600
[Install]
diff --git a/geo-replication/gsyncd.conf.in b/geo-replication/gsyncd.conf.in
index 11e57fdf54f..9688c79fab7 100644
--- a/geo-replication/gsyncd.conf.in
+++ b/geo-replication/gsyncd.conf.in
@@ -123,7 +123,7 @@ type=bool
help=Use this to set Active Passive mode to meta-volume.
[meta-volume-mnt]
-value=/var/run/gluster/shared_storage
+value=/run/gluster/shared_storage
help=Meta Volume or Shared Volume mount path
[allow-network]
diff --git a/geo-replication/syncdaemon/master.py b/geo-replication/syncdaemon/master.py
index 03d871415e6..9501aeae6b5 100644
--- a/geo-replication/syncdaemon/master.py
+++ b/geo-replication/syncdaemon/master.py
@@ -27,7 +27,8 @@ from rconf import rconf
from syncdutils import (Thread, GsyncdError, escape_space_newline,
unescape_space_newline, gauxpfx, escape,
lstat, errno_wrap, FreeObject, lf, matching_disk_gfid,
- NoStimeAvailable, PartialHistoryAvailable)
+ NoStimeAvailable, PartialHistoryAvailable,
+ host_brick_split)
URXTIME = (-1, 0)
@@ -1466,7 +1467,7 @@ class GMasterChangelogMixin(GMasterCommon):
node = rconf.args.resource_remote
node_data = node.split("@")
node = node_data[-1]
- remote_node_ip = node.split(":")[0]
+ remote_node_ip, _ = host_brick_split(node)
self.status.set_slave_node(remote_node_ip)
def changelogs_batch_process(self, changes):
diff --git a/geo-replication/syncdaemon/syncdutils.py b/geo-replication/syncdaemon/syncdutils.py
index 6362c13d5e3..a3df103e76c 100644
--- a/geo-replication/syncdaemon/syncdutils.py
+++ b/geo-replication/syncdaemon/syncdutils.py
@@ -725,7 +725,7 @@ def get_slv_dir_path(slv_host, slv_volume, gfid):
if not isinstance(realpath, int):
basename = os.path.basename(realpath).rstrip('\x00')
dirpath = os.path.dirname(realpath)
- if dirpath is "/":
+ if dirpath == "/":
pargfid = ROOT_GFID
else:
dirpath = dirpath.strip("/")
@@ -894,6 +894,19 @@ class Popen(subprocess.Popen):
self.errfail()
+def host_brick_split(value):
+ """
+ IPv6 compatible way to split and get the host
+ and brick information. Example inputs:
+ node1.example.com:/exports/bricks/brick1/brick
+ fe80::af0f:df82:844f:ef66%utun0:/exports/bricks/brick1/brick
+ """
+ parts = value.split(":")
+ brick = parts[-1]
+ hostparts = parts[0:-1]
+ return (":".join(hostparts), brick)
+
+
class Volinfo(object):
def __init__(self, vol, host='localhost', prelude=[], master=True):
@@ -936,7 +949,7 @@ class Volinfo(object):
@memoize
def bricks(self):
def bparse(b):
- host, dirp = b.find("name").text.split(':', 2)
+ host, dirp = host_brick_split(b.find("name").text)
return {'host': host, 'dir': dirp, 'uuid': b.find("hostUuid").text}
return [bparse(b) for b in self.get('brick')]
@@ -1012,6 +1025,16 @@ class VolinfoFromGconf(object):
def is_hot(self, brickpath):
return False
+ def is_uuid(self, value):
+ try:
+ uuid.UUID(value)
+ return True
+ except ValueError:
+ return False
+
+ def possible_path(self, value):
+ return "/" in value
+
@property
@memoize
def bricks(self):
@@ -1025,8 +1048,22 @@ class VolinfoFromGconf(object):
out = []
for b in bricks_data:
parts = b.split(":")
- bpath = parts[2] if len(parts) == 3 else ""
- out.append({"host": parts[1], "dir": bpath, "uuid": parts[0]})
+ b_uuid = None
+ if self.is_uuid(parts[0]):
+ b_uuid = parts[0]
+ # Set all parts except first
+ parts = parts[1:]
+
+ if self.possible_path(parts[-1]):
+ bpath = parts[-1]
+ # Set all parts except last
+ parts = parts[0:-1]
+
+ out.append({
+ "host": ":".join(parts), # if remaining parts are IPv6 name
+ "dir": bpath,
+ "uuid": b_uuid
+ })
return out
diff --git a/glusterfs.spec.in b/glusterfs.spec.in
index 8ec9e85bf95..b6d63146e14 100644
--- a/glusterfs.spec.in
+++ b/glusterfs.spec.in
@@ -111,6 +111,12 @@
## All %%global definitions should be placed here and keep them sorted
##
+# selinux booleans whose defalut value needs modification
+# these booleans will be consumed by "%%selinux_set_booleans" macro.
+%if ( 0%{?rhel} && 0%{?rhel} >= 8 )
+%global selinuxbooleans rsync_full_access=1 rsync_client=1
+%endif
+
%if ( 0%{?fedora} ) || ( 0%{?rhel} && 0%{?rhel} > 6 )
%global _with_systemd true
%endif
@@ -422,6 +428,14 @@ Requires: python%{_pythonver}-gluster = %{version}-%{release}
Requires: rsync
Requires: util-linux
+# required for setting selinux bools
+%if ( 0%{?rhel} && 0%{?rhel} >= 8 )
+Requires(post): policycoreutils-python-utils
+Requires(postun): policycoreutils-python-utils
+Requires: selinux-policy-targeted
+Requires(post): selinux-policy-targeted
+BuildRequires: selinux-policy-devel
+%endif
%description geo-replication
GlusterFS is a distributed file-system capable of scaling to several
@@ -692,6 +706,9 @@ Requires: %{name}%{?_isa} = %{version}-%{release}
Requires: %{name}-cli%{?_isa} = %{version}-%{release}
Requires: libglusterfs0%{?_isa} = %{version}-%{release}
Requires: libgfchangelog0%{?_isa} = %{version}-%{release}
+%if ( 0%{?fedora} && 0%{?fedora} >= 30 || ( 0%{?rhel} && 0%{?rhel} >= 8 ) )
+Requires: glusterfs-selinux >= 0.1.0-2
+%endif
# some daemons (like quota) use a fuse-mount, glusterfsd is part of -fuse
Requires: %{name}-fuse%{?_isa} = %{version}-%{release}
# self-heal daemon, rebalance, nfs-server etc. are actually clients
@@ -978,6 +995,9 @@ exit 0
%if ( 0%{!?_without_georeplication:1} )
%post geo-replication
+%if ( 0%{?rhel} && 0%{?rhel} >= 8 )
+%selinux_set_booleans %{selinuxbooleans}
+%endif
if [ $1 -ge 1 ]; then
%systemd_postun_with_restart glusterd
fi
@@ -1604,6 +1624,9 @@ exit 0
* Thu May 14 2020 Kaleb S. KEITHLEY <kkeithle@redhat.com>
- refactor, common practice, Issue #1126
+* Mon May 11 2020 Sunny Kumar <sunkumar@redhat.com>
+- added requires policycoreutils-python-utils on rhel8 for geo-replication
+
* Wed Oct 9 2019 Kaleb S. KEITHLEY <kkeithle@redhat.com>
- remove leftover bd xlator cruft
diff --git a/glusterfsd/src/Makefile.am b/glusterfsd/src/Makefile.am
index 7b8d1dbf1fb..a0a778158d8 100644
--- a/glusterfsd/src/Makefile.am
+++ b/glusterfsd/src/Makefile.am
@@ -6,14 +6,15 @@ endif
glusterfsd_SOURCES = glusterfsd.c glusterfsd-mgmt.c
glusterfsd_LDADD = $(top_builddir)/libglusterfs/src/libglusterfs.la \
$(top_builddir)/rpc/rpc-lib/src/libgfrpc.la \
- $(top_builddir)/rpc/xdr/src/libgfxdr.la ${GF_LDADD}
-glusterfsd_LDFLAGS = $(GF_LDFLAGS) $(LIB_DL)
+ $(top_builddir)/rpc/xdr/src/libgfxdr.la $(GF_LDADD) $(LIB_DL)
+glusterfsd_LDFLAGS = $(GF_LDFLAGS)
gf_attach_SOURCES = gf_attach.c
gf_attach_LDADD = $(top_builddir)/libglusterfs/src/libglusterfs.la \
$(top_builddir)/api/src/libgfapi.la \
$(top_builddir)/rpc/rpc-lib/src/libgfrpc.la \
$(top_builddir)/rpc/xdr/src/libgfxdr.la
+gf_attach_LDFLAGS = $(GF_LDFLAGS)
noinst_HEADERS = glusterfsd.h glusterfsd-mem-types.h glusterfsd-messages.h
diff --git a/glusterfsd/src/gf_attach.c b/glusterfsd/src/gf_attach.c
index cb6c0f59b0b..c553b0b1f61 100644
--- a/glusterfsd/src/gf_attach.c
+++ b/glusterfsd/src/gf_attach.c
@@ -19,9 +19,16 @@
#include "xdr-generic.h"
#include "glusterd1-xdr.h"
+/* In seconds */
+#define CONNECT_TIMEOUT 60
+#define REPLY_TIMEOUT 120
+
int done = 0;
int rpc_status;
+pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
+pthread_cond_t cond = PTHREAD_COND_INITIALIZER;
+
struct rpc_clnt_procedure gf_attach_actors[GLUSTERD_BRICK_MAXVALUE] = {
[GLUSTERD_BRICK_NULL] = {"NULL", NULL},
[GLUSTERD_BRICK_OP] = {"BRICK_OP", NULL},
@@ -38,8 +45,12 @@ struct rpc_clnt_program gf_attach_prog = {
int32_t
my_callback(struct rpc_req *req, struct iovec *iov, int count, void *frame)
{
+ pthread_mutex_lock(&mutex);
rpc_status = req->rpc_status;
done = 1;
+ /* Signal main thread which is the only waiter */
+ pthread_cond_signal(&cond);
+ pthread_mutex_unlock(&mutex);
return 0;
}
@@ -48,6 +59,7 @@ int
send_brick_req(xlator_t *this, struct rpc_clnt *rpc, char *path, int op)
{
int ret = -1;
+ struct timespec ts;
struct iobuf *iobuf = NULL;
struct iobref *iobref = NULL;
struct iovec iov = {
@@ -57,7 +69,6 @@ send_brick_req(xlator_t *this, struct rpc_clnt *rpc, char *path, int op)
call_frame_t *frame = NULL;
gd1_mgmt_brick_op_req brick_req;
void *req = &brick_req;
- int i;
brick_req.op = op;
brick_req.name = path;
@@ -87,12 +98,20 @@ send_brick_req(xlator_t *this, struct rpc_clnt *rpc, char *path, int op)
iov.iov_len = ret;
- for (i = 0; i < 60; ++i) {
- if (rpc->conn.connected) {
- break;
- }
- sleep(1);
+ /* Wait for connection */
+ timespec_now_realtime(&ts);
+ ts.tv_sec += CONNECT_TIMEOUT;
+ pthread_mutex_lock(&rpc->conn.lock);
+ {
+ while (!rpc->conn.connected)
+ if (pthread_cond_timedwait(&rpc->conn.cond, &rpc->conn.lock, &ts) ==
+ ETIMEDOUT) {
+ fprintf(stderr, "timeout waiting for RPC connection\n");
+ pthread_mutex_unlock(&rpc->conn.lock);
+ return EXIT_FAILURE;
+ }
}
+ pthread_mutex_unlock(&rpc->conn.lock);
frame = create_frame(this, this->ctx->pool);
if (!frame) {
@@ -104,9 +123,19 @@ send_brick_req(xlator_t *this, struct rpc_clnt *rpc, char *path, int op)
ret = rpc_clnt_submit(rpc, &gf_attach_prog, op, my_callback, &iov, 1, NULL,
0, iobref, frame, NULL, 0, NULL, 0, NULL);
if (!ret) {
- for (i = 0; !done && (i < 120); ++i) {
- sleep(1);
+ /* OK, wait for callback */
+ timespec_now_realtime(&ts);
+ ts.tv_sec += REPLY_TIMEOUT;
+ pthread_mutex_lock(&mutex);
+ {
+ while (!done)
+ if (pthread_cond_timedwait(&cond, &mutex, &ts) == ETIMEDOUT) {
+ fprintf(stderr, "timeout waiting for RPC reply\n");
+ pthread_mutex_unlock(&mutex);
+ return EXIT_FAILURE;
+ }
}
+ pthread_mutex_unlock(&mutex);
}
out:
diff --git a/glusterfsd/src/glusterfsd-mgmt.c b/glusterfsd/src/glusterfsd-mgmt.c
index 45431450185..eaf6796e4c3 100644
--- a/glusterfsd/src/glusterfsd-mgmt.c
+++ b/glusterfsd/src/glusterfsd-mgmt.c
@@ -205,6 +205,7 @@ glusterfs_serialize_reply(rpcsvc_request_t *req, void *arg,
retlen = xdr_serialize_generic(*outmsg, arg, xdrproc);
if (retlen == -1) {
gf_log(THIS->name, GF_LOG_ERROR, "Failed to encode message");
+ GF_FREE(iob);
goto ret;
}
@@ -628,7 +629,7 @@ glusterfs_volume_top_perf(const char *brick_path, dict_t *dict,
goto out;
}
- time = (end.tv_sec - begin.tv_sec) * 1e6 + (end.tv_usec - begin.tv_usec);
+ time = gf_tvdiff(&begin, &end);
throughput = total_blks / time;
gf_log("glusterd", GF_LOG_INFO,
"Throughput %.2f Mbps time %.2f secs "
@@ -684,7 +685,7 @@ glusterfs_volume_top_perf(const char *brick_path, dict_t *dict,
goto out;
}
- time = (end.tv_sec - begin.tv_sec) * 1e6 + (end.tv_usec - begin.tv_usec);
+ time = gf_tvdiff(&begin, &end);
throughput = total_blks / time;
gf_log("glusterd", GF_LOG_INFO,
"Throughput %.2f Mbps time %.2f secs "
diff --git a/glusterfsd/src/glusterfsd.c b/glusterfsd/src/glusterfsd.c
index 5257908bd89..dae41f33fef 100644
--- a/glusterfsd/src/glusterfsd.c
+++ b/glusterfsd/src/glusterfsd.c
@@ -47,12 +47,6 @@
#include <malloc.h>
#endif
-#ifdef HAVE_MALLOC_STATS
-#ifdef DEBUG
-#include <mcheck.h>
-#endif
-#endif
-
#include <glusterfs/xlator.h>
#include <glusterfs/glusterfs.h>
#include <glusterfs/compat.h>
@@ -1985,7 +1979,7 @@ parse_cmdline(int argc, char *argv[], glusterfs_ctx_t *ctx)
struct stat stbuf = {
0,
};
- char timestr[32];
+ char timestr[GF_TIMESTR_SIZE];
char tmp_logfile[1024] = {0};
char *tmp_logfile_dyn = NULL;
char *tmp_logfilebase = NULL;
@@ -2089,8 +2083,8 @@ parse_cmdline(int argc, char *argv[], glusterfs_ctx_t *ctx)
if (((ret == 0) &&
(S_ISREG(stbuf.st_mode) || S_ISLNK(stbuf.st_mode))) ||
(ret == -1)) {
- /* Have separate logfile per run */
- gf_time_fmt(timestr, sizeof timestr, time(NULL), gf_timefmt_FT);
+ /* Have separate logfile per run. */
+ gf_time_fmt(timestr, sizeof timestr, gf_time(), gf_timefmt_FT);
sprintf(tmp_logfile, "%s.%s.%d", cmd_args->log_file, timestr,
getpid());
diff --git a/heal/src/glfs-heal.c b/heal/src/glfs-heal.c
index ef53dfc0ccb..bf4b47f8760 100644
--- a/heal/src/glfs-heal.c
+++ b/heal/src/glfs-heal.c
@@ -1055,6 +1055,10 @@ glfsh_set_heal_options(glfs_t *fs, gf_xl_afr_op_t heal_op)
if (ret)
goto out;
+ ret = glfs_set_xlator_option(fs, "*-replicate-*", "halo-enabled", "off");
+ if (ret)
+ goto out;
+
if ((heal_op != GF_SHD_OP_SBRAIN_HEAL_FROM_BIGGER_FILE) &&
(heal_op != GF_SHD_OP_SBRAIN_HEAL_FROM_BRICK) &&
(heal_op != GF_SHD_OP_SBRAIN_HEAL_FROM_LATEST_MTIME))
diff --git a/libglusterfs/src/Makefile.am b/libglusterfs/src/Makefile.am
index cd9d28e3cdc..385e8ef4600 100644
--- a/libglusterfs/src/Makefile.am
+++ b/libglusterfs/src/Makefile.am
@@ -12,7 +12,8 @@ libglusterfs_la_CPPFLAGS = $(GF_CPPFLAGS) -D__USE_FILE_OFFSET64 \
-DSBIN_DIR=\"$(sbindir)\" -I$(CONTRIBDIR)/timer-wheel \
-I$(CONTRIBDIR)/xxhash
-libglusterfs_la_LIBADD = $(ZLIB_LIBS) $(MATH_LIB) $(UUID_LIBS)
+libglusterfs_la_LIBADD = $(ZLIB_LIBS) $(MATH_LIB) $(UUID_LIBS) $(LIB_DL) \
+ $(URCU_LIBS) $(URCU_CDS_LIBS)
libglusterfs_la_LDFLAGS = -version-info $(LIBGLUSTERFS_LT_VERSION) $(GF_LDFLAGS) \
-export-symbols $(top_srcdir)/libglusterfs/src/libglusterfs.sym
diff --git a/libglusterfs/src/client_t.c b/libglusterfs/src/client_t.c
index 4cd22f4920c..9d377c3c2e1 100644
--- a/libglusterfs/src/client_t.c
+++ b/libglusterfs/src/client_t.c
@@ -314,8 +314,6 @@ client_destroy(client_t *client)
clienttable = client->this->ctx->clienttable;
- LOCK_DESTROY(&client->scratch_ctx.lock);
-
LOCK(&clienttable->lock);
{
clienttable->cliententries[client->tbl_index].client = NULL;
@@ -333,6 +331,8 @@ client_destroy(client_t *client)
if (client->subdir_inode)
inode_unref(client->subdir_inode);
+ LOCK_DESTROY(&client->scratch_ctx.lock);
+
GF_FREE(client->auth.data);
GF_FREE(client->auth.username);
GF_FREE(client->auth.passwd);
diff --git a/libglusterfs/src/common-utils.c b/libglusterfs/src/common-utils.c
index 585952d7641..682cbf28055 100644
--- a/libglusterfs/src/common-utils.c
+++ b/libglusterfs/src/common-utils.c
@@ -37,6 +37,9 @@
#ifndef GF_LINUX_HOST_OS
#include <sys/resource.h>
#endif
+#ifdef HAVE_SYNCFS_SYS
+#include <sys/syscall.h>
+#endif
#include "glusterfs/compat-errno.h"
#include "glusterfs/common-utils.h"
@@ -50,6 +53,7 @@
#include "xxhash.h"
#include <ifaddrs.h>
#include "glusterfs/libglusterfs-messages.h"
+#include "glusterfs/glusterfs-acl.h"
#ifdef __FreeBSD__
#include <pthread_np.h>
#undef BIT_SET
@@ -75,6 +79,15 @@ char *vol_type_str[] = {
typedef int32_t (*rw_op_t)(int32_t fd, char *buf, int32_t size);
typedef int32_t (*rwv_op_t)(int32_t fd, const struct iovec *buf, int32_t size);
+char *xattrs_to_heal[] = {"user.",
+ POSIX_ACL_ACCESS_XATTR,
+ POSIX_ACL_DEFAULT_XATTR,
+ QUOTA_LIMIT_KEY,
+ QUOTA_LIMIT_OBJECTS_KEY,
+ GF_SELINUX_XATTR_KEY,
+ GF_XATTR_MDATA_KEY,
+ NULL};
+
void
gf_xxh64_wrapper(const unsigned char *data, size_t const len,
unsigned long long const seed, char *xxh64)
@@ -425,7 +438,7 @@ gf_resolve_path_parent(const char *path)
GF_VALIDATE_OR_GOTO(THIS->name, path, out);
- if (strlen(path) <= 0) {
+ if (0 == strlen(path)) {
gf_msg_callingfn(THIS->name, GF_LOG_DEBUG, 0, LG_MSG_INVALID_STRING,
"invalid string for 'path'");
goto out;
@@ -564,8 +577,14 @@ struct dnscache *
gf_dnscache_init(time_t ttl)
{
struct dnscache *cache = GF_MALLOC(sizeof(*cache), gf_common_mt_dnscache);
- if (cache) {
- cache->cache_dict = NULL;
+ if (!cache)
+ return NULL;
+
+ cache->cache_dict = dict_new();
+ if (!cache->cache_dict) {
+ GF_FREE(cache);
+ cache = NULL;
+ } else {
cache->ttl = ttl;
}
@@ -573,6 +592,20 @@ gf_dnscache_init(time_t ttl)
}
/**
+ * gf_dnscache_deinit -- cleanup resources used by struct dnscache
+ */
+void
+gf_dnscache_deinit(struct dnscache *cache)
+{
+ if (!cache) {
+ gf_msg_plain(GF_LOG_WARNING, "dnscache is NULL");
+ return;
+ }
+ dict_unref(cache->cache_dict);
+ GF_FREE(cache);
+}
+
+/**
* gf_dnscache_entry_init -- Initialize a dnscache entry
*
* @return: SUCCESS: Pointer to an allocated dnscache entry struct
@@ -620,12 +653,6 @@ gf_rev_dns_lookup_cached(const char *ip, struct dnscache *dnscache)
if (!dnscache)
goto out;
- if (!dnscache->cache_dict) {
- dnscache->cache_dict = dict_new();
- if (!dnscache->cache_dict) {
- goto out;
- }
- }
cache = dnscache->cache_dict;
/* Quick cache lookup to see if we already hold it */
@@ -633,7 +660,7 @@ gf_rev_dns_lookup_cached(const char *ip, struct dnscache *dnscache)
if (entrydata) {
dnsentry = (struct dnscache_entry *)entrydata->data;
/* First check the TTL & timestamp */
- if (time(NULL) - dnsentry->timestamp > dnscache->ttl) {
+ if (gf_time() - dnsentry->timestamp > dnscache->ttl) {
gf_dnscache_entry_deinit(dnsentry);
entrydata->data = NULL; /* Mark this as 'null' so
* dict_del () doesn't try free
@@ -664,23 +691,16 @@ gf_rev_dns_lookup_cached(const char *ip, struct dnscache *dnscache)
from_cache = _gf_false;
out:
/* Insert into the cache */
- if (fqdn && !from_cache) {
+ if (fqdn && !from_cache && ip) {
struct dnscache_entry *entry = gf_dnscache_entry_init();
- if (!entry) {
- goto out;
- }
- entry->fqdn = fqdn;
- if (!ip) {
- gf_dnscache_entry_deinit(entry);
- goto out;
+ if (entry) {
+ entry->fqdn = fqdn;
+ entry->ip = gf_strdup(ip);
+ entry->timestamp = gf_time();
+ entrydata = bin_to_data(entry, sizeof(*entry));
+ dict_set(cache, (char *)ip, entrydata);
}
-
- entry->ip = gf_strdup(ip);
- entry->timestamp = time(NULL);
-
- entrydata = bin_to_data(entry, sizeof(*entry));
- dict_set(cache, (char *)ip, entrydata);
}
return fqdn;
}
@@ -889,7 +909,7 @@ gf_print_trace(int32_t signum, glusterfs_ctx_t *ctx)
char msg[1024] = {
0,
};
- char timestr[64] = {
+ char timestr[GF_TIMESTR_SIZE] = {
0,
};
call_stack_t *stack = NULL;
@@ -929,7 +949,7 @@ gf_print_trace(int32_t signum, glusterfs_ctx_t *ctx)
{
/* Dump the timestamp of the crash too, so the previous logs
can be related */
- gf_time_fmt(timestr, sizeof timestr, time(NULL), gf_timefmt_FT);
+ gf_time_fmt(timestr, sizeof timestr, gf_time(), gf_timefmt_FT);
gf_msg_plain_nomem(GF_LOG_ALERT, "time of crash: ");
gf_msg_plain_nomem(GF_LOG_ALERT, timestr);
}
@@ -3113,7 +3133,7 @@ get_mem_size()
memsize = page_size * num_pages;
#endif
-#if defined GF_DARWIN_HOST_OS
+#if defined GF_DARWIN_HOST_OS || defined __FreeBSD__
size_t len = sizeof(memsize);
int name[] = {CTL_HW, HW_PHYSMEM};
@@ -4127,6 +4147,14 @@ gf_skip_header_section(int fd, int header_len)
gf_boolean_t
gf_is_pid_running(int pid)
{
+#ifdef __FreeBSD__
+ int ret = -1;
+
+ ret = sys_kill(pid, 0);
+ if (ret < 0) {
+ return _gf_false;
+ }
+#else
char fname[32] = {
0,
};
@@ -4140,6 +4168,7 @@ gf_is_pid_running(int pid)
}
sys_close(fd);
+#endif
return _gf_true;
}
@@ -4381,7 +4410,7 @@ gf_backtrace_end(char *buf, size_t frames)
frames = min(frames, GF_BACKTRACE_LEN - pos - 1);
- if (frames <= 0)
+ if (0 == frames)
return;
memset(buf + pos, ')', frames);
@@ -4704,8 +4733,9 @@ recursive_rmdir(const char *delete_path)
goto out;
}
- GF_SKIP_IRRELEVANT_ENTRIES(entry, dir, scratch);
- while (entry) {
+ while ((entry = sys_readdir(dir, scratch))) {
+ if (gf_irrelevant_entry(entry))
+ continue;
snprintf(path, PATH_MAX, "%s/%s", delete_path, entry->d_name);
ret = sys_lstat(path, &st);
if (ret == -1) {
@@ -4731,8 +4761,6 @@ recursive_rmdir(const char *delete_path)
gf_msg_debug(this->name, 0, "%s %s",
ret ? "Failed to remove" : "Removed", entry->d_name);
-
- GF_SKIP_IRRELEVANT_ENTRIES(entry, dir, scratch);
}
ret = sys_closedir(dir);
@@ -5412,3 +5440,26 @@ gf_nanosleep(uint64_t nsec)
return ret;
}
+
+int
+gf_syncfs(int fd)
+{
+ int ret = 0;
+#if defined(HAVE_SYNCFS)
+ /* Linux with glibc recent enough. */
+ ret = syncfs(fd);
+#elif defined(HAVE_SYNCFS_SYS)
+ /* Linux with no library function. */
+ ret = syscall(SYS_syncfs, fd);
+#else
+ /* Fallback to generic UNIX stuff. */
+ sync();
+#endif
+ return ret;
+}
+
+char **
+get_xattrs_to_heal()
+{
+ return xattrs_to_heal;
+}
diff --git a/libglusterfs/src/ctx.c b/libglusterfs/src/ctx.c
index 4a001c29209..3d890b04ec9 100644
--- a/libglusterfs/src/ctx.c
+++ b/libglusterfs/src/ctx.c
@@ -37,8 +37,12 @@ glusterfs_ctx_new()
ctx->log.loglevel = DEFAULT_LOG_LEVEL;
-#ifdef RUN_WITH_VALGRIND
- ctx->cmd_args.valgrind = _gf_true;
+#if defined(RUN_WITH_MEMCHECK)
+ ctx->cmd_args.vgtool = _gf_memcheck;
+#elif defined(RUN_WITH_DRD)
+ ctx->cmd_args.vgtool = _gf_drd;
+#else
+ ctx->cmd_args.vgtool = _gf_none;
#endif
/* lock is never destroyed! */
diff --git a/libglusterfs/src/dict.c b/libglusterfs/src/dict.c
index a1a6ad12527..1d9be9217a6 100644
--- a/libglusterfs/src/dict.c
+++ b/libglusterfs/src/dict.c
@@ -98,6 +98,7 @@ get_new_dict_full(int size_hint)
}
dict->free_pair.key = NULL;
+ dict->totkvlen = 0;
LOCK_INIT(&dict->lock);
return dict;
@@ -410,6 +411,7 @@ dict_set_lk(dict_t *this, char *key, const int key_len, data_t *value,
if (pair) {
data_t *unref_data = pair->value;
pair->value = data_ref(value);
+ this->totkvlen += (value->len - unref_data->len);
data_unref(unref_data);
if (key_free)
GF_FREE(key);
@@ -445,6 +447,7 @@ dict_set_lk(dict_t *this, char *key, const int key_len, data_t *value,
}
pair->key_hash = key_hash;
pair->value = data_ref(value);
+ this->totkvlen += (keylen + 1 + value->len);
/* If the divisor is 1, the modulo is always 0,
* in such case avoid hash calculation.
@@ -642,6 +645,7 @@ dict_deln(dict_t *this, char *key, const int keylen)
else
this->members[hashval] = pair->hash_next;
+ this->totkvlen -= pair->value->len;
data_unref(pair->value);
if (pair->prev)
@@ -652,6 +656,7 @@ dict_deln(dict_t *this, char *key, const int keylen)
if (pair->next)
pair->next->prev = pair->prev;
+ this->totkvlen -= (strlen(pair->key) + 1);
GF_FREE(pair->key);
if (pair == &this->free_pair) {
this->free_pair.key = NULL;
@@ -701,6 +706,7 @@ dict_destroy(dict_t *this)
prev = pair;
}
+ this->totkvlen = 0;
if (this->members != &this->members_internal) {
mem_put(this->members);
}
@@ -1105,117 +1111,146 @@ data_to_int64(data_t *data)
{
VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_INT, "null", -1);
- return (int64_t)strtoull(data->data, NULL, 0);
+ char *endptr = NULL;
+ int64_t value = 0;
+
+ errno = 0;
+ value = strtoll(data->data, &endptr, 0);
+
+ if (endptr && *endptr != '\0')
+ /* Unrecognized characters at the end of string. */
+ errno = EINVAL;
+ if (errno) {
+ gf_msg_callingfn("dict", GF_LOG_WARNING, errno,
+ LG_MSG_DATA_CONVERSION_ERROR,
+ "Error in data conversion: '%s' can't "
+ "be represented as int64_t",
+ data->data);
+ return -1;
+ }
+ return value;
}
+/* Like above but implies signed range check. */
+
+#define DATA_TO_RANGED_SIGNED(endptr, value, data, type, min, max) \
+ do { \
+ errno = 0; \
+ value = strtoll(data->data, &endptr, 0); \
+ if (endptr && *endptr != '\0') \
+ errno = EINVAL; \
+ if (errno || value > max || value < min) { \
+ gf_msg_callingfn("dict", GF_LOG_WARNING, errno, \
+ LG_MSG_DATA_CONVERSION_ERROR, \
+ "Error in data conversion: '%s' can't " \
+ "be represented as " #type, \
+ data->data); \
+ return -1; \
+ } \
+ return (type)value; \
+ } while (0)
+
int32_t
data_to_int32(data_t *data)
{
- VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_INT, "null", -1);
+ char *endptr = NULL;
+ int64_t value = 0;
- return strtoul(data->data, NULL, 0);
+ VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_INT, "null", -1);
+ DATA_TO_RANGED_SIGNED(endptr, value, data, int32_t, INT_MIN, INT_MAX);
}
int16_t
data_to_int16(data_t *data)
{
- VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_INT, "null", -1);
+ char *endptr = NULL;
+ int64_t value = 0;
- int16_t value = 0;
-
- errno = 0;
- value = strtol(data->data, NULL, 0);
-
- if ((value > SHRT_MAX) || (value < SHRT_MIN)) {
- errno = ERANGE;
- gf_msg_callingfn("dict", GF_LOG_WARNING, errno,
- LG_MSG_DATA_CONVERSION_ERROR,
- "Error in data"
- " conversion: detected overflow");
- return -1;
- }
-
- return (int16_t)value;
+ VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_INT, "null", -1);
+ DATA_TO_RANGED_SIGNED(endptr, value, data, int16_t, SHRT_MIN, SHRT_MAX);
}
int8_t
data_to_int8(data_t *data)
{
+ char *endptr = NULL;
+ int64_t value = 0;
+
VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_INT, "null", -1);
+ DATA_TO_RANGED_SIGNED(endptr, value, data, int8_t, CHAR_MIN, CHAR_MAX);
+}
+
+uint64_t
+data_to_uint64(data_t *data)
+{
+ VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_UINT, "null", -1);
- int8_t value = 0;
+ char *endptr = NULL;
+ uint64_t value = 0;
errno = 0;
- value = strtol(data->data, NULL, 0);
+ value = strtoull(data->data, &endptr, 0);
- if ((value > SCHAR_MAX) || (value < SCHAR_MIN)) {
- errno = ERANGE;
+ if (endptr && *endptr != '\0')
+ errno = EINVAL;
+ if (errno) {
gf_msg_callingfn("dict", GF_LOG_WARNING, errno,
LG_MSG_DATA_CONVERSION_ERROR,
- "Error in data"
- " conversion: detected overflow");
+ "Error in data conversion: '%s' can't "
+ "be represented as uint64_t",
+ data->data);
return -1;
}
-
- return (int8_t)value;
+ return value;
}
-uint64_t
-data_to_uint64(data_t *data)
-{
- VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_UINT, "null", -1);
+/* Like above but implies unsigned range check. */
- return strtoll(data->data, NULL, 0);
-}
+#define DATA_TO_RANGED_UNSIGNED(endptr, value, data, type, max) \
+ do { \
+ errno = 0; \
+ value = strtoull(data->data, &endptr, 0); \
+ if (endptr && *endptr != '\0') \
+ errno = EINVAL; \
+ if (errno || value > max) { \
+ gf_msg_callingfn("dict", GF_LOG_WARNING, errno, \
+ LG_MSG_DATA_CONVERSION_ERROR, \
+ "Error in data conversion: '%s' can't " \
+ "be represented as " #type, \
+ data->data); \
+ return -1; \
+ } \
+ return (type)value; \
+ } while (0)
uint32_t
data_to_uint32(data_t *data)
{
- VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_UINT, "null", -1);
+ char *endptr = NULL;
+ uint64_t value = 0;
- return strtol(data->data, NULL, 0);
+ VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_UINT, "null", -1);
+ DATA_TO_RANGED_UNSIGNED(endptr, value, data, uint32_t, UINT_MAX);
}
uint16_t
data_to_uint16(data_t *data)
{
- VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_UINT, "null", -1);
-
- uint16_t value = 0;
-
- errno = 0;
- value = strtol(data->data, NULL, 0);
+ char *endptr = NULL;
+ uint64_t value = 0;
- if ((USHRT_MAX - value) < 0) {
- errno = ERANGE;
- gf_msg_callingfn("dict", GF_LOG_WARNING, errno,
- LG_MSG_DATA_CONVERSION_ERROR,
- "Error in data conversion: "
- "overflow detected");
- return -1;
- }
-
- return (uint16_t)value;
+ VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_UINT, "null", -1);
+ DATA_TO_RANGED_UNSIGNED(endptr, value, data, uint16_t, USHRT_MAX);
}
uint8_t
data_to_uint8(data_t *data)
{
- VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_UINT, "null", -1);
-
- errno = 0;
- uint32_t value = strtol(data->data, NULL, 0);
-
- if ((UCHAR_MAX - (uint8_t)value) < 0) {
- errno = ERANGE;
- gf_msg_callingfn("dict", GF_LOG_WARNING, errno,
- LG_MSG_DATA_CONVERSION_ERROR,
- "data "
- "conversion overflow detected");
- return -1;
- }
+ char *endptr = NULL;
+ uint64_t value = 0;
- return (uint8_t)value;
+ VALIDATE_DATA_AND_LOG(data, GF_DATA_TYPE_UINT, "null", -1);
+ DATA_TO_RANGED_UNSIGNED(endptr, value, data, uint8_t, UCHAR_MAX);
}
char *
@@ -1460,32 +1495,13 @@ fail:
* -val error, val = errno
*/
-int
-dict_get_with_ref(dict_t *this, char *key, data_t **data)
-{
- if (!this || !key || !data) {
- gf_msg_callingfn("dict", GF_LOG_WARNING, EINVAL, LG_MSG_INVALID_ARG,
- "dict OR key (%s) is NULL", key);
- return -EINVAL;
- }
-
- return dict_get_with_refn(this, key, strlen(key), data);
-}
-
-int
+static int
dict_get_with_refn(dict_t *this, char *key, const int keylen, data_t **data)
{
data_pair_t *pair = NULL;
int ret = -ENOENT;
uint32_t hash;
- if (!this || !key || !data) {
- gf_msg_callingfn("dict", GF_LOG_WARNING, EINVAL, LG_MSG_INVALID_ARG,
- "dict OR key (%s) is NULL", key);
- ret = -EINVAL;
- goto err;
- }
-
hash = (uint32_t)XXH64(key, keylen, 0);
LOCK(&this->lock);
@@ -1498,10 +1514,22 @@ dict_get_with_refn(dict_t *this, char *key, const int keylen, data_t **data)
}
}
UNLOCK(&this->lock);
-err:
+
return ret;
}
+int
+dict_get_with_ref(dict_t *this, char *key, data_t **data)
+{
+ if (!this || !key || !data) {
+ gf_msg_callingfn("dict", GF_LOG_WARNING, EINVAL, LG_MSG_INVALID_ARG,
+ "dict OR key (%s) is NULL", key);
+ return -EINVAL;
+ }
+
+ return dict_get_with_refn(this, key, strlen(key), data);
+}
+
static int
data_to_ptr_common(data_t *data, void **val)
{
@@ -1675,7 +1703,7 @@ dict_get_int8(dict_t *this, char *key, int8_t *val)
data_t *data = NULL;
int ret = 0;
- if (!this || !key || !val) {
+ if (!val) {
ret = -EINVAL;
goto err;
}
@@ -1721,7 +1749,7 @@ dict_get_int16(dict_t *this, char *key, int16_t *val)
data_t *data = NULL;
int ret = 0;
- if (!this || !key || !val) {
+ if (!val) {
ret = -EINVAL;
goto err;
}
@@ -1793,7 +1821,7 @@ dict_get_int32(dict_t *this, char *key, int32_t *val)
data_t *data = NULL;
int ret = 0;
- if (!this || !key || !val) {
+ if (!val) {
ret = -EINVAL;
goto err;
}
@@ -1858,7 +1886,7 @@ dict_get_int64(dict_t *this, char *key, int64_t *val)
data_t *data = NULL;
int ret = 0;
- if (!this || !key || !val) {
+ if (!val) {
ret = -EINVAL;
goto err;
}
@@ -1903,7 +1931,7 @@ dict_get_uint16(dict_t *this, char *key, uint16_t *val)
data_t *data = NULL;
int ret = 0;
- if (!this || !key || !val) {
+ if (!val) {
ret = -EINVAL;
goto err;
}
@@ -1948,7 +1976,7 @@ dict_get_uint32(dict_t *this, char *key, uint32_t *val)
data_t *data = NULL;
int ret = 0;
- if (!this || !key || !val) {
+ if (!val) {
ret = -EINVAL;
goto err;
}
@@ -1993,7 +2021,7 @@ dict_get_uint64(dict_t *this, char *key, uint64_t *val)
data_t *data = NULL;
int ret = 0;
- if (!this || !key || !val) {
+ if (!val) {
ret = -EINVAL;
goto err;
}
@@ -2146,7 +2174,7 @@ _dict_modify_flag(dict_t *this, char *key, int flag, int op)
strcpy(pair->key, key);
pair->key_hash = hash;
pair->value = data_ref(data);
-
+ this->totkvlen += (strlen(key) + 1 + data->len);
hashval = hash % this->hash_size;
pair->hash_next = this->members[hashval];
this->members[hashval] = pair;
@@ -2216,7 +2244,7 @@ dict_get_double(dict_t *this, char *key, double *val)
data_t *data = NULL;
int ret = 0;
- if (!this || !key || !val) {
+ if (!val) {
ret = -EINVAL;
goto err;
}
@@ -2299,7 +2327,7 @@ dict_get_ptr(dict_t *this, char *key, void **ptr)
data_t *data = NULL;
int ret = 0;
- if (!this || !key || !ptr) {
+ if (!ptr) {
ret = -EINVAL;
goto err;
}
@@ -2329,7 +2357,7 @@ dict_get_ptr_and_len(dict_t *this, char *key, void **ptr, int *len)
data_t *data = NULL;
int ret = 0;
- if (!this || !key || !ptr) {
+ if (!ptr) {
ret = -EINVAL;
goto err;
}
@@ -2387,7 +2415,7 @@ dict_get_str(dict_t *this, char *key, char **str)
data_t *data = NULL;
int ret = -EINVAL;
- if (!this || !key || !str) {
+ if (!str) {
goto err;
}
ret = dict_get_with_ref(this, key, &data);
@@ -2561,7 +2589,7 @@ dict_get_bin(dict_t *this, char *key, void **bin)
data_t *data = NULL;
int ret = -EINVAL;
- if (!this || !key || !bin) {
+ if (!bin) {
goto err;
}
@@ -2664,7 +2692,7 @@ dict_get_gfuuid(dict_t *this, char *key, uuid_t *gfid)
data_t *data = NULL;
int ret = -EINVAL;
- if (!this || !key || !gfid) {
+ if (!gfid) {
goto err;
}
ret = dict_get_with_ref(this, key, &data);
@@ -2697,7 +2725,7 @@ dict_get_mdata(dict_t *this, char *key, struct mdata_iatt *mdata)
data_t *data = NULL;
int ret = -EINVAL;
- if (!this || !key || !mdata) {
+ if (!mdata) {
goto err;
}
ret = dict_get_with_ref(this, key, &data);
@@ -2735,7 +2763,7 @@ dict_get_iatt(dict_t *this, char *key, struct iatt *iatt)
data_t *data = NULL;
int ret = -EINVAL;
- if (!this || !key || !iatt) {
+ if (!iatt) {
goto err;
}
ret = dict_get_with_ref(this, key, &data);
@@ -2872,8 +2900,7 @@ dict_serialized_length_lk(dict_t *this)
{
int ret = -EINVAL;
int count = this->count;
- int len = DICT_HDR_LEN;
- data_pair_t *pair = this->members_list;
+ const int keyhdrlen = DICT_DATA_HDR_KEY_LEN + DICT_DATA_HDR_VAL_LEN;
if (count < 0) {
gf_smsg("dict", GF_LOG_ERROR, EINVAL, LG_MSG_COUNT_LESS_THAN_ZERO,
@@ -2881,41 +2908,7 @@ dict_serialized_length_lk(dict_t *this)
goto out;
}
- while (count) {
- if (!pair) {
- gf_smsg("dict", GF_LOG_ERROR, EINVAL, LG_MSG_PAIRS_LESS_THAN_COUNT,
- NULL);
- goto out;
- }
-
- len += DICT_DATA_HDR_KEY_LEN + DICT_DATA_HDR_VAL_LEN;
-
- if (!pair->key) {
- gf_smsg("dict", GF_LOG_ERROR, EINVAL, LG_MSG_NULL_PTR, NULL);
- goto out;
- }
-
- len += strlen(pair->key) + 1 /* for '\0' */;
-
- if (!pair->value) {
- gf_smsg("dict", GF_LOG_ERROR, EINVAL, LG_MSG_NULL_PTR, NULL);
- goto out;
- }
-
- if (pair->value->len < 0) {
- gf_smsg("dict", GF_LOG_ERROR, EINVAL,
- LG_MSG_VALUE_LENGTH_LESS_THAN_ZERO, "len=%d",
- pair->value->len, NULL);
- goto out;
- }
-
- len += pair->value->len;
-
- pair = pair->next;
- count--;
- }
-
- ret = len;
+ ret = DICT_HDR_LEN + this->totkvlen + (count * keyhdrlen);
out:
return ret;
}
diff --git a/libglusterfs/src/event.c b/libglusterfs/src/event.c
index 235128b6044..402c253ca25 100644
--- a/libglusterfs/src/event.c
+++ b/libglusterfs/src/event.c
@@ -17,6 +17,7 @@
#include <string.h>
#include "glusterfs/gf-event.h"
+#include "glusterfs/timespec.h"
#include "glusterfs/common-utils.h"
#include "glusterfs/libglusterfs-messages.h"
#include "glusterfs/syscall.h"
@@ -266,7 +267,7 @@ gf_event_dispatch_destroy(struct event_pool *event_pool)
if (sys_write(fd[1], "dummy", 6) == -1) {
break;
}
- clock_gettime(CLOCK_REALTIME, &sleep_till);
+ timespec_now_realtime(&sleep_till);
sleep_till.tv_sec += 1;
ret = pthread_cond_timedwait(&event_pool->cond, &event_pool->mutex,
&sleep_till);
diff --git a/libglusterfs/src/events.c b/libglusterfs/src/events.c
index 6d1e3836477..33157549897 100644
--- a/libglusterfs/src/events.c
+++ b/libglusterfs/src/events.c
@@ -40,6 +40,7 @@ _gf_event(eventtypes_t event, const char *fmt, ...)
char *host = NULL;
struct addrinfo hints;
struct addrinfo *result = NULL;
+ struct addrinfo *iter_result_ptr = NULL;
xlator_t *this = THIS;
char *volfile_server_transport = NULL;
@@ -51,13 +52,6 @@ _gf_event(eventtypes_t event, const char *fmt, ...)
goto out;
}
- /* Initialize UDP socket */
- sock = socket(AF_INET, SOCK_DGRAM, 0);
- if (sock < 0) {
- ret = EVENT_ERROR_SOCKET;
- goto out;
- }
-
if (ctx) {
volfile_server_transport = ctx->cmd_args.volfile_server_transport;
}
@@ -66,7 +60,6 @@ _gf_event(eventtypes_t event, const char *fmt, ...)
}
/* host = NULL returns localhost */
- host = NULL;
if (ctx && ctx->cmd_args.volfile_server &&
(strcmp(volfile_server_transport, "unix"))) {
/* If it is client code then volfile_server is set
@@ -84,6 +77,24 @@ _gf_event(eventtypes_t event, const char *fmt, ...)
goto out;
}
+ // iterate over the result and break when socket creation is success.
+ for (iter_result_ptr = result; iter_result_ptr != NULL;
+ iter_result_ptr = iter_result_ptr->ai_next) {
+ sock = socket(iter_result_ptr->ai_family, iter_result_ptr->ai_socktype,
+ iter_result_ptr->ai_protocol);
+ if (sock != -1) {
+ break;
+ }
+ }
+ /*
+ * If none of the addrinfo structures lead to a successful socket
+ * creation, socket creation has failed.
+ */
+ if (sock < 0) {
+ ret = EVENT_ERROR_SOCKET;
+ goto out;
+ }
+
va_start(arguments, fmt);
ret = gf_vasprintf(&msg, fmt, arguments);
va_end(arguments);
@@ -93,7 +104,7 @@ _gf_event(eventtypes_t event, const char *fmt, ...)
goto out;
}
- ret = gf_asprintf(&eventstr, "%u %d %s", (unsigned)time(NULL), event, msg);
+ ret = gf_asprintf(&eventstr, "%u %d %s", (unsigned)gf_time(), event, msg);
GF_FREE(msg);
if (ret <= 0) {
ret = EVENT_ERROR_MSG_FORMAT;
diff --git a/libglusterfs/src/gidcache.c b/libglusterfs/src/gidcache.c
index 40fcffbb35e..64a93802f76 100644
--- a/libglusterfs/src/gidcache.c
+++ b/libglusterfs/src/gidcache.c
@@ -10,6 +10,7 @@
#include "glusterfs/gidcache.h"
#include "glusterfs/mem-pool.h"
+#include "glusterfs/common-utils.h"
/*
* We treat this as a very simple set-associative LRU cache, with entries aged
@@ -64,7 +65,7 @@ gid_cache_lookup(gid_cache_t *cache, uint64_t id, uint64_t uid, uint64_t gid)
time_t now;
const gid_list_t *agl;
- now = time(NULL);
+ now = gf_time();
LOCK(&cache->gc_lock);
bucket = id % cache->gc_nbuckets;
agl = BUCKET_START(cache->gc_cache, bucket);
@@ -132,7 +133,7 @@ gid_cache_add(gid_cache_t *cache, gid_list_t *gl)
if (!cache->gc_max_age)
return 0;
- now = time(NULL);
+ now = gf_time();
LOCK(&cache->gc_lock);
/*
diff --git a/libglusterfs/src/glusterfs/common-utils.h b/libglusterfs/src/glusterfs/common-utils.h
index 13e413c3632..f297fdab5c9 100644
--- a/libglusterfs/src/glusterfs/common-utils.h
+++ b/libglusterfs/src/glusterfs/common-utils.h
@@ -18,6 +18,7 @@
#include <string.h>
#include <assert.h>
#include <pthread.h>
+#include <unistd.h>
#include <openssl/md5.h>
#ifndef GF_BSD_HOST_OS
#include <alloca.h>
@@ -26,6 +27,11 @@
#include <fnmatch.h>
#include <uuid/uuid.h>
+/* FreeBSD, etc. */
+#ifndef __BITS_PER_LONG
+#define __BITS_PER_LONG (CHAR_BIT * (sizeof(long)))
+#endif
+
#ifndef ffsll
#define ffsll(x) __builtin_ffsll(x)
#endif
@@ -148,6 +154,9 @@ trap(void);
#define GF_THREAD_NAME_LIMIT 16
#define GF_THREAD_NAME_PREFIX "glfs_"
+/* Advisory buffer size for formatted timestamps (see gf_time_fmt) */
+#define GF_TIMESTR_SIZE 256
+
/*
* we could have initialized these as +ve values and treated
* them as negative while comparing etc.. (which would have
@@ -180,6 +189,12 @@ enum _gf_xlator_ipc_targets {
typedef enum _gf_special_pid gf_special_pid_t;
typedef enum _gf_xlator_ipc_targets _gf_xlator_ipc_targets_t;
+/* Array to hold custom xattr keys */
+extern char *xattrs_to_heal[];
+
+char **
+get_xattrs_to_heal();
+
/* The DHT file rename operation is not a straightforward rename.
* It involves creating linkto and linkfiles, and can unlink or rename the
* source file depending on the hashed and cached subvols for the source
@@ -240,6 +255,8 @@ list_node_del(struct list_node *node);
struct dnscache *
gf_dnscache_init(time_t ttl);
+void
+gf_dnscache_deinit(struct dnscache *cache);
struct dnscache_entry *
gf_dnscache_entry_init(void);
void
@@ -438,6 +455,15 @@ BIT_VALUE(unsigned char *array, unsigned int index)
} while (0)
#endif
+/* Compile-time assert, borrowed from Linux kernel. */
+#ifdef HAVE_STATIC_ASSERT
+#define GF_STATIC_ASSERT(expr, ...) \
+ __gf_static_assert(expr, ##__VA_ARGS__, #expr)
+#define __gf_static_assert(expr, msg, ...) _Static_assert(expr, msg)
+#else
+#define GF_STATIC_ASSERT(expr, ...)
+#endif
+
#define GF_ABORT(msg...) \
do { \
gf_msg_callingfn("", GF_LOG_CRITICAL, 0, LG_MSG_ASSERTION_FAILED, \
@@ -468,18 +494,15 @@ union gf_sock_union {
#define IOV_MIN(n) min(IOV_MAX, n)
-#define GF_SKIP_IRRELEVANT_ENTRIES(entry, dir, scr) \
- do { \
- entry = NULL; \
- if (dir) { \
- entry = sys_readdir(dir, scr); \
- while (entry && (!strcmp(entry->d_name, ".") || \
- !fnmatch("*.tmp", entry->d_name, 0) || \
- !strcmp(entry->d_name, ".."))) { \
- entry = sys_readdir(dir, scr); \
- } \
- } \
- } while (0)
+static inline gf_boolean_t
+gf_irrelevant_entry(struct dirent *entry)
+{
+ GF_ASSERT(entry);
+
+ return (!strcmp(entry->d_name, ".") ||
+ !fnmatch("*.tmp", entry->d_name, 0) ||
+ !strcmp(entry->d_name, ".."));
+}
static inline void
iov_free(struct iovec *vector, int count)
@@ -790,7 +813,7 @@ typedef enum {
} gf_timefmts;
static inline char *
-gf_time_fmt(char *dst, size_t sz_dst, time_t utime, unsigned int fmt)
+gf_time_fmt_tv(char *dst, size_t sz_dst, struct timeval *tv, unsigned int fmt)
{
extern void _gf_timestuff(const char ***, const char ***);
static gf_timefmts timefmt_last = (gf_timefmts)-1;
@@ -798,6 +821,8 @@ gf_time_fmt(char *dst, size_t sz_dst, time_t utime, unsigned int fmt)
static const char **zeros;
struct tm tm, *res;
int localtime = 0;
+ int len = 0;
+ int pos = 0;
if (timefmt_last == ((gf_timefmts)-1)) {
_gf_timestuff(&fmts, &zeros);
@@ -807,15 +832,35 @@ gf_time_fmt(char *dst, size_t sz_dst, time_t utime, unsigned int fmt)
fmt = gf_timefmt_default;
}
localtime = gf_log_get_localtime();
- res = localtime ? localtime_r(&utime, &tm) : gmtime_r(&utime, &tm);
- if (utime && (res != NULL)) {
- strftime(dst, sz_dst, fmts[fmt], &tm);
+ res = localtime ? localtime_r(&tv->tv_sec, &tm)
+ : gmtime_r(&tv->tv_sec, &tm);
+ if (tv->tv_sec && (res != NULL)) {
+ len = strftime(dst, sz_dst, fmts[fmt], &tm);
+ if (len == 0)
+ return dst;
+ pos += len;
+ if (tv->tv_usec >= 0) {
+ len = snprintf(dst + pos, sz_dst - pos, ".%" GF_PRI_SUSECONDS,
+ tv->tv_usec);
+ if (len >= sz_dst - pos)
+ return dst;
+ pos += len;
+ }
+ strftime(dst + pos, sz_dst - pos, " %z", &tm);
} else {
strncpy(dst, "N/A", sz_dst);
}
return dst;
}
+static inline char *
+gf_time_fmt(char *dst, size_t sz_dst, time_t utime, unsigned int fmt)
+{
+ struct timeval tv = {utime, -1};
+
+ return gf_time_fmt_tv(dst, sz_dst, &tv, fmt);
+}
+
/* This function helps us use gfid (unique identity) to generate inode's unique
* number in glusterfs.
*/
@@ -1165,6 +1210,47 @@ int
gf_d_type_from_ia_type(ia_type_t type);
int
+gf_syncfs(int fd);
+
+int
gf_nanosleep(uint64_t nsec);
+static inline time_t
+gf_time(void)
+{
+ return time(NULL);
+}
+
+/* Return delta value in microseconds. */
+
+static inline double
+gf_tvdiff(struct timeval *start, struct timeval *end)
+{
+ struct timeval t;
+
+ if (start->tv_usec > end->tv_usec)
+ t.tv_sec = end->tv_sec - 1, t.tv_usec = end->tv_usec + 1000000;
+ else
+ t.tv_sec = end->tv_sec, t.tv_usec = end->tv_usec;
+
+ return (double)(t.tv_sec - start->tv_sec) * 1e6 +
+ (double)(t.tv_usec - start->tv_usec);
+}
+
+/* Return delta value in nanoseconds. */
+
+static inline double
+gf_tsdiff(struct timespec *start, struct timespec *end)
+{
+ struct timespec t;
+
+ if (start->tv_nsec > end->tv_nsec)
+ t.tv_sec = end->tv_sec - 1, t.tv_nsec = end->tv_nsec + 1000000000;
+ else
+ t.tv_sec = end->tv_sec, t.tv_nsec = end->tv_nsec;
+
+ return (double)(t.tv_sec - start->tv_sec) * 1e9 +
+ (double)(t.tv_nsec - start->tv_nsec);
+}
+
#endif /* _COMMON_UTILS_H */
diff --git a/libglusterfs/src/glusterfs/dict.h b/libglusterfs/src/glusterfs/dict.h
index ce1c2100276..d0467c6dfb6 100644
--- a/libglusterfs/src/glusterfs/dict.h
+++ b/libglusterfs/src/glusterfs/dict.h
@@ -25,9 +25,6 @@ typedef struct _data_pair data_pair_t;
#define dict_add_sizen(this, key, value) dict_addn(this, key, SLEN(key), value)
-#define dict_get_with_ref_sizen(this, key, value) \
- dict_get_with_refn(this, key, SLEN(key), value)
-
#define dict_get_sizen(this, key) dict_getn(this, key, SLEN(key))
#define dict_del_sizen(this, key) dict_deln(this, key, SLEN(key))
@@ -98,7 +95,7 @@ struct _data {
char *data;
gf_atomic_t refcount;
gf_dict_data_type_t data_type;
- int32_t len;
+ uint32_t len;
gf_boolean_t is_static;
};
@@ -122,6 +119,8 @@ struct _dict {
gf_lock_t lock;
data_pair_t *members_internal;
data_pair_t free_pair;
+ /* Variable to store total keylen + value->len */
+ uint32_t totkvlen;
};
typedef gf_boolean_t (*dict_match_t)(dict_t *d, char *k, data_t *v, void *data);
@@ -136,6 +135,7 @@ int32_t
dict_set(dict_t *this, char *key, data_t *value);
int32_t
dict_setn(dict_t *this, char *key, const int keylen, data_t *value);
+
/* function to set a new key/value pair (without checking for duplicate) */
int32_t
dict_add(dict_t *this, char *key, data_t *value);
@@ -143,8 +143,6 @@ int32_t
dict_addn(dict_t *this, char *key, const int keylen, data_t *value);
int
dict_get_with_ref(dict_t *this, char *key, data_t **data);
-int
-dict_get_with_refn(dict_t *this, char *key, const int keylen, data_t **data);
data_t *
dict_get(dict_t *this, char *key);
data_t *
diff --git a/libglusterfs/src/glusterfs/gf-event.h b/libglusterfs/src/glusterfs/gf-event.h
index c0f05e7c83b..40f8fbdf10a 100644
--- a/libglusterfs/src/glusterfs/gf-event.h
+++ b/libglusterfs/src/glusterfs/gf-event.h
@@ -12,6 +12,7 @@
#define _GF_EVENT_H_
#include <pthread.h>
+#include "common-utils.h"
#include "list.h"
struct event_pool;
@@ -31,6 +32,9 @@ typedef void (*event_handler_t)(int fd, int idx, int gen, void *data,
#define EVENT_EPOLL_SLOTS 1024
#define EVENT_MAX_THREADS 1024
+/* See rpcsvc.h to check why. */
+GF_STATIC_ASSERT(EVENT_MAX_THREADS % __BITS_PER_LONG == 0);
+
struct event_pool {
struct event_ops *ops;
diff --git a/libglusterfs/src/glusterfs/globals.h b/libglusterfs/src/glusterfs/globals.h
index e50ee3046af..b22eaae6c2f 100644
--- a/libglusterfs/src/glusterfs/globals.h
+++ b/libglusterfs/src/glusterfs/globals.h
@@ -45,7 +45,7 @@
1 /* MIN is the fresh start op-version, mostly \
should not change */
#define GD_OP_VERSION_MAX \
- GD_OP_VERSION_8_0 /* MAX VERSION is the maximum \
+ GD_OP_VERSION_9_0 /* MAX VERSION is the maximum \
count in VME table, should \
keep changing with \
introduction of newer \
@@ -122,6 +122,8 @@
#define GD_OP_VERSION_8_0 80000 /* Op-version for GlusterFS 8.0 */
+#define GD_OP_VERSION_9_0 90000 /* Op-version for GlusterFS 9.0 */
+
#define GD_OP_VER_PERSISTENT_AFR_XATTRS GD_OP_VERSION_3_6_0
#include "glusterfs/xlator.h"
diff --git a/libglusterfs/src/glusterfs/glusterfs-acl.h b/libglusterfs/src/glusterfs/glusterfs-acl.h
index cae55e8062f..987bf5fab0b 100644
--- a/libglusterfs/src/glusterfs/glusterfs-acl.h
+++ b/libglusterfs/src/glusterfs/glusterfs-acl.h
@@ -143,7 +143,7 @@ gf_posix_acl_get_key(const acl_type_t type)
return acl_key;
}
-static inline const acl_type_t
+static inline acl_type_t
gf_posix_acl_get_type(const char *key)
{
acl_type_t type = 0;
diff --git a/libglusterfs/src/glusterfs/glusterfs.h b/libglusterfs/src/glusterfs/glusterfs.h
index a54a4ed42a7..e6425618b7f 100644
--- a/libglusterfs/src/glusterfs/glusterfs.h
+++ b/libglusterfs/src/glusterfs/glusterfs.h
@@ -43,6 +43,9 @@
#define GF_YES 1
#define GF_NO 0
+#define IS_ERROR(ret) ((ret) < 0)
+#define IS_SUCCESS(ret) ((ret) >= 0)
+
#ifndef O_LARGEFILE
/* savannah bug #20053, patch for compiling on darwin */
#define O_LARGEFILE 0100000 /* from bits/fcntl.h */
@@ -422,7 +425,7 @@ static const char *const FOP_PRI_STRINGS[] = {"HIGH", "NORMAL", "LOW", "LEAST"};
static inline const char *
fop_pri_to_string(gf_fop_pri_t pri)
{
- if (pri < 0)
+ if (IS_ERROR(pri))
return "UNSPEC";
if (pri >= GF_FOP_PRI_MAX)
@@ -463,6 +466,8 @@ typedef struct _server_cmdline server_cmdline_t;
#define GF_OPTION_DISABLE _gf_false
#define GF_OPTION_DEFERRED 2
+typedef enum { _gf_none, _gf_memcheck, _gf_drd } gf_valgrind_tool;
+
struct _cmd_args {
/* basic options */
char *volfile_server;
@@ -555,7 +560,8 @@ struct _cmd_args {
/* Run this process with valgrind? Might want to prevent calling
* functions that prevent valgrind from working correctly, like
* dlclose(). */
- int valgrind;
+ gf_valgrind_tool vgtool;
+
int localtime_logging;
/* For the subdir mount */
diff --git a/libglusterfs/src/glusterfs/latency.h b/libglusterfs/src/glusterfs/latency.h
index ed47b1f0cbc..4d601bbcbd6 100644
--- a/libglusterfs/src/glusterfs/latency.h
+++ b/libglusterfs/src/glusterfs/latency.h
@@ -11,13 +11,23 @@
#ifndef __LATENCY_H__
#define __LATENCY_H__
-#include "glusterfs/glusterfs.h"
+#include <inttypes.h>
+#include <time.h>
-typedef struct fop_latency {
- double min; /* min time for the call (microseconds) */
- double max; /* max time for the call (microseconds) */
- double total; /* total time (microseconds) */
+typedef struct _gf_latency {
+ uint64_t min; /* min time for the call (nanoseconds) */
+ uint64_t max; /* max time for the call (nanoseconds) */
+ uint64_t total; /* total time (nanoseconds) */
uint64_t count;
-} fop_latency_t;
+} gf_latency_t;
+gf_latency_t *
+gf_latency_new(size_t n);
+
+void
+gf_latency_reset(gf_latency_t *lat);
+
+void
+gf_latency_update(gf_latency_t *lat, struct timespec *begin,
+ struct timespec *end);
#endif /* __LATENCY_H__ */
diff --git a/libglusterfs/src/glusterfs/mem-pool.h b/libglusterfs/src/glusterfs/mem-pool.h
index 90fb8820c74..e5b3276d047 100644
--- a/libglusterfs/src/glusterfs/mem-pool.h
+++ b/libglusterfs/src/glusterfs/mem-pool.h
@@ -202,6 +202,24 @@ out:
return dup_mem;
}
+#ifdef GF_DISABLE_MEMPOOL
+
+/* No-op memory pool enough to fit current API without massive redesign. */
+
+struct mem_pool {
+ unsigned long sizeof_type;
+};
+
+#define mem_pools_init() \
+ do { \
+ } while (0)
+#define mem_pools_fini() \
+ do { \
+ } while (0)
+#define mem_pool_thread_destructor(pool_list) (void)pool_list
+
+#else /* !GF_DISABLE_MEMPOOL */
+
/* kind of 'header' for the actual mem_pool_shared structure, this might make
* it possible to dump some more details in a statedump */
struct mem_pool {
@@ -209,10 +227,11 @@ struct mem_pool {
unsigned long sizeof_type;
unsigned long count; /* requested pool size (unused) */
char *name;
- gf_atomic_t active; /* current allocations */
+ char *xl_name;
+ gf_atomic_t active; /* current allocations */
#ifdef DEBUG
- gf_atomic_t hit; /* number of allocations served from pt_pool */
- gf_atomic_t miss; /* number of std allocs due to miss */
+ gf_atomic_t hit; /* number of allocations served from pt_pool */
+ gf_atomic_t miss; /* number of std allocs due to miss */
#endif
struct list_head owner; /* glusterfs_ctx_t->mempool_list */
glusterfs_ctx_t *ctx; /* take ctx->lock when updating owner */
@@ -286,6 +305,10 @@ void
mem_pools_init(void); /* start the pool_sweeper thread */
void
mem_pools_fini(void); /* cleanup memory pools */
+void
+mem_pool_thread_destructor(per_thread_pool_list_t *pool_list);
+
+#endif /* GF_DISABLE_MEMPOOL */
struct mem_pool *
mem_pool_new_fn(glusterfs_ctx_t *ctx, unsigned long sizeof_type,
@@ -308,9 +331,6 @@ void
mem_pool_destroy(struct mem_pool *pool);
void
-mem_pool_thread_destructor(per_thread_pool_list_t *pool_list);
-
-void
gf_mem_acct_enable_set(void *ctx);
#endif /* _MEM_POOL_H */
diff --git a/libglusterfs/src/glusterfs/mem-types.h b/libglusterfs/src/glusterfs/mem-types.h
index 36cf7820ad5..d45d5b68c91 100644
--- a/libglusterfs/src/glusterfs/mem-types.h
+++ b/libglusterfs/src/glusterfs/mem-types.h
@@ -133,6 +133,7 @@ enum gf_common_mem_types_ {
gf_common_volfile_t,
gf_common_mt_mgmt_v3_lock_timer_t, /* used only in one location */
gf_common_mt_server_cmdline_t, /* used only in one location */
+ gf_common_mt_latency_t,
gf_common_mt_end
};
#endif
diff --git a/libglusterfs/src/glusterfs/stack.h b/libglusterfs/src/glusterfs/stack.h
index 17585508a22..536a330d38b 100644
--- a/libglusterfs/src/glusterfs/stack.h
+++ b/libglusterfs/src/glusterfs/stack.h
@@ -45,6 +45,9 @@ typedef int32_t (*ret_fn_t)(call_frame_t *frame, call_frame_t *prev_frame,
xlator_t *this, int32_t op_ret, int32_t op_errno,
...);
+void
+gf_frame_latency_update(call_frame_t *frame);
+
struct call_pool {
union {
struct list_head all_frames;
@@ -149,8 +152,6 @@ struct _call_stack {
} while (0);
struct xlator_fops;
-void
-gf_update_latency(call_frame_t *frame);
static inline void
FRAME_DESTROY(call_frame_t *frame)
@@ -158,7 +159,7 @@ FRAME_DESTROY(call_frame_t *frame)
void *local = NULL;
if (frame->root->ctx->measure_latency)
- gf_update_latency(frame);
+ gf_frame_latency_update(frame);
list_del_init(&frame->frames);
if (frame->local) {
@@ -429,6 +430,7 @@ call_stack_alloc_groups(call_stack_t *stack, int ngrps)
if (ngrps <= SMALL_GROUP_COUNT) {
stack->groups = stack->groups_small;
} else {
+ GF_FREE(stack->groups_large);
stack->groups_large = GF_CALLOC(ngrps, sizeof(gid_t),
gf_common_mt_groups_t);
if (!stack->groups_large)
@@ -442,6 +444,12 @@ call_stack_alloc_groups(call_stack_t *stack, int ngrps)
}
static inline int
+call_stack_groups_capacity(call_stack_t *stack)
+{
+ return max(stack->ngrps, SMALL_GROUP_COUNT);
+}
+
+static inline int
call_frames_count(call_stack_t *call_stack)
{
call_frame_t *pos;
diff --git a/libglusterfs/src/glusterfs/statedump.h b/libglusterfs/src/glusterfs/statedump.h
index 89d04f94587..ce082706bdf 100644
--- a/libglusterfs/src/glusterfs/statedump.h
+++ b/libglusterfs/src/glusterfs/statedump.h
@@ -127,4 +127,6 @@ gf_proc_dump_xlator_meminfo(xlator_t *this, strfd_t *strfd);
void
gf_proc_dump_xlator_profile(xlator_t *this, strfd_t *strfd);
+void
+gf_latency_statedump_and_reset(char *key, gf_latency_t *lat);
#endif /* STATEDUMP_H */
diff --git a/libglusterfs/src/glusterfs/store.h b/libglusterfs/src/glusterfs/store.h
index c8be544e164..a1f70c7b840 100644
--- a/libglusterfs/src/glusterfs/store.h
+++ b/libglusterfs/src/glusterfs/store.h
@@ -95,7 +95,7 @@ int32_t
gf_store_iter_get_matching(gf_store_iter_t *iter, char *key, char **value);
int32_t
-gf_store_iter_destroy(gf_store_iter_t *iter);
+gf_store_iter_destroy(gf_store_iter_t **iter);
char *
gf_store_strerror(gf_store_op_errno_t op_errno);
diff --git a/libglusterfs/src/glusterfs/syncop.h b/libglusterfs/src/glusterfs/syncop.h
index 1a801a91444..4e9241a32fc 100644
--- a/libglusterfs/src/glusterfs/syncop.h
+++ b/libglusterfs/src/glusterfs/syncop.h
@@ -31,6 +31,11 @@
#define SYNCOPCTX_PID 0x00000008
#define SYNCOPCTX_LKOWNER 0x00000010
+#ifdef HAVE_TSAN_API
+/* Currently hardcoded within thread context maintained by the sanitizer. */
+#define TSAN_THREAD_NAMELEN 64
+#endif
+
struct synctask;
struct syncproc;
struct syncenv;
@@ -71,6 +76,13 @@ struct synctask {
uid_t uid;
gid_t gid;
+#ifdef HAVE_TSAN_API
+ struct {
+ void *fiber;
+ char name[TSAN_THREAD_NAMELEN];
+ } tsan;
+#endif
+
ucontext_t ctx;
struct syncproc *proc;
@@ -83,6 +95,14 @@ struct synctask {
struct syncproc {
pthread_t processor;
+
+#ifdef HAVE_TSAN_API
+ struct {
+ void *fiber;
+ char name[TSAN_THREAD_NAMELEN];
+ } tsan;
+#endif
+
ucontext_t sched;
struct syncenv *env;
struct synctask *current;
diff --git a/libglusterfs/src/glusterfs/syscall.h b/libglusterfs/src/glusterfs/syscall.h
index 91e921aea50..b6d3ab4f2ad 100644
--- a/libglusterfs/src/glusterfs/syscall.h
+++ b/libglusterfs/src/glusterfs/syscall.h
@@ -266,4 +266,13 @@ ssize_t
sys_copy_file_range(int fd_in, off64_t *off_in, int fd_out, off64_t *off_out,
size_t len, unsigned int flags);
+int
+sys_kill(pid_t pid, int sig);
+
+#ifdef __FreeBSD__
+int
+sys_sysctl(const int *name, u_int namelen, void *oldp, size_t *oldlenp,
+ const void *newp, size_t newlen);
+#endif
+
#endif /* __SYSCALL_H__ */
diff --git a/libglusterfs/src/glusterfs/xlator.h b/libglusterfs/src/glusterfs/xlator.h
index 23004ab2245..4fd3abdaeff 100644
--- a/libglusterfs/src/glusterfs/xlator.h
+++ b/libglusterfs/src/glusterfs/xlator.h
@@ -805,7 +805,7 @@ struct _xlator {
struct {
/* for latency measurement */
- fop_latency_t latencies[GF_FOP_MAXVALUE];
+ gf_latency_t latencies[GF_FOP_MAXVALUE];
/* for latency measurement */
fop_metrics_t metrics[GF_FOP_MAXVALUE];
diff --git a/libglusterfs/src/graph.c b/libglusterfs/src/graph.c
index 64e77e095a2..13f298eb3bd 100644
--- a/libglusterfs/src/graph.c
+++ b/libglusterfs/src/graph.c
@@ -41,7 +41,7 @@ _gf_dump_details (int argc, char **argv)
{
extern FILE *gf_log_logfile;
int i = 0;
- char timestr[64];
+ char timestr[GF_TIMESTR_SIZE];
time_t utime = 0;
pid_t mypid = 0;
struct utsname uname_buf = {{0, }, };
@@ -482,7 +482,7 @@ fill_uuid(char *uuid, int size, struct timeval tv)
char hostname[50] = {
0,
};
- char now_str[64];
+ char now_str[GF_TIMESTR_SIZE];
if (gethostname(hostname, sizeof(hostname) - 1) != 0) {
gf_msg("graph", GF_LOG_ERROR, errno, LG_MSG_GETHOSTNAME_FAILED,
@@ -490,9 +490,8 @@ fill_uuid(char *uuid, int size, struct timeval tv)
hostname[sizeof(hostname) - 1] = '\0';
}
- gf_time_fmt(now_str, sizeof now_str, tv.tv_sec, gf_timefmt_dirent);
- snprintf(uuid, size, "%s-%d-%s:%" GF_PRI_SUSECONDS, hostname, getpid(),
- now_str, tv.tv_usec);
+ gf_time_fmt_tv(now_str, sizeof now_str, &tv, gf_timefmt_dirent);
+ snprintf(uuid, size, "%s-%d-%s", hostname, getpid(), now_str);
return;
}
diff --git a/libglusterfs/src/latency.c b/libglusterfs/src/latency.c
index 15b397c3799..ce4b0e8255d 100644
--- a/libglusterfs/src/latency.c
+++ b/libglusterfs/src/latency.c
@@ -16,35 +16,32 @@
#include "glusterfs/glusterfs.h"
#include "glusterfs/statedump.h"
-void
-gf_update_latency(call_frame_t *frame)
+gf_latency_t *
+gf_latency_new(size_t n)
{
- double elapsed;
- struct timespec *begin, *end;
-
- fop_latency_t *lat;
-
- begin = &frame->begin;
- end = &frame->end;
+ int i = 0;
+ gf_latency_t *lat = NULL;
- if (!(begin->tv_sec && end->tv_sec))
- goto out;
+ lat = GF_MALLOC(n * sizeof(*lat), gf_common_mt_latency_t);
+ if (!lat)
+ return NULL;
- elapsed = (end->tv_sec - begin->tv_sec) * 1e9 +
- (end->tv_nsec - begin->tv_nsec);
+ for (i = 0; i < n; i++) {
+ gf_latency_reset(lat + i);
+ }
+ return lat;
+}
- if (frame->op < 0 || frame->op >= GF_FOP_MAXVALUE) {
- gf_log("[core]", GF_LOG_WARNING, "Invalid frame op value: %d",
- frame->op);
+void
+gf_latency_update(gf_latency_t *lat, struct timespec *begin,
+ struct timespec *end)
+{
+ if (!(begin->tv_sec && end->tv_sec)) {
+ /*Measure latency might have been enabled/disabled during the op*/
return;
}
- /* Can happen mostly at initiator xlator, as STACK_WIND/UNWIND macros
- set it right anyways for those frames */
- if (!frame->op)
- frame->op = frame->root->op;
-
- lat = &frame->this->stats.interval.latencies[frame->op];
+ double elapsed = gf_tsdiff(begin, end);
if (lat->max < elapsed)
lat->max = elapsed;
@@ -54,42 +51,34 @@ gf_update_latency(call_frame_t *frame)
lat->total += elapsed;
lat->count++;
-out:
- return;
}
void
-gf_proc_dump_latency_info(xlator_t *xl)
+gf_latency_reset(gf_latency_t *lat)
{
- char key_prefix[GF_DUMP_MAX_BUF_LEN];
- char key[GF_DUMP_MAX_BUF_LEN];
- int i;
-
- snprintf(key_prefix, GF_DUMP_MAX_BUF_LEN, "%s.latency", xl->name);
- gf_proc_dump_add_section("%s", key_prefix);
-
- for (i = 0; i < GF_FOP_MAXVALUE; i++) {
- gf_proc_dump_build_key(key, key_prefix, "%s", (char *)gf_fop_list[i]);
-
- fop_latency_t *lat = &xl->stats.interval.latencies[i];
+ if (!lat)
+ return;
+ memset(lat, 0, sizeof(*lat));
+ lat->min = ULLONG_MAX;
+ /* make sure 'min' is set to high value, so it would be
+ properly set later */
+}
- /* Doesn't make sense to continue if there are no fops
- came in the given interval */
- if (!lat->count)
- continue;
+void
+gf_frame_latency_update(call_frame_t *frame)
+{
+ gf_latency_t *lat;
+ /* Can happen mostly at initiator xlator, as STACK_WIND/UNWIND macros
+ set it right anyways for those frames */
+ if (!frame->op)
+ frame->op = frame->root->op;
- gf_proc_dump_write(
- key, "AVG:%.03f,CNT:%" PRId64 ",TOTAL:%.03f,MIN:%.03f,MAX:%.03f",
- (lat->total / lat->count), lat->count, lat->total, lat->min,
- lat->max);
+ if (frame->op < 0 || frame->op >= GF_FOP_MAXVALUE) {
+ gf_log("[core]", GF_LOG_WARNING, "Invalid frame op value: %d",
+ frame->op);
+ return;
}
- memset(xl->stats.interval.latencies, 0,
- sizeof(xl->stats.interval.latencies));
-
- /* make sure 'min' is set to high value, so it would be
- properly set later */
- for (i = 0; i < GF_FOP_MAXVALUE; i++) {
- xl->stats.interval.latencies[i].min = 0xffffffff;
- }
+ lat = &frame->this->stats.interval.latencies[frame->op];
+ gf_latency_update(lat, &frame->begin, &frame->end);
}
diff --git a/libglusterfs/src/libglusterfs.sym b/libglusterfs/src/libglusterfs.sym
index c80b892df25..5f18cd56cbe 100644
--- a/libglusterfs/src/libglusterfs.sym
+++ b/libglusterfs/src/libglusterfs.sym
@@ -585,6 +585,7 @@ gf_dirent_free
gf_dirent_orig_offset
gf_dm_hashfn
gf_dnscache_init
+gf_dnscache_deinit
gf_errno_to_error
gf_error_to_errno
_gf_event
@@ -1080,6 +1081,8 @@ sys_write
sys_writev
sys_socket
sys_accept
+sys_kill
+sys_sysctl
tbf_init
tbf_throttle
timespec_now
@@ -1180,4 +1183,11 @@ glusterfs_process_svc_detach
mgmt_is_multiplexed_daemon
xlator_is_cleanup_starting
gf_nanosleep
+gf_syncfs
graph_total_client_xlator
+get_xattrs_to_heal
+gf_latency_statedump_and_reset
+gf_latency_new
+gf_latency_reset
+gf_latency_update
+gf_frame_latency_update
diff --git a/libglusterfs/src/logging.c b/libglusterfs/src/logging.c
index 3d7d31772df..a930d3e3b63 100644
--- a/libglusterfs/src/logging.c
+++ b/libglusterfs/src/logging.c
@@ -35,7 +35,6 @@
#define GF_LOG_CONTROL_FILE "/etc/glusterfs/logger.conf"
#define GF_LOG_BACKTRACE_DEPTH 5
#define GF_LOG_BACKTRACE_SIZE 4096
-#define GF_LOG_TIMESTR_SIZE 256
#define GF_MAX_SLOG_PAIR_COUNT 100
#include "glusterfs/logging.h"
@@ -760,7 +759,7 @@ _gf_log_callingfn(const char *domain, const char *file, const char *function,
xlator_t *this = THIS;
char *logline = NULL;
char *msg = NULL;
- char timestr[GF_LOG_TIMESTR_SIZE] = {
+ char timestr[GF_TIMESTR_SIZE] = {
0,
};
char *callstr = NULL;
@@ -820,12 +819,12 @@ _gf_log_callingfn(const char *domain, const char *file, const char *function,
if (-1 == ret)
goto out;
- gf_time_fmt(timestr, sizeof timestr, tv.tv_sec, gf_timefmt_FT);
+ gf_time_fmt_tv(timestr, sizeof timestr, &tv, gf_timefmt_FT);
- ret = gf_asprintf(
- &logline, "[%s.%" GF_PRI_SUSECONDS "] %c [%s:%d:%s] %s %d-%s: %s\n",
- timestr, tv.tv_usec, gf_level_strings[level], basename, line, function,
- callstr, ((this->graph) ? this->graph->id : 0), domain, msg);
+ ret = gf_asprintf(&logline, "[%s] %c [%s:%d:%s] %s %d-%s: %s\n", timestr,
+ gf_level_strings[level], basename, line, function,
+ callstr, ((this->graph) ? this->graph->id : 0), domain,
+ msg);
if (-1 == ret) {
goto out;
}
@@ -1088,7 +1087,7 @@ _gf_msg_nomem(const char *domain, const char *file, const char *function,
char msg[2048] = {
0,
};
- char timestr[GF_LOG_TIMESTR_SIZE] = {
+ char timestr[GF_TIMESTR_SIZE] = {
0,
};
glusterfs_ctx_t *ctx = NULL;
@@ -1116,20 +1115,20 @@ _gf_msg_nomem(const char *domain, const char *file, const char *function,
ret = gettimeofday(&tv, NULL);
if (-1 == ret)
goto out;
- gf_time_fmt(timestr, sizeof timestr, tv.tv_sec, gf_timefmt_FT);
+ gf_time_fmt_tv(timestr, sizeof timestr, &tv, gf_timefmt_FT);
/* TODO: Currently we print in the enhanced format, with a message ID
* of 0. Need to enhance this to support format as configured */
wlen = snprintf(
msg, sizeof msg,
- "[%s.%" GF_PRI_SUSECONDS "] %c [MSGID: %" PRIu64
+ "[%s] %c [MSGID: %" PRIu64
"]"
" [%s:%d:%s] %s: no memory "
"available for size (%" GF_PRI_SIZET
") current memory usage in kilobytes %ld"
" [call stack follows]\n",
- timestr, tv.tv_usec, gf_level_strings[level], (uint64_t)0, basename,
- line, function, domain, size,
+ timestr, gf_level_strings[level], (uint64_t)0, basename, line, function,
+ domain, size,
(!getrusage(RUSAGE_SELF, &r_usage) ? r_usage.ru_maxrss : 0));
if (-1 == wlen) {
ret = -1;
@@ -1275,7 +1274,7 @@ gf_log_glusterlog(glusterfs_ctx_t *ctx, const char *domain, const char *file,
int errnum, uint64_t msgid, char **appmsgstr, char *callstr,
struct timeval tv, int graph_id, gf_log_format_t fmt)
{
- char timestr[GF_LOG_TIMESTR_SIZE] = {
+ char timestr[GF_TIMESTR_SIZE] = {
0,
};
char *header = NULL;
@@ -1286,7 +1285,7 @@ gf_log_glusterlog(glusterfs_ctx_t *ctx, const char *domain, const char *file,
gf_log_rotate(ctx);
/* format the time stamp */
- gf_time_fmt(timestr, sizeof timestr, tv.tv_sec, gf_timefmt_FT);
+ gf_time_fmt_tv(timestr, sizeof timestr, &tv, gf_timefmt_FT);
/* generate footer */
if (errnum) {
@@ -1302,40 +1301,35 @@ gf_log_glusterlog(glusterfs_ctx_t *ctx, const char *domain, const char *file,
if (fmt == gf_logformat_traditional) {
if (!callstr) {
ret = gf_asprintf(&header,
- "[%s.%" GF_PRI_SUSECONDS
- "] %c [%s:%d:%s]"
+ "[%s] %c [%s:%d:%s]"
" %d-%s: %s",
- timestr, tv.tv_usec, gf_level_strings[level],
- file, line, function, graph_id, domain,
- *appmsgstr);
+ timestr, gf_level_strings[level], file, line,
+ function, graph_id, domain, *appmsgstr);
} else {
ret = gf_asprintf(&header,
- "[%s.%" GF_PRI_SUSECONDS
- "] %c [%s:%d:%s] %s"
+ "[%s] %c [%s:%d:%s] %s"
" %d-%s: %s",
- timestr, tv.tv_usec, gf_level_strings[level],
- file, line, function, callstr, graph_id, domain,
- *appmsgstr);
+ timestr, gf_level_strings[level], file, line,
+ function, callstr, graph_id, domain, *appmsgstr);
}
} else { /* gf_logformat_withmsgid */
/* CEE log format unsupported in logger_glusterlog, so just
* print enhanced log format */
if (!callstr) {
ret = gf_asprintf(&header,
- "[%s.%" GF_PRI_SUSECONDS "] %c [MSGID: %" PRIu64
+ "[%s] %c [MSGID: %" PRIu64
"]"
" [%s:%d:%s] %d-%s: %s",
- timestr, tv.tv_usec, gf_level_strings[level],
- msgid, file, line, function, graph_id, domain,
- *appmsgstr);
+ timestr, gf_level_strings[level], msgid, file,
+ line, function, graph_id, domain, *appmsgstr);
} else {
ret = gf_asprintf(&header,
- "[%s.%" GF_PRI_SUSECONDS "] %c [MSGID: %" PRIu64
+ "[%s] %c [MSGID: %" PRIu64
"]"
" [%s:%d:%s] %s %d-%s: %s",
- timestr, tv.tv_usec, gf_level_strings[level],
- msgid, file, line, function, callstr, graph_id,
- domain, *appmsgstr);
+ timestr, gf_level_strings[level], msgid, file,
+ line, function, callstr, graph_id, domain,
+ *appmsgstr);
}
}
if (-1 == ret) {
@@ -1386,39 +1380,36 @@ gf_syslog_log_repetitions(const char *domain, const char *file,
int graph_id)
{
int priority;
- char timestr_latest[GF_LOG_TIMESTR_SIZE] = {
+ char timestr_latest[GF_TIMESTR_SIZE] = {
0,
};
- char timestr_oldest[GF_LOG_TIMESTR_SIZE] = {
+ char timestr_oldest[GF_TIMESTR_SIZE] = {
0,
};
SET_LOG_PRIO(level, priority);
- gf_time_fmt(timestr_latest, sizeof timestr_latest, latest.tv_sec,
- gf_timefmt_FT);
- gf_time_fmt(timestr_oldest, sizeof timestr_oldest, oldest.tv_sec,
- gf_timefmt_FT);
+ gf_time_fmt_tv(timestr_latest, sizeof timestr_latest, &latest,
+ gf_timefmt_FT);
+ gf_time_fmt_tv(timestr_oldest, sizeof timestr_oldest, &oldest,
+ gf_timefmt_FT);
if (errnum) {
- syslog(
- priority,
- "The message \"[MSGID: %" PRIu64
- "] [%s:%d:%s] "
- "%d-%s: %s [%s] \" repeated %d times between %s.%" GF_PRI_SUSECONDS
- " and %s.%" GF_PRI_SUSECONDS,
- msgid, file, line, function, graph_id, domain, *appmsgstr,
- strerror(errnum), refcount, timestr_oldest, oldest.tv_usec,
- timestr_latest, latest.tv_usec);
+ syslog(priority,
+ "The message \"[MSGID: %" PRIu64
+ "] [%s:%d:%s] "
+ "%d-%s: %s [%s] \" repeated %d times between %s"
+ " and %s",
+ msgid, file, line, function, graph_id, domain, *appmsgstr,
+ strerror(errnum), refcount, timestr_oldest, timestr_latest);
} else {
syslog(priority,
"The message \"[MSGID: %" PRIu64
"] [%s:%d:%s] "
- "%d-%s: %s \" repeated %d times between %s.%" GF_PRI_SUSECONDS
- " and %s.%" GF_PRI_SUSECONDS,
+ "%d-%s: %s \" repeated %d times between %s"
+ " and %s",
msgid, file, line, function, graph_id, domain, *appmsgstr,
- refcount, timestr_oldest, oldest.tv_usec, timestr_latest,
- latest.tv_usec);
+ refcount, timestr_oldest, timestr_latest);
}
return 0;
}
@@ -1432,10 +1423,10 @@ gf_glusterlog_log_repetitions(glusterfs_ctx_t *ctx, const char *domain,
struct timeval latest, int graph_id)
{
int ret = 0;
- char timestr_latest[GF_LOG_TIMESTR_SIZE] = {
+ char timestr_latest[GF_TIMESTR_SIZE] = {
0,
};
- char timestr_oldest[GF_LOG_TIMESTR_SIZE] = {
+ char timestr_oldest[GF_TIMESTR_SIZE] = {
0,
};
char errstr[256] = {
@@ -1459,21 +1450,17 @@ gf_glusterlog_log_repetitions(glusterfs_ctx_t *ctx, const char *domain,
goto err;
}
- gf_time_fmt(timestr_latest, sizeof timestr_latest, latest.tv_sec,
- gf_timefmt_FT);
+ gf_time_fmt_tv(timestr_latest, sizeof timestr_latest, &latest,
+ gf_timefmt_FT);
- gf_time_fmt(timestr_oldest, sizeof timestr_oldest, oldest.tv_sec,
- gf_timefmt_FT);
+ gf_time_fmt_tv(timestr_oldest, sizeof timestr_oldest, &oldest,
+ gf_timefmt_FT);
if (errnum)
snprintf(errstr, sizeof(errstr) - 1, " [%s]", strerror(errnum));
- ret = gf_asprintf(&footer,
- "%s\" repeated %d times between"
- " [%s.%" GF_PRI_SUSECONDS "] and [%s.%" GF_PRI_SUSECONDS
- "]",
- errstr, refcount, timestr_oldest, oldest.tv_usec,
- timestr_latest, latest.tv_usec);
+ ret = gf_asprintf(&footer, "%s\" repeated %d times between [%s] and [%s]",
+ errstr, refcount, timestr_oldest, timestr_latest);
if (-1 == ret) {
ret = -1;
goto err;
@@ -2013,7 +2000,7 @@ _gf_log(const char *domain, const char *file, const char *function, int line,
const char *basename = NULL;
FILE *new_logfile = NULL;
va_list ap;
- char timestr[GF_LOG_TIMESTR_SIZE] = {
+ char timestr[GF_TIMESTR_SIZE] = {
0,
};
struct timeval tv = {
@@ -2100,12 +2087,11 @@ log:
if (-1 == ret)
goto out;
- gf_time_fmt(timestr, sizeof timestr, tv.tv_sec, gf_timefmt_FT);
+ gf_time_fmt_tv(timestr, sizeof timestr, &tv, gf_timefmt_FT);
- ret = gf_asprintf(
- &logline, "[%s.%" GF_PRI_SUSECONDS "] %c [%s:%d:%s] %d-%s: %s\n",
- timestr, tv.tv_usec, gf_level_strings[level], basename, line, function,
- ((this->graph) ? this->graph->id : 0), domain, msg);
+ ret = gf_asprintf(&logline, "[%s] %c [%s:%d:%s] %d-%s: %s\n", timestr,
+ gf_level_strings[level], basename, line, function,
+ ((this->graph) ? this->graph->id : 0), domain, msg);
if (-1 == ret) {
goto err;
}
@@ -2225,7 +2211,7 @@ int
gf_cmd_log(const char *domain, const char *fmt, ...)
{
va_list ap;
- char timestr[64];
+ char timestr[GF_TIMESTR_SIZE];
struct timeval tv = {
0,
};
@@ -2258,10 +2244,9 @@ gf_cmd_log(const char *domain, const char *fmt, ...)
goto out;
}
- gf_time_fmt(timestr, sizeof timestr, tv.tv_sec, gf_timefmt_FT);
+ gf_time_fmt_tv(timestr, sizeof timestr, &tv, gf_timefmt_FT);
- ret = gf_asprintf(&logline, "[%s.%" GF_PRI_SUSECONDS "] %s : %s\n", timestr,
- tv.tv_usec, domain, msg);
+ ret = gf_asprintf(&logline, "[%s] %s : %s\n", timestr, domain, msg);
if (ret == -1) {
goto out;
}
diff --git a/libglusterfs/src/mem-pool.c b/libglusterfs/src/mem-pool.c
index 1c31c27d95b..2d5a12b0a00 100644
--- a/libglusterfs/src/mem-pool.c
+++ b/libglusterfs/src/mem-pool.c
@@ -362,6 +362,30 @@ free:
FREE(ptr);
}
+#if defined(GF_DISABLE_MEMPOOL)
+
+struct mem_pool *
+mem_pool_new_fn(glusterfs_ctx_t *ctx, unsigned long sizeof_type,
+ unsigned long count, char *name)
+{
+ struct mem_pool *new;
+
+ new = GF_MALLOC(sizeof(struct mem_pool), gf_common_mt_mem_pool);
+ if (!new)
+ return NULL;
+
+ new->sizeof_type = sizeof_type;
+ return new;
+}
+
+void
+mem_pool_destroy(struct mem_pool *pool)
+{
+ GF_FREE(pool);
+}
+
+#else /* !GF_DISABLE_MEMPOOL */
+
static pthread_mutex_t pool_lock = PTHREAD_MUTEX_INITIALIZER;
static struct list_head pool_threads;
static pthread_mutex_t pool_free_lock = PTHREAD_MUTEX_INITIALIZER;
@@ -371,7 +395,6 @@ static size_t pool_list_size;
static __thread per_thread_pool_list_t *thread_pool_list = NULL;
-#if !defined(GF_DISABLE_MEMPOOL)
#define N_COLD_LISTS 1024
#define POOL_SWEEP_SECS 30
@@ -617,21 +640,29 @@ mem_pools_fini(void)
pthread_mutex_unlock(&init_mutex);
}
-#else
void
-mem_pools_init(void)
-{
-}
-void
-mem_pools_fini(void)
-{
-}
-void
-mem_pool_thread_destructor(per_thread_pool_list_t *pool_list)
+mem_pool_destroy(struct mem_pool *pool)
{
-}
+ if (!pool)
+ return;
-#endif
+ /* remove this pool from the owner (glusterfs_ctx_t) */
+ LOCK(&pool->ctx->lock);
+ {
+ list_del(&pool->owner);
+ }
+ UNLOCK(&pool->ctx->lock);
+
+ /* free this pool, but keep the mem_pool_shared */
+ GF_FREE(pool);
+
+ /*
+ * Pools are now permanent, so the mem_pool->pool is kept around. All
+ * of the objects *in* the pool will eventually be freed via the
+ * pool-sweeper thread, and this way we don't have to add a lot of
+ * reference-counting complexity.
+ */
+}
struct mem_pool *
mem_pool_new_fn(glusterfs_ctx_t *ctx, unsigned long sizeof_type,
@@ -682,6 +713,7 @@ mem_pool_new_fn(glusterfs_ctx_t *ctx, unsigned long sizeof_type,
new->sizeof_type = sizeof_type;
new->count = count;
new->name = name;
+ new->xl_name = THIS->name;
new->pool = pool;
GF_ATOMIC_INIT(new->active, 0);
#ifdef DEBUG
@@ -699,21 +731,6 @@ mem_pool_new_fn(glusterfs_ctx_t *ctx, unsigned long sizeof_type,
return new;
}
-void *
-mem_get0(struct mem_pool *mem_pool)
-{
- void *ptr = mem_get(mem_pool);
- if (ptr) {
-#if defined(GF_DISABLE_MEMPOOL)
- memset(ptr, 0, mem_pool->sizeof_type);
-#else
- memset(ptr, 0, AVAILABLE_SIZE(mem_pool->pool->power_of_two));
-#endif
- }
-
- return ptr;
-}
-
per_thread_pool_list_t *
mem_get_pool_list(void)
{
@@ -822,6 +839,23 @@ mem_get_from_pool(struct mem_pool *mem_pool)
return retval;
}
+#endif /* GF_DISABLE_MEMPOOL */
+
+void *
+mem_get0(struct mem_pool *mem_pool)
+{
+ void *ptr = mem_get(mem_pool);
+ if (ptr) {
+#if defined(GF_DISABLE_MEMPOOL)
+ memset(ptr, 0, mem_pool->sizeof_type);
+#else
+ memset(ptr, 0, AVAILABLE_SIZE(mem_pool->pool->power_of_two));
+#endif
+ }
+
+ return ptr;
+}
+
void *
mem_get(struct mem_pool *mem_pool)
{
@@ -896,27 +930,3 @@ mem_put(void *ptr)
}
#endif /* GF_DISABLE_MEMPOOL */
}
-
-void
-mem_pool_destroy(struct mem_pool *pool)
-{
- if (!pool)
- return;
-
- /* remove this pool from the owner (glusterfs_ctx_t) */
- LOCK(&pool->ctx->lock);
- {
- list_del(&pool->owner);
- }
- UNLOCK(&pool->ctx->lock);
-
- /* free this pool, but keep the mem_pool_shared */
- GF_FREE(pool);
-
- /*
- * Pools are now permanent, so the mem_pool->pool is kept around. All
- * of the objects *in* the pool will eventually be freed via the
- * pool-sweeper thread, and this way we don't have to add a lot of
- * reference-counting complexity.
- */
-}
diff --git a/libglusterfs/src/monitoring.c b/libglusterfs/src/monitoring.c
index 45e3d776903..fbb68dc8622 100644
--- a/libglusterfs/src/monitoring.c
+++ b/libglusterfs/src/monitoring.c
@@ -113,15 +113,15 @@ dump_latency_and_count(xlator_t *xl, int fd)
dprintf(fd, "%s.interval.%s.fail_count %" PRIu64 "\n", xl->name,
gf_fop_list[index], cbk);
}
- if (xl->stats.interval.latencies[index].count != 0.0) {
+ if (xl->stats.interval.latencies[index].count != 0) {
dprintf(fd, "%s.interval.%s.latency %lf\n", xl->name,
gf_fop_list[index],
- (xl->stats.interval.latencies[index].total /
+ (((double)xl->stats.interval.latencies[index].total) /
xl->stats.interval.latencies[index].count));
- dprintf(fd, "%s.interval.%s.max %lf\n", xl->name,
+ dprintf(fd, "%s.interval.%s.max %" PRIu64 "\n", xl->name,
gf_fop_list[index],
xl->stats.interval.latencies[index].max);
- dprintf(fd, "%s.interval.%s.min %lf\n", xl->name,
+ dprintf(fd, "%s.interval.%s.min %" PRIu64 "\n", xl->name,
gf_fop_list[index],
xl->stats.interval.latencies[index].min);
}
diff --git a/libglusterfs/src/stack.c b/libglusterfs/src/stack.c
index 371f60c4436..1531f0da43f 100644
--- a/libglusterfs/src/stack.c
+++ b/libglusterfs/src/stack.c
@@ -92,7 +92,7 @@ gf_proc_dump_call_frame(call_frame_t *call_frame, const char *key_buf, ...)
};
int ret = -1;
- char timestr[256] = {
+ char timestr[GF_TIMESTR_SIZE] = {
0,
};
int len;
@@ -162,7 +162,7 @@ gf_proc_dump_call_stack(call_stack_t *call_stack, const char *key_buf, ...)
va_list ap;
call_frame_t *trav;
int32_t i = 1, cnt = 0;
- char timestr[256] = {
+ char timestr[GF_TIMESTR_SIZE] = {
0,
};
int len;
diff --git a/libglusterfs/src/statedump.c b/libglusterfs/src/statedump.c
index 5e0f04f3217..65f0eb5c7f3 100644
--- a/libglusterfs/src/statedump.c
+++ b/libglusterfs/src/statedump.c
@@ -199,6 +199,40 @@ gf_proc_dump_write(char *key, char *value, ...)
return ret;
}
+void
+gf_latency_statedump_and_reset(char *key, gf_latency_t *lat)
+{
+ /* Doesn't make sense to continue if there are no fops
+ came in the given interval */
+ if (!lat || !lat->count)
+ return;
+ gf_proc_dump_write(key,
+ "AVG:%lf CNT:%" PRIu64 " TOTAL:%" PRIu64 " MIN:%" PRIu64
+ " MAX:%" PRIu64,
+ (((double)lat->total) / lat->count), lat->count,
+ lat->total, lat->min, lat->max);
+ gf_latency_reset(lat);
+}
+
+void
+gf_proc_dump_xl_latency_info(xlator_t *xl)
+{
+ char key_prefix[GF_DUMP_MAX_BUF_LEN];
+ char key[GF_DUMP_MAX_BUF_LEN];
+ int i;
+
+ snprintf(key_prefix, GF_DUMP_MAX_BUF_LEN, "%s.latency", xl->name);
+ gf_proc_dump_add_section("%s", key_prefix);
+
+ for (i = 0; i < GF_FOP_MAXVALUE; i++) {
+ gf_proc_dump_build_key(key, key_prefix, "%s", (char *)gf_fop_list[i]);
+
+ gf_latency_t *lat = &xl->stats.interval.latencies[i];
+
+ gf_latency_statedump_and_reset(key, lat);
+ }
+}
+
static void
gf_proc_dump_xlator_mem_info(xlator_t *xl)
{
@@ -270,7 +304,7 @@ gf_proc_dump_xlator_mem_info_only_in_use(xlator_t *xl)
void
gf_proc_dump_mem_info()
{
-#ifdef HAVE_MALLOC_STATS
+#ifdef HAVE_MALLINFO
struct mallinfo info;
memset(&info, 0, sizeof(struct mallinfo));
@@ -296,7 +330,7 @@ gf_proc_dump_mem_info_to_dict(dict_t *dict)
{
if (!dict)
return;
-#ifdef HAVE_MALLOC_STATS
+#ifdef HAVE_MALLINFO
struct mallinfo info;
int ret = -1;
@@ -349,26 +383,13 @@ gf_proc_dump_mem_info_to_dict(dict_t *dict)
void
gf_proc_dump_mempool_info(glusterfs_ctx_t *ctx)
{
+#ifdef GF_DISABLE_MEMPOOL
+ gf_proc_dump_write("built with --disable-mempool", " so no memory pools");
+#else
struct mem_pool *pool = NULL;
gf_proc_dump_add_section("mempool");
-#if defined(OLD_MEM_POOLS)
- list_for_each_entry(pool, &ctx->mempool_list, global_list)
- {
- gf_proc_dump_write("-----", "-----");
- gf_proc_dump_write("pool-name", "%s", pool->name);
- gf_proc_dump_write("hot-count", "%d", pool->hot_count);
- gf_proc_dump_write("cold-count", "%d", pool->cold_count);
- gf_proc_dump_write("padded_sizeof", "%lu", pool->padded_sizeof_type);
- gf_proc_dump_write("alloc-count", "%" PRIu64, pool->alloc_count);
- gf_proc_dump_write("max-alloc", "%d", pool->max_alloc);
-
- gf_proc_dump_write("pool-misses", "%" PRIu64, pool->pool_misses);
- gf_proc_dump_write("cur-stdalloc", "%d", pool->curr_stdalloc);
- gf_proc_dump_write("max-stdalloc", "%d", pool->max_stdalloc);
- }
-#else
LOCK(&ctx->lock);
{
list_for_each_entry(pool, &ctx->mempool_list, owner)
@@ -377,6 +398,7 @@ gf_proc_dump_mempool_info(glusterfs_ctx_t *ctx)
gf_proc_dump_write("-----", "-----");
gf_proc_dump_write("pool-name", "%s", pool->name);
+ gf_proc_dump_write("xlator-name", "%s", pool->xl_name);
gf_proc_dump_write("active-count", "%" GF_PRI_ATOMIC, active);
gf_proc_dump_write("sizeof-type", "%lu", pool->sizeof_type);
gf_proc_dump_write("padded-sizeof", "%d",
@@ -387,15 +409,13 @@ gf_proc_dump_mempool_info(glusterfs_ctx_t *ctx)
}
}
UNLOCK(&ctx->lock);
-
- /* TODO: details of (struct mem_pool_shared) pool->pool */
-#endif
+#endif /* GF_DISABLE_MEMPOOL */
}
void
gf_proc_dump_mempool_info_to_dict(glusterfs_ctx_t *ctx, dict_t *dict)
{
-#if defined(OLD_MEM_POOLS)
+#ifndef GF_DISABLE_MEMPOOL
struct mem_pool *pool = NULL;
char key[GF_DUMP_MAX_BUF_LEN] = {
0,
@@ -406,51 +426,47 @@ gf_proc_dump_mempool_info_to_dict(glusterfs_ctx_t *ctx, dict_t *dict)
if (!ctx || !dict)
return;
- list_for_each_entry(pool, &ctx->mempool_list, global_list)
+ LOCK(&ctx->lock);
{
- snprintf(key, sizeof(key), "pool%d.name", count);
- ret = dict_set_str(dict, key, pool->name);
- if (ret)
- return;
-
- snprintf(key, sizeof(key), "pool%d.hotcount", count);
- ret = dict_set_int32(dict, key, pool->hot_count);
- if (ret)
- return;
-
- snprintf(key, sizeof(key), "pool%d.coldcount", count);
- ret = dict_set_int32(dict, key, pool->cold_count);
- if (ret)
- return;
-
- snprintf(key, sizeof(key), "pool%d.paddedsizeof", count);
- ret = dict_set_uint64(dict, key, pool->padded_sizeof_type);
- if (ret)
- return;
-
- snprintf(key, sizeof(key), "pool%d.alloccount", count);
- ret = dict_set_uint64(dict, key, pool->alloc_count);
- if (ret)
- return;
-
- snprintf(key, sizeof(key), "pool%d.max_alloc", count);
- ret = dict_set_int32(dict, key, pool->max_alloc);
- if (ret)
- return;
-
- snprintf(key, sizeof(key), "pool%d.max-stdalloc", count);
- ret = dict_set_int32(dict, key, pool->max_stdalloc);
- if (ret)
- return;
-
- snprintf(key, sizeof(key), "pool%d.pool-misses", count);
- ret = dict_set_uint64(dict, key, pool->pool_misses);
- if (ret)
- return;
- count++;
+ list_for_each_entry(pool, &ctx->mempool_list, owner)
+ {
+ int64_t active = GF_ATOMIC_GET(pool->active);
+
+ snprintf(key, sizeof(key), "pool%d.name", count);
+ ret = dict_set_str(dict, key, pool->name);
+ if (ret)
+ goto out;
+
+ snprintf(key, sizeof(key), "pool%d.active-count", count);
+ ret = dict_set_uint64(dict, key, active);
+ if (ret)
+ goto out;
+
+ snprintf(key, sizeof(key), "pool%d.sizeof-type", count);
+ ret = dict_set_uint64(dict, key, pool->sizeof_type);
+ if (ret)
+ goto out;
+
+ snprintf(key, sizeof(key), "pool%d.padded-sizeof", count);
+ ret = dict_set_uint64(dict, key, 1 << pool->pool->power_of_two);
+ if (ret)
+ goto out;
+
+ snprintf(key, sizeof(key), "pool%d.size", count);
+ ret = dict_set_uint64(dict, key,
+ (1 << pool->pool->power_of_two) * active);
+ if (ret)
+ goto out;
+
+ snprintf(key, sizeof(key), "pool%d.shared-pool", count);
+ ret = dict_set_static_ptr(dict, key, pool->pool);
+ if (ret)
+ goto out;
+ }
}
- ret = dict_set_int32(dict, "mempool-count", count);
-#endif
+out:
+ UNLOCK(&ctx->lock);
+#endif /* !GF_DISABLE_MEMPOOL */
}
void
@@ -485,7 +501,7 @@ gf_proc_dump_single_xlator_info(xlator_t *trav)
return;
if (ctx->measure_latency)
- gf_proc_dump_latency_info(trav);
+ gf_proc_dump_xl_latency_info(trav);
gf_proc_dump_xlator_mem_info(trav);
@@ -782,7 +798,7 @@ gf_proc_dump_info(int signum, glusterfs_ctx_t *ctx)
char brick_name[PATH_MAX] = {
0,
};
- char timestr[256] = {
+ char timestr[GF_TIMESTR_SIZE] = {
0,
};
char sign_string[512] = {
@@ -842,7 +858,7 @@ gf_proc_dump_info(int signum, glusterfs_ctx_t *ctx)
? dump_options.dump_path
: ((ctx->statedump_path != NULL) ? ctx->statedump_path
: DEFAULT_VAR_RUN_DIRECTORY)),
- brick_name, getpid(), (uint64_t)time(NULL));
+ brick_name, getpid(), (uint64_t)gf_time());
if ((ret < 0) || (ret >= sizeof(path))) {
goto out;
}
@@ -861,10 +877,7 @@ gf_proc_dump_info(int signum, glusterfs_ctx_t *ctx)
// continue even though gettimeofday() has failed
ret = gettimeofday(&tv, NULL);
if (0 == ret) {
- gf_time_fmt(timestr, sizeof timestr, tv.tv_sec, gf_timefmt_FT);
- len = strlen(timestr);
- snprintf(timestr + len, sizeof timestr - len, ".%" GF_PRI_SUSECONDS,
- tv.tv_usec);
+ gf_time_fmt_tv(timestr, sizeof timestr, &tv, gf_timefmt_FT);
}
len = snprintf(sign_string, sizeof(sign_string), "DUMP-START-TIME: %s\n",
@@ -913,10 +926,7 @@ gf_proc_dump_info(int signum, glusterfs_ctx_t *ctx)
ret = gettimeofday(&tv, NULL);
if (0 == ret) {
- gf_time_fmt(timestr, sizeof timestr, tv.tv_sec, gf_timefmt_FT);
- len = strlen(timestr);
- snprintf(timestr + len, sizeof timestr - len, ".%" GF_PRI_SUSECONDS,
- tv.tv_usec);
+ gf_time_fmt_tv(timestr, sizeof timestr, &tv, gf_timefmt_FT);
}
len = snprintf(sign_string, sizeof(sign_string), "\nDUMP-END-TIME: %s",
@@ -1035,7 +1045,7 @@ gf_proc_dump_xlator_profile(xlator_t *this, strfd_t *strfd)
{
gf_dump_strfd = strfd;
- gf_proc_dump_latency_info(this);
+ gf_proc_dump_xl_latency_info(this);
gf_dump_strfd = NULL;
}
diff --git a/libglusterfs/src/store.c b/libglusterfs/src/store.c
index f444741ef0c..5c316b9291a 100644
--- a/libglusterfs/src/store.c
+++ b/libglusterfs/src/store.c
@@ -649,23 +649,25 @@ out:
}
int32_t
-gf_store_iter_destroy(gf_store_iter_t *iter)
+gf_store_iter_destroy(gf_store_iter_t **iter)
{
int32_t ret = -1;
- if (!iter)
+ if (!(*iter))
return 0;
/* gf_store_iter_new will not return a valid iter object with iter->file
* being NULL*/
- ret = fclose(iter->file);
+ ret = fclose((*iter)->file);
if (ret)
gf_msg("", GF_LOG_ERROR, errno, LG_MSG_FILE_OP_FAILED,
"Unable"
" to close file: %s, ret: %d",
- iter->filepath, ret);
+ (*iter)->filepath, ret);
+
+ GF_FREE(*iter);
+ *iter = NULL;
- GF_FREE(iter);
return ret;
}
diff --git a/libglusterfs/src/syncop.c b/libglusterfs/src/syncop.c
index e20cf006493..df20cec559f 100644
--- a/libglusterfs/src/syncop.c
+++ b/libglusterfs/src/syncop.c
@@ -11,6 +11,10 @@
#include "glusterfs/syncop.h"
#include "glusterfs/libglusterfs-messages.h"
+#ifdef HAVE_TSAN_API
+#include <sanitizer/tsan_interface.h>
+#endif
+
int
syncopctx_setfsuid(void *uid)
{
@@ -262,6 +266,11 @@ synctask_yield(struct synctask *task, struct timespec *delta)
if (task->state != SYNCTASK_DONE) {
task->state = SYNCTASK_SUSPEND;
}
+
+#ifdef HAVE_TSAN_API
+ __tsan_switch_to_fiber(task->proc->tsan.fiber, 0);
+#endif
+
if (swapcontext(&task->ctx, &task->proc->sched) < 0) {
gf_msg("syncop", GF_LOG_ERROR, errno, LG_MSG_SWAPCONTEXT_FAILED,
"swapcontext failed");
@@ -357,6 +366,10 @@ synctask_destroy(struct synctask *task)
pthread_cond_destroy(&task->cond);
}
+#ifdef HAVE_TSAN_API
+ __tsan_destroy_fiber(task->tsan.fiber);
+#endif
+
GF_FREE(task);
}
@@ -467,6 +480,13 @@ synctask_create(struct syncenv *env, size_t stacksize, synctask_fn_t fn,
makecontext(&newtask->ctx, (void (*)(void))synctask_wrap, 0);
+#ifdef HAVE_TSAN_API
+ newtask->tsan.fiber = __tsan_create_fiber(0);
+ snprintf(newtask->tsan.name, TSAN_THREAD_NAMELEN, "<synctask of %s>",
+ this->name);
+ __tsan_set_fiber_name(newtask->tsan.fiber, newtask->tsan.name);
+#endif
+
newtask->state = SYNCTASK_INIT;
newtask->slept = 1;
@@ -573,7 +593,7 @@ syncenv_task(struct syncproc *proc)
env->procs_idle++;
- sleep_till.tv_sec = time(NULL) + SYNCPROC_IDLE_TIME;
+ sleep_till.tv_sec = gf_time() + SYNCPROC_IDLE_TIME;
ret = pthread_cond_timedwait(&env->cond, &env->mutex, &sleep_till);
env->procs_idle--;
@@ -638,6 +658,10 @@ synctask_switchto(struct synctask *task)
task->ctx.uc_flags &= ~_UC_TLSBASE;
#endif
+#ifdef HAVE_TSAN_API
+ __tsan_switch_to_fiber(task->tsan.fiber, 0);
+#endif
+
if (swapcontext(&task->proc->sched, &task->ctx) < 0) {
gf_msg("syncop", GF_LOG_ERROR, errno, LG_MSG_SWAPCONTEXT_FAILED,
"swapcontext failed");
@@ -675,10 +699,21 @@ syncenv_processor(void *thdata)
proc = thdata;
+#ifdef HAVE_TSAN_API
+ proc->tsan.fiber = __tsan_create_fiber(0);
+ snprintf(proc->tsan.name, TSAN_THREAD_NAMELEN, "<sched of syncenv@%p>",
+ proc);
+ __tsan_set_fiber_name(proc->tsan.fiber, proc->tsan.name);
+#endif
+
while ((task = syncenv_task(proc)) != NULL) {
synctask_switchto(task);
}
+#ifdef HAVE_TSAN_API
+ __tsan_destroy_fiber(proc->tsan.fiber);
+#endif
+
return NULL;
}
@@ -3055,12 +3090,13 @@ syncop_seek(xlator_t *subvol, fd_t *fd, off_t offset, gf_seek_what_t what,
SYNCOP(subvol, (&args), syncop_seek_cbk, subvol->fops->seek, fd, offset,
what, xdata_in);
- if (*off)
- *off = args.offset;
-
- if (args.op_ret == -1)
+ if (args.op_ret < 0) {
return -args.op_errno;
- return args.op_ret;
+ } else {
+ if (off)
+ *off = args.offset;
+ return args.op_ret;
+ }
}
int
diff --git a/libglusterfs/src/syscall.c b/libglusterfs/src/syscall.c
index 007e3e84eea..04400f98b6c 100644
--- a/libglusterfs/src/syscall.c
+++ b/libglusterfs/src/syscall.c
@@ -13,6 +13,10 @@
#include "glusterfs/mem-pool.h"
#include "glusterfs/libglusterfs-messages.h"
+#ifdef __FreeBSD__
+#include <sys/sysctl.h>
+#include <signal.h>
+#endif
#include <sys/types.h>
#include <utime.h>
#include <sys/time.h>
@@ -506,7 +510,7 @@ sys_lsetxattr(const char *path, const char *name, const void *value,
#endif
#ifdef GF_BSD_HOST_OS
- return FS_RET_CHECK0(
+ return FS_RET_CHECK(
extattr_set_link(path, EXTATTR_NAMESPACE_USER, name, value, size),
errno);
#endif
@@ -624,7 +628,7 @@ sys_fsetxattr(int filedes, const char *name, const void *value, size_t size,
#endif
#ifdef GF_BSD_HOST_OS
- return FS_RET_CHECK0(
+ return FS_RET_CHECK(
extattr_set_fd(filedes, EXTATTR_NAMESPACE_USER, name, value, size),
errno);
#endif
@@ -854,3 +858,19 @@ sys_copy_file_range(int fd_in, off64_t *off_in, int fd_out, off64_t *off_out,
#endif /* HAVE_COPY_FILE_RANGE_SYS */
#endif /* HAVE_COPY_FILE_RANGE */
}
+
+#ifdef __FreeBSD__
+int
+sys_kill(pid_t pid, int sig)
+{
+ return FS_RET_CHECK0(kill(pid, sig), errno);
+}
+
+int
+sys_sysctl(const int *name, u_int namelen, void *oldp, size_t *oldlenp,
+ const void *newp, size_t newlen)
+{
+ return FS_RET_CHECK0(sysctl(name, namelen, oldp, oldlenp, newp, newlen),
+ errno);
+}
+#endif
diff --git a/libglusterfs/src/xlator.c b/libglusterfs/src/xlator.c
index e793e1e3d68..9a2582d45d5 100644
--- a/libglusterfs/src/xlator.c
+++ b/libglusterfs/src/xlator.c
@@ -184,9 +184,11 @@ xlator_volopt_dynload(char *xlator_type, void **dl_handle,
volume_opt_list_t *opt_list)
{
int ret = -1;
+ int flag = 0;
char *name = NULL;
void *handle = NULL;
xlator_api_t *xlapi = NULL;
+ volume_option_t *opt = NULL;
GF_VALIDATE_OR_GOTO("xlator", xlator_type, out);
@@ -194,8 +196,10 @@ xlator_volopt_dynload(char *xlator_type, void **dl_handle,
* need this check */
if (!strstr(xlator_type, "rpc-transport"))
ret = gf_asprintf(&name, "%s/%s.so", XLATORDIR, xlator_type);
- else
+ else {
+ flag = 1;
ret = gf_asprintf(&name, "%s/%s.so", XLATORPARENTDIR, xlator_type);
+ }
if (-1 == ret) {
goto out;
}
@@ -211,18 +215,29 @@ xlator_volopt_dynload(char *xlator_type, void **dl_handle,
goto out;
}
- /* check new struct first, and then check this */
- xlapi = dlsym(handle, "xlator_api");
- if (!xlapi) {
- gf_smsg("xlator", GF_LOG_ERROR, 0, LG_MSG_DLSYM_ERROR, "error=%s",
- dlerror(), NULL);
- goto out;
- }
+ if (flag == 0) {
+ /* check new struct first, and then check this */
+ xlapi = dlsym(handle, "xlator_api");
+ if (!xlapi) {
+ gf_smsg("xlator", GF_LOG_ERROR, 0, LG_MSG_DLSYM_ERROR, "error=%s",
+ dlerror(), NULL);
+ goto out;
+ }
- opt_list->given_opt = xlapi->options;
- if (!opt_list->given_opt) {
- gf_smsg("xlator", GF_LOG_ERROR, 0, LG_MSG_LOAD_FAILED, NULL);
- goto out;
+ opt_list->given_opt = xlapi->options;
+ if (!opt_list->given_opt) {
+ gf_smsg("xlator", GF_LOG_ERROR, 0, LG_MSG_LOAD_FAILED, NULL);
+ goto out;
+ }
+ } else {
+ opt = dlsym(handle, "options");
+ if (!opt) {
+ gf_smsg("xlator", GF_LOG_ERROR, 0, LG_MSG_DLSYM_ERROR, "error=%s",
+ dlerror(), NULL);
+ goto out;
+ }
+
+ opt_list->given_opt = opt;
}
*dl_handle = handle;
@@ -245,6 +260,7 @@ xlator_dynload_apis(xlator_t *xl)
void *handle = NULL;
volume_opt_list_t *vol_opt = NULL;
xlator_api_t *xlapi = NULL;
+ int i = 0;
handle = xl->dlhandle;
@@ -342,6 +358,10 @@ xlator_dynload_apis(xlator_t *xl)
memcpy(xl->op_version, xlapi->op_version,
sizeof(uint32_t) * GF_MAX_RELEASES);
+ for (i = 0; i < GF_FOP_MAXVALUE; i++) {
+ gf_latency_reset(&xl->stats.interval.latencies[i]);
+ }
+
ret = 0;
out:
return ret;
@@ -806,7 +826,7 @@ xlator_members_free(xlator_t *xl)
GF_FREE(xl->name);
GF_FREE(xl->type);
- if (!(xl->ctx && xl->ctx->cmd_args.valgrind) && xl->dlhandle)
+ if (!(xl->ctx && xl->ctx->cmd_args.vgtool != _gf_none) && xl->dlhandle)
dlclose(xl->dlhandle);
if (xl->options)
dict_unref(xl->options);
diff --git a/rfc.sh b/rfc.sh
index 223ebc2825b..e7faec9ea0f 100755
--- a/rfc.sh
+++ b/rfc.sh
@@ -4,7 +4,29 @@
# i.e. where we are interested in the result of a command,
# we have to run the command in an if-statement.
-ORIGIN=${GLUSTER_ORIGIN:-origin}
+UPSTREAM=${GLUSTER_UPSTREAM}
+if [ "x$UPSTREAM" -eq "x" ]; then
+ for rmt in $(git remote); do
+ rmt_repo=$(git remote show $rmt -n | grep Fetch | awk '{ print $3 }');
+ if [ $rmt_repo -eq "git@github:gluster/glusterfs" ]; then
+ UPSTREAM=$rmt
+ echo "Picked $rmt as upstream remote"
+ break
+ fi
+ done
+fi
+
+USER_REPO=${GLUSTER_USER_REPO:-origin}
+if [ "x${USER_REPO}" -eq "x${UPSTREAM}" ] ; then
+ echo "When you submit patches, it should get submitted to your fork, not to upstream directly"
+ echo "If you are not sure, check `for rmt in $(git remote); do git remote show $rmt -n; done`"
+ echo "And pick the correct remote you would like to push to and do `export GLUSTER_USER_REPO=$rmt`"
+ echo ""
+ echo "Exiting..."
+ exit 1
+fi
+
+
while getopts "v" opt; do
case $opt in
@@ -18,7 +40,7 @@ done
shift $((OPTIND-1))
-branch="release-8";
+branch="devel";
set_hooks_commit_msg()
{
@@ -50,21 +72,21 @@ is_num()
backport_id_message()
{
echo ""
- echo "This commit is to a non-master branch, and hence is treated as a backport."
+ echo "This commit is to a non-devel branch, and hence is treated as a backport."
echo ""
echo "For backports we would like to retain the same gerrit Change-Id across"
echo "branches. On auto inspection it is found that a gerrit Change-Id is"
- echo "missing, or the Change-Id is not found on your local master"
+ echo "missing, or the Change-Id is not found on your local devel branch"
echo ""
echo "This could mean a few things:"
echo " 1. This is not a backport, hence choose Y on the prompt to proceed"
- echo " 2. Your $ORIGIN master is not up to date, hence the script is unable"
- echo " to find the corresponding Change-Id on master. Either choose N,"
+ echo " 2. Your $USER_REPO/devel is not up to date, hence the script is unable"
+ echo " to find the corresponding Change-Id on devel. Either choose N,"
echo " 'git fetch', and try again, OR if you are sure you used the"
echo " same Change-Id, choose Y at the prompt to proceed"
echo " 3. You commented or removed the Change-Id in your commit message after"
echo " cherry picking the commit. Choose N, fix the commit message to"
- echo " use the same Change-Id as master (git commit --amend), resubmit"
+ echo " use the same Change-Id as 'devel' (git commit --amend), resubmit"
echo ""
}
@@ -72,8 +94,8 @@ check_backport()
{
moveon='N'
- # Backports are never made to master
- if [ $branch = "master" ]; then
+ # Backports are never made to 'devel'
+ if [ $branch = "devel" ]; then
return;
fi
@@ -86,22 +108,22 @@ check_backport()
echo -n "Did not find a Change-Id for a possible backport. Continue (y/N): "
read moveon
else
- # Search master for the same change ID (rebase_changes has run, so we
+ # Search 'devel' for the same change ID (rebase_changes has run, so we
# should never not find a Change-Id)
- mchangeid=$(git log $ORIGIN/master --format='%b' --grep="^Change-Id: ${changeid}" | grep ${changeid} | awk '{print $2}')
+ mchangeid=$(git log $UPSTREAM/devel --format='%b' --grep="^Change-Id: ${changeid}" | grep ${changeid} | awk '{print $2}')
- # Check if we found the change ID on master, else throw a message to
+ # Check if we found the change ID on 'devel', else throw a message to
# decide if we should continue.
- # NOTE: If master was not rebased, we will not find the Change-ID and
+ # NOTE: If 'devel' was not rebased, we will not find the Change-ID and
# could hit a false positive case here (or if someone checks out some
- # other branch as master).
+ # other branch as 'devel').
if [ "${mchangeid}" = "${changeid}" ]; then
moveon="Y"
else
backport_id_message;
echo "Change-Id of commit: $changeid"
- echo "Change-Id on master: $mchangeid"
- echo -n "Did not find mentioned Change-Id on master for a possible backport. Continue (y/N): "
+ echo "Change-Id on devel: $mchangeid"
+ echo -n "Did not find mentioned Change-Id on 'devel' for a possible backport. Continue (y/N): "
read moveon
fi
fi
@@ -116,7 +138,7 @@ check_backport()
rebase_changes()
{
- GIT_EDITOR=$0 git rebase -i $ORIGIN/$branch;
+ GIT_EDITOR=$0 git rebase -i $UPSTREAM/$branch;
}
@@ -154,15 +176,12 @@ rebase_changes()
# [[:space:]]+
# Followed by 1 or more spaces
#
-# (gluster\/glusterfs|bz)?
-# Followed by nothing, 'gluster/glusterfs' or 'bz'
-#
# #
# Followed by #
#
# [[:digit:]]+
# Followed by 1 or more digits
-REFRE="^[[:space:]]*(Fixes|Updates)(:)?[[:space:]]+(gluster\/glusterfs|bz)?#[[:digit:]]+"
+REFRE="^[[:space:]]*(Fixes|Updates)(:)?[[:space:]]+#[[:digit:]]+"
editor_mode()
{
@@ -181,22 +200,17 @@ editor_mode()
while true; do
echo Commit: "\"$(head -n 1 $1)\""
- echo -n "Reference (Bugzilla ID or Github Issue ID): "
- read bug
- if [ -z "$bug" ]; then
+ echo -n "Github Issue ID: "
+ read issue
+ if [ -z "$issue" ]; then
return;
fi
- if ! is_num "$bug"; then
- echo "Invalid reference ID ($bug)!!!";
+ if ! is_num "$issue"; then
+ echo "Invalid Github Issue ID!!!";
continue;
fi
- bz_string="bz"
- if [ $bug -lt 742000 ]; then
- bz_string=""
- fi
-
- echo "Select yes '(y)' if this patch fixes the bug/feature completely,"
+ echo "Select yes '(y)' if this patch fixes the issue/feature completely,"
echo -n "or is the last of the patchset which brings feature (Y/n): "
read fixes
fixes_string="Fixes"
@@ -204,7 +218,7 @@ editor_mode()
fixes_string="Updates"
fi
- sed "/^Change-Id:/{p; s/^.*$/${fixes_string}: ${bz_string}#${bug}/;}" $1 > $1.new && \
+ sed "/^Change-Id:/{p; s/^.*$/${fixes_string}: #${issue}/;}" $1 > $1.new && \
mv $1.new $1;
return;
done
@@ -220,7 +234,7 @@ EOF
assert_diverge()
{
- git diff $ORIGIN/$branch..HEAD | grep -q .;
+ git diff $UPSTREAM/$branch..HEAD | grep -q .;
}
@@ -229,34 +243,24 @@ warn_reference_missing()
echo ""
echo "=== Missing a reference in commit! ==="
echo ""
- echo "Gluster commits are made with a reference to a bug or a github issue"
- echo ""
- echo "Submissions that are enhancements (IOW, not functional"
- echo "bug fixes, but improvements of any nature to the code) are tracked"
- echo "using github issues [1]."
+ echo "Gluster commits are made with a reference to a github issue"
echo ""
- echo "Submissions that are bug fixes are tracked using Bugzilla [2]."
+ echo "A check on the commit message, reveals that there is no "
+ echo "github issue referenced in the commit message."
echo ""
- echo "A check on the commit message, reveals that there is no bug or"
- echo "github issue referenced in the commit message"
+ echo "https://github.com/gluster/glusterfs/issues/new"
echo ""
- echo "[1] https://github.com/gluster/glusterfs/issues/new"
- echo "[2] https://bugzilla.redhat.com/enter_bug.cgi?product=GlusterFS"
+ echo "Please open an issue and reference the same in the commit message "
+ echo "using the following tags:"
echo ""
- echo "Please file an issue or a bug report and reference the same in the"
- echo "commit message using the following tags:"
- echo "GitHub Issues:"
- echo "\"Fixes: gluster/glusterfs#n\" OR \"Updates: gluster/glusterfs#n\","
- echo "\"Fixes: #n\" OR \"Updates: #n\","
- echo "Bugzilla ID:"
- echo "\"Fixes: bz#n\" OR \"Updates: bz#n\","
- echo "where n is the issue or bug number"
+ echo "\"Fixes: #NNNN\" OR \"Updates: #NNNN\","
+ echo "where NNNN is the issue id"
echo ""
echo "You may abort the submission choosing 'N' below and use"
echo "'git commit --amend' to add the issue reference before posting"
echo "to gerrit."
echo ""
- echo -n "Missing reference to a bug or a github issue. Continue (y/N): "
+ echo -n "Missing reference to a github issue. Continue (y/N): "
read moveon
if [ "${moveon}" = 'Y' ] || [ "${moveon}" = 'y' ]; then
return;
@@ -276,7 +280,7 @@ main()
return;
fi
- git fetch $ORIGIN;
+ git fetch $UPSTREAM;
rebase_changes;
@@ -287,9 +291,9 @@ main()
# see note above variable REFRE for regex elaboration
reference=$(git log -n1 --format='%b' | grep -iow -E "${REFRE}" | awk -F '#' '{print $2}');
- # If this is a commit against master and does not have a bug ID or a github
+ # If this is a commit against 'devel' and does not have a github
# issue reference. Warn the contributor that one of the 2 is required
- if [ -z "${reference}" ] && [ $branch = "master" ]; then
+ if [ -z "${reference}" ] && [ $branch = "devel" ]; then
warn_reference_missing;
fi
@@ -317,9 +321,9 @@ main()
fi
if [ -z "${reference}" ]; then
- $drier git push $ORIGIN HEAD:refs/for/$branch/rfc;
+ $drier git push $USER_REPO HEAD:temp_${branch}/$(date +%Y-%m-%d_%s);
else
- $drier git push $ORIGIN HEAD:refs/for/$branch/ref-${reference};
+ $drier git push $USER_REPO HEAD:issue${reference}_${branch};
fi
}
diff --git a/rpc/rpc-lib/src/Makefile.am b/rpc/rpc-lib/src/Makefile.am
index 81a96476883..35c9db07e7f 100644
--- a/rpc/rpc-lib/src/Makefile.am
+++ b/rpc/rpc-lib/src/Makefile.am
@@ -2,7 +2,7 @@ lib_LTLIBRARIES = libgfrpc.la
libgfrpc_la_SOURCES = auth-unix.c rpcsvc-auth.c rpcsvc.c auth-null.c \
rpc-transport.c xdr-rpc.c xdr-rpcclnt.c rpc-clnt.c auth-glusterfs.c \
- rpc-drc.c $(CONTRIBDIR)/sunrpc/xdr_sizeof.c rpc-clnt-ping.c \
+ rpc-drc.c rpc-clnt-ping.c \
autoscale-threads.c mgmt-pmap.c
EXTRA_DIST = libgfrpc.sym
diff --git a/rpc/rpc-lib/src/libgfrpc.sym b/rpc/rpc-lib/src/libgfrpc.sym
index 54d1be1112f..e026d80259b 100644
--- a/rpc/rpc-lib/src/libgfrpc.sym
+++ b/rpc/rpc-lib/src/libgfrpc.sym
@@ -65,3 +65,4 @@ rpc_transport_unix_options_build
rpc_transport_unref
rpc_clnt_mgmt_pmap_signout
rpcsvc_autoscale_threads
+rpcsvc_statedump
diff --git a/rpc/rpc-lib/src/protocol-common.h b/rpc/rpc-lib/src/protocol-common.h
index 7275d7568b6..0cb5862e9a9 100644
--- a/rpc/rpc-lib/src/protocol-common.h
+++ b/rpc/rpc-lib/src/protocol-common.h
@@ -309,6 +309,7 @@ enum glusterd_mgmt_v3_procnum {
GLUSTERD_MGMT_V3_PRE_VALIDATE,
GLUSTERD_MGMT_V3_BRICK_OP,
GLUSTERD_MGMT_V3_COMMIT,
+ GLUSTERD_MGMT_V3_POST_COMMIT,
GLUSTERD_MGMT_V3_POST_VALIDATE,
GLUSTERD_MGMT_V3_UNLOCK,
GLUSTERD_MGMT_V3_MAXVALUE,
diff --git a/rpc/rpc-lib/src/rpc-clnt-ping.c b/rpc/rpc-lib/src/rpc-clnt-ping.c
index 2298ef6394f..31f17841bea 100644
--- a/rpc/rpc-lib/src/rpc-clnt-ping.c
+++ b/rpc/rpc-lib/src/rpc-clnt-ping.c
@@ -122,7 +122,7 @@ rpc_clnt_ping_timer_expired(void *rpc_ptr)
goto out;
}
- clock_gettime(CLOCK_REALTIME, &current);
+ timespec_now_realtime(&current);
pthread_mutex_lock(&conn->lock);
{
unref = rpc_clnt_remove_ping_timer_locked(rpc);
diff --git a/rpc/rpc-lib/src/rpc-clnt.c b/rpc/rpc-lib/src/rpc-clnt.c
index dac707664df..517037c4a5d 100644
--- a/rpc/rpc-lib/src/rpc-clnt.c
+++ b/rpc/rpc-lib/src/rpc-clnt.c
@@ -97,7 +97,7 @@ call_bail(void *data)
struct saved_frame *saved_frame = NULL;
struct saved_frame *trav = NULL;
struct saved_frame *tmp = NULL;
- char frame_sent[256] = {
+ char frame_sent[GF_TIMESTR_SIZE] = {
0,
};
struct timespec timeout = {
@@ -105,7 +105,6 @@ call_bail(void *data)
};
char peerid[UNIX_PATH_MAX] = {0};
gf_boolean_t need_unref = _gf_false;
- int len;
GF_VALIDATE_OR_GOTO("client", data, out);
@@ -165,11 +164,8 @@ call_bail(void *data)
list_for_each_entry_safe(trav, tmp, &list, list)
{
- gf_time_fmt(frame_sent, sizeof frame_sent, trav->saved_at.tv_sec,
- gf_timefmt_FT);
- len = strlen(frame_sent);
- snprintf(frame_sent + len, sizeof(frame_sent) - len,
- ".%" GF_PRI_SUSECONDS, trav->saved_at.tv_usec);
+ gf_time_fmt_tv(frame_sent, sizeof frame_sent, &trav->saved_at,
+ gf_timefmt_FT);
gf_log(conn->name, GF_LOG_ERROR,
"bailing out frame type(%s), op(%s(%d)), xid = 0x%x, "
@@ -317,20 +313,15 @@ saved_frames_unwind(struct saved_frames *saved_frames)
{
struct saved_frame *trav = NULL;
struct saved_frame *tmp = NULL;
- char timestr[1024] = {
+ char timestr[GF_TIMESTR_SIZE] = {
0,
};
- int len;
list_splice_init(&saved_frames->lk_sf.list, &saved_frames->sf.list);
list_for_each_entry_safe(trav, tmp, &saved_frames->sf.list, list)
{
- gf_time_fmt(timestr, sizeof timestr, trav->saved_at.tv_sec,
- gf_timefmt_FT);
- len = strlen(timestr);
- snprintf(timestr + len, sizeof(timestr) - len, ".%" GF_PRI_SUSECONDS,
- trav->saved_at.tv_usec);
+ gf_time_fmt_tv(timestr, sizeof timestr, &trav->saved_at, gf_timefmt_FT);
if (!trav->rpcreq || !trav->rpcreq->prog)
continue;
@@ -928,7 +919,7 @@ rpc_clnt_notify(rpc_transport_t *trans, void *mydata,
}
case RPC_TRANSPORT_MSG_RECEIVED: {
- clock_gettime(CLOCK_REALTIME, &conn->last_received);
+ timespec_now_realtime(&conn->last_received);
pollin = data;
if (pollin->is_reply)
@@ -942,8 +933,7 @@ rpc_clnt_notify(rpc_transport_t *trans, void *mydata,
}
case RPC_TRANSPORT_MSG_SENT: {
- clock_gettime(CLOCK_REALTIME, &conn->last_sent);
-
+ timespec_now_realtime(&conn->last_sent);
ret = 0;
break;
}
@@ -960,6 +950,7 @@ rpc_clnt_notify(rpc_transport_t *trans, void *mydata,
conn->config.remote_port = 0;
conn->connected = 1;
conn->disconnected = 0;
+ pthread_cond_broadcast(&conn->cond);
}
pthread_mutex_unlock(&conn->lock);
@@ -1005,6 +996,7 @@ rpc_clnt_connection_init(struct rpc_clnt *clnt, glusterfs_ctx_t *ctx,
conn = &clnt->conn;
pthread_mutex_init(&clnt->conn.lock, NULL);
+ pthread_cond_init(&clnt->conn.cond, NULL);
conn->name = gf_strdup(name);
if (!conn->name) {
@@ -1830,6 +1822,7 @@ rpc_clnt_destroy(struct rpc_clnt *rpc)
saved_frames_destroy(saved_frames);
pthread_mutex_destroy(&rpc->lock);
pthread_mutex_destroy(&rpc->conn.lock);
+ pthread_cond_destroy(&rpc->conn.cond);
/* mem-pool should be destroyed, otherwise,
it will cause huge memory leaks */
diff --git a/rpc/rpc-lib/src/rpc-clnt.h b/rpc/rpc-lib/src/rpc-clnt.h
index 0d0b115578c..2945265200b 100644
--- a/rpc/rpc-lib/src/rpc-clnt.h
+++ b/rpc/rpc-lib/src/rpc-clnt.h
@@ -127,6 +127,7 @@ struct rpc_clnt_config {
struct rpc_clnt_connection {
pthread_mutex_t lock;
+ pthread_cond_t cond;
rpc_transport_t *trans;
struct rpc_clnt_config config;
gf_timer_t *reconnect;
diff --git a/rpc/rpc-lib/src/rpcsvc.c b/rpc/rpc-lib/src/rpcsvc.c
index 81d40c8ec0e..39910d481bf 100644
--- a/rpc/rpc-lib/src/rpcsvc.c
+++ b/rpc/rpc-lib/src/rpcsvc.c
@@ -13,6 +13,7 @@
#include <glusterfs/dict.h>
#include <glusterfs/byte-order.h>
#include <glusterfs/compat-errno.h>
+#include <glusterfs/statedump.h>
#include "xdr-rpc.h"
#include <glusterfs/iobuf.h>
#include "xdr-common.h"
@@ -41,6 +42,10 @@
#include "xdr-rpcclnt.h"
#include <glusterfs/glusterfs-acl.h>
+#ifndef PTHREAD_MUTEX_ADAPTIVE_NP
+#define PTHREAD_MUTEX_ADAPTIVE_NP PTHREAD_MUTEX_DEFAULT
+#endif
+
static struct rpcsvc_program gluster_dump_prog;
#define rpcsvc_alloc_request(svc, request) \
@@ -66,59 +71,33 @@ rpcsvc_request_handler(void *arg);
static int
rpcsvc_match_subnet_v4(const char *addrtok, const char *ipaddr);
-void
+static void
rpcsvc_toggle_queue_status(rpcsvc_program_t *prog,
- rpcsvc_request_queue_t *queue, char status[])
+ rpcsvc_request_queue_t *queue,
+ unsigned long status[])
{
- int queue_index = 0, status_index = 0, set_bit = 0;
-
- if (queue != &prog->request_queue[0]) {
- queue_index = (queue - &prog->request_queue[0]);
- }
-
- status_index = queue_index / 8;
- set_bit = queue_index % 8;
+ unsigned queue_index = queue - prog->request_queue;
- status[status_index] ^= (1 << set_bit);
-
- return;
+ status[queue_index / __BITS_PER_LONG] ^= (1UL << (queue_index %
+ __BITS_PER_LONG));
}
int
rpcsvc_get_free_queue_index(rpcsvc_program_t *prog)
{
- int queue_index = 0, max_index = 0, i = 0;
- unsigned int right_most_unset_bit = 0;
+ unsigned i, j = 0;
- right_most_unset_bit = 8;
-
- max_index = gf_roof(EVENT_MAX_THREADS, 8) / 8;
- for (i = 0; i < max_index; i++) {
- if (prog->request_queue_status[i] == 0) {
- right_most_unset_bit = 0;
+ for (i = 0; i < EVENT_MAX_THREADS / __BITS_PER_LONG; i++)
+ if (prog->request_queue_status[i] != ULONG_MAX) {
+ j = __builtin_ctzl(~prog->request_queue_status[i]);
break;
- } else {
- /* get_rightmost_set_bit (sic)*/
- right_most_unset_bit = __builtin_ctz(
- ~prog->request_queue_status[i]);
- if (right_most_unset_bit < 8) {
- break;
- }
}
- }
- if (right_most_unset_bit > 7) {
- queue_index = -1;
- } else {
- queue_index = i * 8;
- queue_index += right_most_unset_bit;
- }
-
- if (queue_index != -1) {
- prog->request_queue_status[i] |= (0x1 << right_most_unset_bit);
- }
+ if (i == EVENT_MAX_THREADS / __BITS_PER_LONG)
+ return -1;
- return queue_index;
+ prog->request_queue_status[i] |= (1UL << j);
+ return i * __BITS_PER_LONG + j;
}
rpcsvc_notify_wrapper_t *
@@ -363,6 +342,10 @@ rpcsvc_program_actor(rpcsvc_request_t *req)
goto err;
}
+ if (svc->xl->ctx->measure_latency) {
+ timespec_now(&req->begin);
+ }
+
req->ownthread = program->ownthread;
req->synctask = program->synctask;
@@ -1512,10 +1495,18 @@ rpcsvc_submit_generic(rpcsvc_request_t *req, struct iovec *proghdr,
size_t hdrlen = 0;
char new_iobref = 0;
rpcsvc_drc_globals_t *drc = NULL;
+ gf_latency_t *lat = NULL;
if ((!req) || (!req->trans))
return -1;
+ if (req->prog && req->begin.tv_sec) {
+ if ((req->procnum >= 0) && (req->procnum < req->prog->numactors)) {
+ timespec_now(&req->end);
+ lat = &req->prog->latencies[req->procnum];
+ gf_latency_update(lat, &req->begin, &req->end);
+ }
+ }
trans = req->trans;
for (i = 0; i < hdrcount; i++) {
@@ -1846,6 +1837,15 @@ rpcsvc_submit_message(rpcsvc_request_t *req, struct iovec *proghdr,
iobref);
}
+void
+rpcsvc_program_destroy(rpcsvc_program_t *program)
+{
+ if (program) {
+ GF_FREE(program->latencies);
+ GF_FREE(program);
+ }
+}
+
int
rpcsvc_program_unregister(rpcsvc_t *svc, rpcsvc_program_t *program)
{
@@ -1903,8 +1903,7 @@ rpcsvc_program_unregister(rpcsvc_t *svc, rpcsvc_program_t *program)
ret = 0;
out:
- if (prog)
- GF_FREE(prog);
+ rpcsvc_program_destroy(prog);
if (ret == -1) {
if (program) {
@@ -2289,6 +2288,11 @@ rpcsvc_program_register(rpcsvc_t *svc, rpcsvc_program_t *program,
}
memcpy(newprog, program, sizeof(*program));
+ newprog->latencies = gf_latency_new(program->numactors);
+ if (!newprog->latencies) {
+ rpcsvc_program_destroy(newprog);
+ goto out;
+ }
INIT_LIST_HEAD(&newprog->program);
pthread_mutexattr_init(&thr_attr);
@@ -3234,6 +3238,48 @@ out:
return ret;
}
+void
+rpcsvc_program_dump(rpcsvc_program_t *prog)
+{
+ char key_prefix[GF_DUMP_MAX_BUF_LEN];
+ char key[GF_DUMP_MAX_BUF_LEN];
+ int i;
+
+ snprintf(key_prefix, GF_DUMP_MAX_BUF_LEN, "%s", prog->progname);
+ gf_proc_dump_add_section("%s", key_prefix);
+
+ gf_proc_dump_build_key(key, key_prefix, "program-number");
+ gf_proc_dump_write(key, "%d", prog->prognum);
+
+ gf_proc_dump_build_key(key, key_prefix, "program-version");
+ gf_proc_dump_write(key, "%d", prog->progver);
+
+ strncat(key_prefix, ".latency",
+ sizeof(key_prefix) - strlen(key_prefix) - 1);
+
+ for (i = 0; i < prog->numactors; i++) {
+ gf_proc_dump_build_key(key, key_prefix, "%s", prog->actors[i].procname);
+ gf_latency_statedump_and_reset(key, &prog->latencies[i]);
+ }
+}
+
+void
+rpcsvc_statedump(rpcsvc_t *svc)
+{
+ rpcsvc_program_t *prog = NULL;
+ int ret = 0;
+ ret = pthread_rwlock_tryrdlock(&svc->rpclock);
+ if (ret)
+ return;
+ {
+ list_for_each_entry(prog, &svc->programs, program)
+ {
+ rpcsvc_program_dump(prog);
+ }
+ }
+ pthread_rwlock_unlock(&svc->rpclock);
+}
+
static rpcsvc_actor_t gluster_dump_actors[GF_DUMP_MAXVALUE] = {
[GF_DUMP_NULL] = {"NULL", NULL, NULL, GF_DUMP_NULL, DRC_NA, 0},
[GF_DUMP_DUMP] = {"DUMP", rpcsvc_dump, NULL, GF_DUMP_DUMP, DRC_NA, 0},
diff --git a/rpc/rpc-lib/src/rpcsvc.h b/rpc/rpc-lib/src/rpcsvc.h
index a88fc841b60..7b3030926c8 100644
--- a/rpc/rpc-lib/src/rpcsvc.h
+++ b/rpc/rpc-lib/src/rpcsvc.h
@@ -264,6 +264,8 @@ struct rpcsvc_request {
gf_boolean_t ownthread;
gf_boolean_t synctask;
+ struct timespec begin; /*req handling start time*/
+ struct timespec end; /*req handling end time*/
};
#define rpcsvc_request_program(req) ((rpcsvc_program_t *)((req)->prog))
@@ -419,6 +421,7 @@ struct rpcsvc_program {
supported by the system. */
/* Program specific state handed to actors */
void *private;
+ gf_latency_t *latencies; /*Tracks latency statistics for the rpc call*/
/* This upcall is provided by the program during registration.
* It is used to notify the program about events like connection being
@@ -457,7 +460,7 @@ struct rpcsvc_program {
gf_boolean_t alive;
gf_boolean_t synctask;
- char request_queue_status[EVENT_MAX_THREADS / 8 + 1];
+ unsigned long request_queue_status[EVENT_MAX_THREADS / __BITS_PER_LONG];
};
typedef struct rpcsvc_cbk_program {
@@ -686,4 +689,6 @@ rpcsvc_autoscale_threads(glusterfs_ctx_t *ctx, rpcsvc_t *rpc, int incr);
extern int
rpcsvc_destroy(rpcsvc_t *svc);
+void
+rpcsvc_statedump(rpcsvc_t *svc);
#endif
diff --git a/rpc/rpc-transport/socket/src/socket.c b/rpc/rpc-transport/socket/src/socket.c
index 5d65fc56de8..ed8b473be23 100644
--- a/rpc/rpc-transport/socket/src/socket.c
+++ b/rpc/rpc-transport/socket/src/socket.c
@@ -1154,7 +1154,7 @@ __socket_reset(rpc_transport_t *this)
SSL_free(priv->ssl_ssl);
priv->ssl_ssl = NULL;
}
- if (priv->use_ssl && priv->ssl_ctx) {
+ if (priv->ssl_ctx) {
SSL_CTX_free(priv->ssl_ctx);
priv->ssl_ctx = NULL;
}
@@ -4168,6 +4168,34 @@ static void __attribute__((destructor)) fini_openssl_mt(void)
ERR_free_strings();
}
+/* The function returns 0 if AES bit is enabled on the CPU */
+static int
+ssl_check_aes_bit(void)
+{
+ FILE *fp = fopen("/proc/cpuinfo", "r");
+ int ret = 1;
+ size_t len = 0;
+ char *line = NULL;
+ char *match = NULL;
+
+ GF_ASSERT(fp != NULL);
+
+ while (getline(&line, &len, fp) > 0) {
+ if (!strncmp(line, "flags", 5)) {
+ match = strstr(line, " aes");
+ if ((match != NULL) && ((match[4] == ' ') || (match[4] == 0))) {
+ ret = 0;
+ break;
+ }
+ }
+ }
+
+ free(line);
+ fclose(fp);
+
+ return ret;
+}
+
static int
ssl_setup_connection_params(rpc_transport_t *this)
{
@@ -4191,6 +4219,10 @@ ssl_setup_connection_params(rpc_transport_t *this)
return 0;
}
+ if (!ssl_check_aes_bit()) {
+ cipher_list = "AES128:" DEFAULT_CIPHER_LIST;
+ }
+
priv->ssl_own_cert = DEFAULT_CERT_PATH;
if (dict_get_str_sizen(this->options, SSL_OWN_CERT_OPT, &optstr) == 0) {
if (!priv->ssl_enabled) {
@@ -4615,7 +4647,7 @@ fini(rpc_transport_t *this)
SSL_free(priv->ssl_ssl);
priv->ssl_ssl = NULL;
}
- if (priv->use_ssl && priv->ssl_ctx) {
+ if (priv->ssl_ctx) {
SSL_CTX_free(priv->ssl_ctx);
priv->ssl_ctx = NULL;
}
diff --git a/rpc/xdr/src/glusterd1-xdr.x b/rpc/xdr/src/glusterd1-xdr.x
index 02ebec26c01..b631dea3502 100644
--- a/rpc/xdr/src/glusterd1-xdr.x
+++ b/rpc/xdr/src/glusterd1-xdr.x
@@ -202,6 +202,21 @@ struct gd1_mgmt_v3_commit_rsp {
string op_errstr<>;
} ;
+struct gd1_mgmt_v3_post_commit_req {
+ unsigned char uuid[16];
+ int op;
+ opaque dict<>;
+} ;
+
+struct gd1_mgmt_v3_post_commit_rsp {
+ unsigned char uuid[16];
+ int op;
+ int op_ret;
+ int op_errno;
+ opaque dict<>;
+ string op_errstr<>;
+} ;
+
struct gd1_mgmt_v3_post_val_req {
unsigned char uuid[16];
int op;
diff --git a/rpc/xdr/src/libgfxdr.sym b/rpc/xdr/src/libgfxdr.sym
index dd4ac8562bc..8fa0e0ddd8a 100644
--- a/rpc/xdr/src/libgfxdr.sym
+++ b/rpc/xdr/src/libgfxdr.sym
@@ -28,6 +28,8 @@ xdr_gd1_mgmt_v3_brick_op_req
xdr_gd1_mgmt_v3_brick_op_rsp
xdr_gd1_mgmt_v3_commit_req
xdr_gd1_mgmt_v3_commit_rsp
+xdr_gd1_mgmt_v3_post_commit_req
+xdr_gd1_mgmt_v3_post_commit_rsp
xdr_gd1_mgmt_v3_lock_req
xdr_gd1_mgmt_v3_lock_rsp
xdr_gd1_mgmt_v3_post_val_req
diff --git a/run-tests.sh b/run-tests.sh
index 2768adb5afd..e2a1655d8e0 100755
--- a/run-tests.sh
+++ b/run-tests.sh
@@ -328,6 +328,7 @@ function get_bug_list_for_disabled_test ()
function run_tests()
{
RES=0
+ FLAKY=''
FAILED=''
TESTS_NEEDED_RETRY=''
GENERATED_CORE=''
@@ -349,14 +350,15 @@ function run_tests()
timeout_cmd_exists="no"
fi
- for t in $(find ${regression_testsdir}/tests -name '*.t' \
- | LC_COLLATE=C sort) ; do
+ all_tests=($(find ${regression_testsdir}/tests -name '*.t' | sort))
+ all_tests_cnt=${#all_tests[@]}
+ for t in "${all_tests[@]}" ; do
old_cores=$(ls /*-*.core 2> /dev/null | wc -l)
total_tests=$((total_tests+1))
if match $t "$@" ; then
selected_tests=$((selected_tests+1))
echo
- echo $section_separator$section_separator
+ echo $section_separator "(${total_tests} / ${all_tests_cnt})" $section_separator
if [[ $(get_test_status $t) =~ "BAD_TEST" ]] && \
[[ $skip_bad_tests == "yes" ]]
then
@@ -432,9 +434,17 @@ function run_tests()
TESTS_NEEDED_RETRY="${TESTS_NEEDED_RETRY}${t} "
fi
+
+
if [ ${TMP_RES} -ne 0 ] ; then
- RES=${TMP_RES}
- FAILED="${FAILED}${t} "
+ if [[ "$t" == *"tests/000-flaky/"* ]]; then
+ FLAKY="${FLAKY}${t} "
+ echo "FAILURE -> SUCCESS: Flaky test"
+ TMP_RES=0
+ else
+ RES=${TMP_RES}
+ FAILED="${FAILED}${t} "
+ fi
fi
new_cores=$(ls /*-*.core 2> /dev/null | wc -l)
@@ -469,8 +479,10 @@ function run_tests()
echo "$key - ${ELAPSEDTIMEMAP["$key"]} second"
done | sort -rn -k3
- # Output the errors into a file
+ # initialize the output file
echo > "${result_output}"
+
+ # Output the errors into a file
if [ ${RES} -ne 0 ] ; then
FAILED=$( echo ${FAILED} | tr ' ' '\n' | sort -u )
FAILED_COUNT=$( echo -n "${FAILED}" | grep -c '^' )
@@ -483,7 +495,13 @@ function run_tests()
TESTS_NEEDED_RETRY=$( echo ${TESTS_NEEDED_RETRY} | tr ' ' '\n' | sort -u )
RETRY_COUNT=$( echo -n "${TESTS_NEEDED_RETRY}" | grep -c '^' )
if [ ${RETRY_COUNT} -ne 0 ] ; then
- echo -e "\n${RETRY_COUNT} test(s) needed retry \n${TESTS_NEEDED_RETRY}"
+ echo -e "\n${RETRY_COUNT} test(s) needed retry \n${TESTS_NEEDED_RETRY}" >> "${result_output}"
+ fi
+
+ FLAKY_TESTS_FAILED=$( echo ${FLAKY} | tr ' ' '\n' | sort -u )
+ RETRY_COUNT=$( echo -n "${FLAKY_TESTS_FAILED}" | grep -c '^' )
+ if [ ${RETRY_COUNT} -ne 0 ] ; then
+ echo -e "\n${RETRY_COUNT} flaky test(s) marked as success even though they failed \n${FLAKY_TESTS_FAILED}" >> "${result_output}"
fi
echo
@@ -512,8 +530,38 @@ function run_head_tests()
run_tests "$htests"
}
-function parse_args () {
- args=`getopt frcbkphHno:t: "$@"`
+function show_usage ()
+{
+ cat <<EOF
+Usage: $0 <opts> [<glob>|<bzid>]...
+
+Options:
+
+-f force
+-h skip tests altering from HEAD
+-H run only tests altering from HEAD
+-r retry failed tests
+-R do not retry failed tests
+-c dont't exit on failure
+-b don't skip bad tests
+-k don't skip known bugs
+-p don't keep logs from preceding runs
+-o OUTPUT
+-t TIMEOUT
+-n skip NFS tests
+--help
+EOF
+}
+
+usage="no"
+
+function parse_args ()
+{
+ args=`getopt -u -l help frRcbkphHno:t: "$@"`
+ if ! [ $? -eq 0 ]; then
+ show_usage
+ exit 1
+ fi
set -- $args
while [ $# -gt 0 ]; do
case "$1" in
@@ -521,6 +569,7 @@ function parse_args () {
-h) head="no" ;;
-H) head="only" ;;
-r) retry="yes" ;;
+ -R) retry="no" ;;
-c) exit_on_failure="no" ;;
-b) skip_bad_tests="no" ;;
-k) skip_known_bugs="no" ;;
@@ -528,6 +577,7 @@ function parse_args () {
-o) result_output="$2"; shift;;
-t) run_timeout="$2"; shift;;
-n) nfs_tests="no";;
+ --help) usage="yes" ;;
--) shift; break;;
esac
shift
@@ -542,6 +592,10 @@ echo
# Get user options
parse_args "$@"
+if [ x"$usage" == x"yes" ]; then
+ show_usage
+ exit 0
+fi
# Make sure we're running as the root user
check_user
diff --git a/tests/00-geo-rep/00-georep-verify-non-root-setup.t b/tests/00-geo-rep/00-georep-verify-non-root-setup.t
index 27ed82732c9..a55fd3e5e6a 100644
--- a/tests/00-geo-rep/00-georep-verify-non-root-setup.t
+++ b/tests/00-geo-rep/00-georep-verify-non-root-setup.t
@@ -5,7 +5,7 @@
. $(dirname $0)/../geo-rep.rc
. $(dirname $0)/../env.rc
-SCRIPT_TIMEOUT=500
+SCRIPT_TIMEOUT=600
### Basic Non-root geo-rep setup test with Distribute Replicate volumes
@@ -144,12 +144,15 @@ TEST pidof glusterd;
TEST $CLI volume create $META_VOL replica 3 $H0:$B0/${META_VOL}{1,2,3};
TEST $CLI volume start $META_VOL
TEST mkdir -p $META_MNT
+EXPECT_WITHIN ${PROCESS_UP_TIMEOUT} "3" brick_count ${META_VOL}
TEST glusterfs -s $H0 --volfile-id $META_VOL $META_MNT
##Mount master
+EXPECT_WITHIN ${PROCESS_UP_TIMEOUT} "4" brick_count $GMV0
TEST glusterfs -s $H0 --volfile-id $GMV0 $M0
##Mount slave
+EXPECT_WITHIN ${PROCESS_UP_TIMEOUT} "4" brick_count $GSV0
TEST glusterfs -s $H0 --volfile-id $GSV0 $M1
## Check status of mount-broker
@@ -189,6 +192,8 @@ TEST gluster-georep-sshkey generate
TEST $GEOREP_CLI $master $slave_url create push-pem
+#check for session creation
+EXPECT_WITHIN $GEO_REP_TIMEOUT 4 check_status_non_root "Created"
#Config gluster-command-dir
TEST $GEOREP_CLI $master $slave_url config gluster-command-dir ${GLUSTER_CMD_DIR}
@@ -240,21 +245,25 @@ TEST ! $GEOREP_CLI $master $slave_url config ssh-port 22a
TEST $GEOREP_CLI $master $slave config ssh-port 22
#Hybrid directory rename test BZ#1763439
+
TEST $GEOREP_CLI $master $slave_url config change_detector xsync
-mkdir ${master_mnt}/dir1
-mkdir ${master_mnt}/dir1/dir2
-mkdir ${master_mnt}/dir1/dir3
-mkdir ${master_mnt}/hybrid_d1
+#verify master and slave mount
-EXPECT_WITHIN $GEO_REP_TIMEOUT 0 directory_ok ${slave_mnt}/hybrid_d1
-EXPECT_WITHIN $GEO_REP_TIMEOUT 0 directory_ok ${slave_mnt}/dir1
-EXPECT_WITHIN $GEO_REP_TIMEOUT 0 directory_ok ${slave_mnt}/dir1/dir2
-EXPECT_WITHIN $GEO_REP_TIMEOUT 0 directory_ok ${slave_mnt}/dir1/dir3
+EXPECT_WITHIN $CHECK_MOUNT_TIMEOUT "^1$" check_mounted ${master_mnt}
+EXPECT_WITHIN $CHECK_MOUNT_TIMEOUT "^1$" check_mounted ${slave_mnt}
+
+#Create test data for hybrid crawl
+TEST mkdir ${master_mnt}/dir1
+TEST mkdir ${master_mnt}/dir1/dir2
+TEST mkdir ${master_mnt}/dir1/dir3
+TEST mkdir ${master_mnt}/hybrid_d1
mv ${master_mnt}/hybrid_d1 ${master_mnt}/hybrid_rn_d1
mv ${master_mnt}/dir1/dir2 ${master_mnt}/rn_dir2
mv ${master_mnt}/dir1/dir3 ${master_mnt}/dir1/rn_dir3
+#Verify hybrid crawl data on slave
+EXPECT_WITHIN $GEO_REP_TIMEOUT 0 directory_ok ${slave_mnt}/dir1
EXPECT_WITHIN $GEO_REP_TIMEOUT 0 directory_ok ${slave_mnt}/hybrid_rn_d1
EXPECT_WITHIN $GEO_REP_TIMEOUT 0 directory_ok ${slave_mnt}/rn_dir2
EXPECT_WITHIN $GEO_REP_TIMEOUT 0 directory_ok ${slave_mnt}/dir1/rn_dir3
diff --git a/tests/00-geo-rep/00-georep-verify-setup.t b/tests/00-geo-rep/00-georep-verify-setup.t
index cdfccd35db5..0d46c04102d 100644
--- a/tests/00-geo-rep/00-georep-verify-setup.t
+++ b/tests/00-geo-rep/00-georep-verify-setup.t
@@ -5,7 +5,8 @@
. $(dirname $0)/../geo-rep.rc
. $(dirname $0)/../env.rc
-SCRIPT_TIMEOUT=300
+SCRIPT_TIMEOUT=400
+GEO_REP_TIMEOUT=200
##Cleanup and start glusterd
cleanup;
diff --git a/tests/00-geo-rep/georep-upgrade.t b/tests/00-geo-rep/georep-upgrade.t
new file mode 100644
index 00000000000..7523068ed50
--- /dev/null
+++ b/tests/00-geo-rep/georep-upgrade.t
@@ -0,0 +1,79 @@
+#!/bin/bash
+
+. $(dirname $0)/../include.rc
+
+SCRIPT_TIMEOUT=500
+
+###############################################################################################
+#Before upgrade
+###############################################################################################
+brick=/bricks/brick1
+epoch1=$(date '+%s')
+sleep 1
+epoch2=$(date '+%s')
+mkdir -p /bricks/brick1/.glusterfs/changelogs/htime
+mkdir -p /bricks/brick1/.glusterfs/changelogs
+
+#multiple htime files(changelog enable/disable scenario)
+TEST touch /bricks/brick1/.glusterfs/changelogs/htime/HTIME.$epoch1
+TEST touch /bricks/brick1/.glusterfs/changelogs/htime/HTIME.$epoch2
+
+#changelog files
+TEST touch /bricks/brick1/.glusterfs/changelogs/CHANGELOG.$epoch1
+TEST touch /bricks/brick1/.glusterfs/changelogs/CHANGELOG.$epoch2
+
+htime_file1=/bricks/brick1/.glusterfs/changelogs/htime/HTIME.$epoch1
+htime_file2=/bricks/brick1/.glusterfs/changelogs/htime/HTIME.$epoch2
+
+#data inside htime files before upgrade
+data1=/bricks/brick1/.glusterfs/changelogs/CHANGELOG.$epoch1
+data2=/bricks/brick1/.glusterfs/changelogs/CHANGELOG.$epoch2
+
+#data inside htime files after upgrade
+updated_data1=/bricks/brick1/.glusterfs/changelogs/`echo $(date '+%Y/%m/%d')`/CHANGELOG.$epoch1
+updated_data2=/bricks/brick1/.glusterfs/changelogs/`echo $(date '+%Y/%m/%d')`/CHANGELOG.$epoch2
+
+echo -n $data1>$htime_file1
+echo -n $data2>$htime_file2
+
+echo "Before upgrade:"
+EXPECT '1' echo $(grep $data1 $htime_file1 | wc -l)
+EXPECT '1' echo $(grep $data2 $htime_file2 | wc -l)
+
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/CHANGELOG.$epoch1 | wc -l)
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/htime/HTIME.$epoch1 | wc -l)
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/CHANGELOG.$epoch2 | wc -l)
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/htime/HTIME.$epoch2 | wc -l)
+###############################################################################################
+#Upgrade
+###############################################################################################
+### This needed to be fixed as this very vague finding a file with name in '/'
+### multiple file with same name can exist
+### for temp fix picking only 1st result
+TEST upgrade_script=$(find / -type f -name glusterfs-georep-upgrade.py -print | head -n 1)
+TEST python3 $upgrade_script $brick
+
+###############################################################################################
+#After upgrade
+###############################################################################################
+echo "After upgrade:"
+EXPECT '1' echo $(grep $updated_data1 $htime_file1 | wc -l)
+EXPECT '1' echo $(grep $updated_data2 $htime_file2 | wc -l)
+
+#Check directory structure inside changelogs
+TEST ! ls /bricks/brick1/.glusterfs/changelogs/CHANGELOG.$epoch1
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/htime/HTIME.$epoch1 | wc -l)
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/htime/HTIME.$epoch1.bak | wc -l)
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/`echo $(date '+%Y')` | wc -l)
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/`echo $(date '+%Y/%m')` | wc -l)
+EXPECT '2' echo $(ls /bricks/brick1/.glusterfs/changelogs/`echo $(date '+%Y/%m/%d')` | wc -l)
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/`echo $(date '+%Y/%m/%d')`/CHANGELOG.$epoch1 | wc -l)
+
+TEST ! ls /bricks/brick1/.glusterfs/changelogs/CHANGELOG.$epoch2
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/htime/HTIME.$epoch2 | wc -l)
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/htime/HTIME.$epoch2.bak | wc -l)
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/`echo $(date '+%Y')` | wc -l)
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/`echo $(date '+%Y/%m')`| wc -l)
+EXPECT '1' echo $(ls /bricks/brick1/.glusterfs/changelogs/`echo $(date '+%Y/%m/%d')`/CHANGELOG.$epoch2 | wc -l)
+
+TEST rm -rf /bricks
diff --git a/tests/basic/afr/split-brain-favorite-child-policy.t b/tests/000-flaky/basic_afr_split-brain-favorite-child-policy.t
index c268c125610..77d82a4996f 100644
--- a/tests/basic/afr/split-brain-favorite-child-policy.t
+++ b/tests/000-flaky/basic_afr_split-brain-favorite-child-policy.t
@@ -1,8 +1,8 @@
#!/bin/bash
#Test the split-brain resolution CLI commands.
-. $(dirname $0)/../../include.rc
-. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
cleanup;
diff --git a/tests/basic/changelog/changelog-snapshot.t b/tests/000-flaky/basic_changelog_changelog-snapshot.t
index 7742db48cdd..f6cd0b04d47 100644
--- a/tests/basic/changelog/changelog-snapshot.t
+++ b/tests/000-flaky/basic_changelog_changelog-snapshot.t
@@ -1,7 +1,7 @@
#!/bin/bash
-. $(dirname $0)/../../include.rc
-. $(dirname $0)/../../snapshot.rc
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../snapshot.rc
cleanup;
ROLLOVER_TIME=3
diff --git a/tests/basic/distribute/rebal-all-nodes-migrate.t b/tests/000-flaky/basic_distribute_rebal-all-nodes-migrate.t
index acc4ffefecc..eb5d3305ac1 100644
--- a/tests/basic/distribute/rebal-all-nodes-migrate.t
+++ b/tests/000-flaky/basic_distribute_rebal-all-nodes-migrate.t
@@ -1,8 +1,8 @@
#!/bin/bash
-. $(dirname $0)/../../include.rc
-. $(dirname $0)/../../cluster.rc
-. $(dirname $0)/../../dht.rc
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../cluster.rc
+. $(dirname $0)/../dht.rc
# Check if every single rebalance process migrated some files
@@ -10,11 +10,9 @@
function cluster_rebal_all_nodes_migrated_files {
val=0
a=$($CLI_1 volume rebalance $V0 status | grep "completed" | awk '{print $2}');
-# echo $a
b=($a)
for i in "${b[@]}"
do
-# echo "$i";
if [ "$i" -eq "0" ]; then
echo "false";
val=1;
diff --git a/tests/basic/ec/ec-quorum-count-partial-failure.t b/tests/000-flaky/basic_ec_ec-quorum-count-partial-failure.t
index 79f5825ae10..42808ce0c0e 100755..100644
--- a/tests/basic/ec/ec-quorum-count-partial-failure.t
+++ b/tests/000-flaky/basic_ec_ec-quorum-count-partial-failure.t
@@ -1,7 +1,7 @@
#!/bin/bash
-. $(dirname $0)/../../include.rc
-. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
#This test checks that partial failure of fop results in main fop failure only
cleanup;
diff --git a/tests/basic/mount-nfs-auth.t b/tests/000-flaky/basic_mount-nfs-auth.t
index 3d4a9cff00b..3d4a9cff00b 100755..100644
--- a/tests/basic/mount-nfs-auth.t
+++ b/tests/000-flaky/basic_mount-nfs-auth.t
diff --git a/tests/bugs/core/multiplex-limit-issue-151.t b/tests/000-flaky/bugs_core_multiplex-limit-issue-151.t
index dc9013061b0..5a88f97d726 100644
--- a/tests/bugs/core/multiplex-limit-issue-151.t
+++ b/tests/000-flaky/bugs_core_multiplex-limit-issue-151.t
@@ -1,8 +1,8 @@
#!/bin/bash
-. $(dirname $0)/../../include.rc
-. $(dirname $0)/../../traps.rc
-. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../traps.rc
+. $(dirname $0)/../volume.rc
function count_up_bricks {
$CLI --xml volume status all | grep '<status>1' | wc -l
diff --git a/tests/bugs/distribute/bug-1117851.t b/tests/000-flaky/bugs_distribute_bug-1117851.t
index 62cb6b66ab4..5980bf2fd4b 100755..100644
--- a/tests/bugs/distribute/bug-1117851.t
+++ b/tests/000-flaky/bugs_distribute_bug-1117851.t
@@ -2,8 +2,8 @@
SCRIPT_TIMEOUT=250
-. $(dirname $0)/../../include.rc
-. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
create_files () {
for i in {1..1000}; do
diff --git a/tests/bugs/distribute/bug-1122443.t b/tests/000-flaky/bugs_distribute_bug-1122443.t
index 906be7072bd..abd37082b33 100644
--- a/tests/bugs/distribute/bug-1122443.t
+++ b/tests/000-flaky/bugs_distribute_bug-1122443.t
@@ -1,8 +1,8 @@
#!/bin/bash
-. $(dirname $0)/../../include.rc
-. $(dirname $0)/../../volume.rc
-. $(dirname $0)/../../dht.rc
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
+. $(dirname $0)/../dht.rc
make_files() {
mkdir $1 && \
@@ -42,8 +42,8 @@ TEST glusterfs -s $H0 --volfile-id $V0 $M0
TEST make_files $M0/subdir
# Get mtime before migration
-BEFORE="$(stat -c %n:%Y $M0/subdir/* | tr '\n' ',')"
-
+BEFORE="$(stat -c %n:%Y $M0/subdir/* | sort | tr '\n' ',')"
+echo $BEFORE
# Migrate brick
TEST $CLI volume add-brick $V0 $H0:$B0/${V0}1
TEST $CLI volume remove-brick $V0 $H0:$B0/${V0}0 start
@@ -51,9 +51,10 @@ EXPECT_WITHIN $REBALANCE_TIMEOUT "completed" remove_brick_status_completed_field
TEST $CLI volume remove-brick $V0 $H0:$B0/${V0}0 commit
# Get mtime after migration
-EXPECT_WITHIN 5 RECONNECTED bug_1113050_workaround $M0/subdir/*
-AFTER="$(stat -c %n:%Y $M0/subdir/* | tr '\n' ',')"
-
+EXPECT_WITHIN 30 RECONNECTED bug_1113050_workaround $M0/subdir/symlink
+sleep 3
+AFTER="$(stat -c %n:%Y $M0/subdir/* | sort | tr '\n' ',')"
+echo $AFTER
# Check if mtime is unchanged
TEST [ "$AFTER" == "$BEFORE" ]
diff --git a/tests/bugs/glusterd/bug-857330/common.rc b/tests/000-flaky/bugs_glusterd_bug-857330/common.rc
index d0aa4b1a640..bd122eff18c 100644
--- a/tests/bugs/glusterd/bug-857330/common.rc
+++ b/tests/000-flaky/bugs_glusterd_bug-857330/common.rc
@@ -1,4 +1,4 @@
-. $(dirname $0)/../../../include.rc
+. $(dirname $0)/../../include.rc
UUID_REGEX='[a-f0-9]{8}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{12}'
diff --git a/tests/bugs/glusterd/bug-857330/normal.t b/tests/000-flaky/bugs_glusterd_bug-857330/normal.t
index ad0c8844fae..6c1cf54ec3c 100755
--- a/tests/bugs/glusterd/bug-857330/normal.t
+++ b/tests/000-flaky/bugs_glusterd_bug-857330/normal.t
@@ -1,7 +1,7 @@
#!/bin/bash
. $(dirname $0)/common.rc
-. $(dirname $0)/../../../volume.rc
+. $(dirname $0)/../../volume.rc
cleanup;
TEST glusterd
@@ -14,7 +14,7 @@ TEST $CLI volume start $V0;
TEST glusterfs -s $H0 --volfile-id=$V0 $M0;
-TEST $PYTHON $(dirname $0)/../../../utils/create-files.py \
+TEST $PYTHON $(dirname $0)/../../utils/create-files.py \
--multi -b 10 -d 10 -n 10 $M0;
EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0
diff --git a/tests/bugs/glusterd/bug-857330/xml.t b/tests/000-flaky/bugs_glusterd_bug-857330/xml.t
index 8383d2a0711..11785adacdb 100755
--- a/tests/bugs/glusterd/bug-857330/xml.t
+++ b/tests/000-flaky/bugs_glusterd_bug-857330/xml.t
@@ -1,7 +1,7 @@
#!/bin/bash
. $(dirname $0)/common.rc
-. $(dirname $0)/../../../volume.rc
+. $(dirname $0)/../../volume.rc
cleanup;
@@ -15,7 +15,7 @@ TEST $CLI volume start $V0;
TEST glusterfs -s $H0 --volfile-id=$V0 $M0;
-TEST $PYTHON $(dirname $0)/../../../utils/create-files.py \
+TEST $PYTHON $(dirname $0)/../../utils/create-files.py \
--multi -b 10 -d 10 -n 10 $M0;
EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0
diff --git a/tests/bugs/glusterd/quorum-value-check.t b/tests/000-flaky/bugs_glusterd_quorum-value-check.t
index c701f08bbd5..a431b8c4fd4 100755..100644
--- a/tests/bugs/glusterd/quorum-value-check.t
+++ b/tests/000-flaky/bugs_glusterd_quorum-value-check.t
@@ -1,7 +1,7 @@
#!/bin/bash
-. $(dirname $0)/../../include.rc
-. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
#G_TESTDEF_TEST_STATUS_CENTOS6=NFS_TEST
diff --git a/tests/bugs/nfs/bug-1116503.t b/tests/000-flaky/bugs_nfs_bug-1116503.t
index dd3998df150..fc50021acc7 100644
--- a/tests/bugs/nfs/bug-1116503.t
+++ b/tests/000-flaky/bugs_nfs_bug-1116503.t
@@ -3,9 +3,9 @@
# Verify that mounting NFS over UDP (MOUNT service only) works.
#
-. $(dirname $0)/../../include.rc
-. $(dirname $0)/../../volume.rc
-. $(dirname $0)/../../nfs.rc
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
+. $(dirname $0)/../nfs.rc
#G_TESTDEF_TEST_STATUS_CENTOS6=NFS_TEST
diff --git a/tests/features/lock-migration/lkmigration-set-option.t b/tests/000-flaky/features_lock-migration_lkmigration-set-option.t
index 4340438591f..1327ef3579f 100644
--- a/tests/features/lock-migration/lkmigration-set-option.t
+++ b/tests/000-flaky/features_lock-migration_lkmigration-set-option.t
@@ -1,7 +1,7 @@
#!/bin/bash
# Test to check
-. $(dirname $0)/../../include.rc
-. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
#Check lock-migration set option sanity
cleanup;
diff --git a/tests/afr.rc b/tests/afr.rc
index 5fc7fa1898d..241789903ba 100644
--- a/tests/afr.rc
+++ b/tests/afr.rc
@@ -115,3 +115,9 @@ function afr_private_key_value()
#xargs at the end will strip leading spaces
grep -E "^${key} = " $m/.meta/graphs/active/${v}-replicate-${replica_id}/private | cut -f2 -d'=' | xargs
}
+
+function afr_anon_entry_count()
+{
+ local b=$1
+ ls $b/.glusterfs-anonymous-inode* | wc -l
+}
diff --git a/tests/basic/afr/afr-anon-inode-no-quorum.t b/tests/basic/afr/afr-anon-inode-no-quorum.t
new file mode 100644
index 00000000000..896ba0c9b2c
--- /dev/null
+++ b/tests/basic/afr/afr-anon-inode-no-quorum.t
@@ -0,0 +1,63 @@
+#!/bin/bash
+
+#Test that anon-inode entry is not cleaned up as long as there exists at least
+#one valid entry
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../afr.rc
+
+cleanup;
+
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{0,1}
+TEST $CLI volume heal $V0 disable
+TEST $CLI volume set $V0 performance.write-behind off
+TEST $CLI volume set $V0 performance.read-ahead off
+TEST $CLI volume set $V0 performance.readdir-ahead off
+TEST $CLI volume set $V0 performance.open-behind off
+TEST $CLI volume set $V0 performance.stat-prefetch off
+TEST $CLI volume set $V0 performance.io-cache off
+TEST $CLI volume set $V0 performance.quick-read off
+TEST $CLI volume set $V0 cluster.entry-self-heal off
+TEST $CLI volume start $V0
+
+TEST $GFS --volfile-id=/$V0 --volfile-server=$H0 $M0
+
+TEST touch $M0/a $M0/b
+
+gfid_a=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/${V0}0/a))
+gfid_b=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/${V0}0/b))
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST mv $M0/a $M0/a-new
+TEST mv $M0/b $M0/b-new
+
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status $V0 0
+TEST ! ls $M0/a
+TEST ! ls $M0/b
+anon_inode_name=$(ls -a $B0/${V0}0 | grep glusterfs-anonymous-inode)
+TEST stat $B0/${V0}0/$anon_inode_name/$gfid_a
+TEST stat $B0/${V0}0/$anon_inode_name/$gfid_b
+#Make sure index heal doesn't happen after enabling heal
+TEST setfattr -x trusted.afr.$V0-client-0 $B0/${V0}1
+TEST rm -f $B0/${V0}1/.glusterfs/indices/xattrop/*
+TEST $CLI volume heal $V0 enable
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" glustershd_up_status
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 1
+TEST $CLI volume heal $V0
+#Allow time for a scan
+sleep 5
+TEST stat $B0/${V0}0/$anon_inode_name/$gfid_a
+TEST stat $B0/${V0}0/$anon_inode_name/$gfid_b
+inum_b=$(STAT_INO $B0/${V0}0/$anon_inode_name/$gfid_b)
+TEST rm -f $M0/a-new
+TEST stat $M0/b-new
+
+TEST $CLI volume heal $V0
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" afr_anon_entry_count $B0/${V0}0
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" afr_anon_entry_count $B0/${V0}1
+EXPECT "$inum_b" STAT_INO $B0/${V0}0/b-new
+
+cleanup
diff --git a/tests/basic/afr/afr-anon-inode.t b/tests/basic/afr/afr-anon-inode.t
new file mode 100644
index 00000000000..f4cf37a2fa0
--- /dev/null
+++ b/tests/basic/afr/afr-anon-inode.t
@@ -0,0 +1,114 @@
+#!/bin/bash
+#Tests that afr-anon-inode test cases work fine as expected
+#These are cases where in entry-heal/name-heal we dont know entry for an inode
+#so these inodes are kept in a special directory
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../afr.rc
+
+cleanup;
+
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 replica 3 $H0:$B0/${V0}{0..2}
+TEST $CLI volume set $V0 performance.quick-read off
+TEST $CLI volume set $V0 performance.io-cache off
+TEST $CLI volume set $V0 performance.write-behind off
+TEST $CLI volume set $V0 performance.stat-prefetch off
+TEST $CLI volume set $V0 performance.read-ahead off
+TEST $CLI volume set $V0 performance.open-behind off
+TEST $CLI volume start $V0
+TEST $GFS --volfile-id=$V0 --volfile-server=$H0 $M0;
+EXPECT "^1$" afr_private_key_value $V0 $M0 0 "use-anonymous-inode"
+TEST $CLI volume set $V0 cluster.use-anonymous-inode no
+EXPECT_WITHIN $CONFIG_UPDATE_TIMEOUT "^0$" afr_private_key_value $V0 $M0 0 "use-anonymous-inode"
+TEST $CLI volume set $V0 cluster.use-anonymous-inode yes
+EXPECT_WITHIN $CONFIG_UPDATE_TIMEOUT "^1$" afr_private_key_value $V0 $M0 0 "use-anonymous-inode"
+TEST mkdir -p $M0/d1/b $M0/d2/a
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST mv $M0/d2/a $M0/d1
+TEST mv $M0/d1/b $M0/d2
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" get_pending_heal_count $V0
+anon_inode_name=$(ls -a $B0/${V0}0 | grep glusterfs-anonymous-inode)
+TEST [[ -d $B0/${V0}1/$anon_inode_name ]]
+TEST [[ -d $B0/${V0}2/$anon_inode_name ]]
+anon_gfid=$(gf_get_gfid_xattr $B0/${V0}0/$anon_inode_name)
+EXPECT "$anon_gfid" gf_get_gfid_xattr $B0/${V0}1/$anon_inode_name
+EXPECT "$anon_gfid" gf_get_gfid_xattr $B0/${V0}2/$anon_inode_name
+
+TEST ! ls $M0/$anon_inode_name
+EXPECT "^4$" echo $(ls -a $M0 | wc -l)
+
+#Test purging code path by shd
+TEST $CLI volume heal $V0 disable
+TEST mkdir $M0/l0 $M0/l1 $M0/l2
+TEST touch $M0/del-file $M0/del-file-nolink $M0/l0/file
+TEST ln $M0/del-file $M0/del-file-link
+TEST ln $M0/l0/file $M0/l1/file-link1
+TEST ln $M0/l0/file $M0/l2/file-link2
+TEST mkdir -p $M0/del-recursive-dir/d1
+
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST rm -f $M0/del-file $M0/del-file-nolink
+TEST rm -rf $M0/del-recursive-dir
+TEST mv $M0/d1/a $M0/d2
+TEST mv $M0/l0/file $M0/l0/renamed-file
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "^1$" afr_child_up_status $V0 0
+
+nolink_gfid=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/${V0}0/del-file-nolink))
+link_gfid=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/${V0}0/del-file))
+dir_gfid=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/${V0}0/del-recursive-dir))
+rename_dir_gfid=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/${V0}0/d1/a))
+rename_file_gfid=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/${V0}0/l0/file))
+TEST ! stat $M0/del-file
+TEST stat $B0/${V0}0/$anon_inode_name/$link_gfid
+TEST ! stat $M0/del-file-nolink
+TEST ! stat $B0/${V0}0/$anon_inode_name/$nolink_gfid
+TEST ! stat $M0/del-recursive-dir
+TEST stat $B0/${V0}0/$anon_inode_name/$dir_gfid
+TEST ! stat $M0/d1/a
+TEST stat $B0/${V0}0/$anon_inode_name/$rename_dir_gfid
+TEST ! stat $M0/l0/file
+TEST stat $B0/${V0}0/$anon_inode_name/$rename_file_gfid
+
+TEST kill_brick $V0 $H0 $B0/${V0}1
+TEST mv $M0/l1/file-link1 $M0/l1/renamed-file-link1
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "^1$" afr_child_up_status $V0 1
+TEST ! stat $M0/l1/file-link1
+TEST stat $B0/${V0}1/$anon_inode_name/$rename_file_gfid
+
+TEST kill_brick $V0 $H0 $B0/${V0}2
+TEST mv $M0/l2/file-link2 $M0/l2/renamed-file-link2
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "^1$" afr_child_up_status $V0 2
+TEST ! stat $M0/l2/file-link2
+TEST stat $B0/${V0}2/$anon_inode_name/$rename_file_gfid
+
+#Simulate only anon-inodes present in all bricks
+TEST rm -f $M0/l0/renamed-file $M0/l1/renamed-file-link1 $M0/l2/renamed-file-link2
+
+#Test that shd doesn't cleanup anon-inodes when some bricks are down
+TEST kill_brick $V0 $H0 $B0/${V0}1
+TEST $CLI volume heal $V0 enable
+$CLI volume heal $V0
+sleep 5 #Allow time for completion of one scan
+TEST stat $B0/${V0}0/$anon_inode_name/$link_gfid
+TEST stat $B0/${V0}0/$anon_inode_name/$rename_dir_gfid
+TEST stat $B0/${V0}0/$anon_inode_name/$dir_gfid
+rename_dir_inum=$(STAT_INO $B0/${V0}0/$anon_inode_name/$rename_dir_gfid)
+
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "^1$" afr_child_up_status $V0 1
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" get_pending_heal_count $V0
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" afr_anon_entry_count $B0/${V0}0
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" afr_anon_entry_count $B0/${V0}1
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" afr_anon_entry_count $B0/${V0}2
+
+#Test that rename indeed happened instead of rmdir/mkdir
+renamed_dir_inum=$(STAT_INO $B0/${V0}0/d2/a)
+EXPECT "$rename_dir_inum" echo $renamed_dir_inum
+cleanup;
diff --git a/tests/basic/afr/entry-self-heal-anon-dir-off.t b/tests/basic/afr/entry-self-heal-anon-dir-off.t
new file mode 100644
index 00000000000..7bb6ee14193
--- /dev/null
+++ b/tests/basic/afr/entry-self-heal-anon-dir-off.t
@@ -0,0 +1,459 @@
+#!/bin/bash
+
+#This file checks if missing entry self-heal and entry self-heal are working
+#as expected.
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../afr.rc
+
+cleanup;
+
+function get_file_type {
+ stat -c "%a:%F:%g:%t:%T:%u" $1
+}
+
+function diff_dirs {
+ diff <(ls $1 | sort) <(ls $2 | sort)
+}
+
+function heal_status {
+ local f1_path="${1}/${3}"
+ local f2_path="${2}/${3}"
+ local insync=""
+ diff_dirs $f1_path $f2_path
+ if [ $? -eq 0 ];
+ then
+ insync="Y"
+ else
+ insync="N"
+ fi
+ local xattr11=$(get_hex_xattr trusted.afr.$V0-client-0 $f1_path)
+ local xattr12=$(get_hex_xattr trusted.afr.$V0-client-1 $f1_path)
+ local xattr21=$(get_hex_xattr trusted.afr.$V0-client-0 $f2_path)
+ local xattr22=$(get_hex_xattr trusted.afr.$V0-client-1 $f2_path)
+ local dirty1=$(get_hex_xattr trusted.afr.dirty $f1_path)
+ local dirty2=$(get_hex_xattr trusted.afr.dirty $f2_path)
+ if [ -z $xattr11 ]; then xattr11="000000000000000000000000"; fi
+ if [ -z $xattr12 ]; then xattr12="000000000000000000000000"; fi
+ if [ -z $xattr21 ]; then xattr21="000000000000000000000000"; fi
+ if [ -z $xattr22 ]; then xattr22="000000000000000000000000"; fi
+ if [ -z $dirty1 ]; then dirty1="000000000000000000000000"; fi
+ if [ -z $dirty2 ]; then dirty2="000000000000000000000000"; fi
+ echo ${insync}${xattr11}${xattr12}${xattr21}${xattr22}${dirty1}${dirty2}
+}
+
+function is_heal_done {
+ local zero_xattr="000000000000000000000000"
+ if [ "$(heal_status $@)" == "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" ];
+ then
+ echo "Y"
+ else
+ echo "N"
+ fi
+}
+
+function print_pending_heals {
+ local result=":"
+ for i in "$@";
+ do
+ if [ "N" == $(is_heal_done $B0/${V0}0 $B0/${V0}1 $i) ];
+ then
+ result="$result:$i"
+ fi
+ done
+#To prevent any match for EXPECT_WITHIN, print a char non-existent in file-names
+ if [ $result == ":" ]; then result="~"; fi
+ echo $result
+}
+
+zero_xattr="000000000000000000000000"
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{0,1}
+TEST $CLI volume heal $V0 disable
+TEST $CLI volume set $V0 cluster.use-anonymous-inode off
+TEST $CLI volume set $V0 performance.write-behind off
+TEST $CLI volume set $V0 performance.read-ahead off
+TEST $CLI volume set $V0 performance.readdir-ahead off
+TEST $CLI volume set $V0 performance.open-behind off
+TEST $CLI volume set $V0 performance.stat-prefetch off
+TEST $CLI volume set $V0 performance.io-cache off
+TEST $CLI volume set $V0 performance.quick-read off
+TEST $CLI volume set $V0 cluster.data-self-heal on
+TEST $CLI volume set $V0 cluster.metadata-self-heal on
+TEST $CLI volume set $V0 cluster.entry-self-heal on
+TEST $CLI volume start $V0
+
+TEST $GFS --volfile-id=/$V0 --volfile-server=$H0 --use-readdirp=no $M0
+cd $M0
+#_me_ is dir on which missing entry self-heal happens, _heal is where dir self-heal happens
+#spb is split-brain, fool is all fool
+
+#source_self_accusing means there exists source and a sink which self-accuses.
+#This simulates failures where fops failed on the bricks without it going down.
+#Something like EACCESS/EDQUOT etc
+
+TEST mkdir spb_heal spb spb_me_heal spb_me fool_heal fool_me v1_fool_heal v1_fool_me source_creations_heal source_deletions_heal source_creations_me source_deletions_me v1_dirty_me v1_dirty_heal source_self_accusing
+TEST mkfifo source_deletions_heal/fifo
+TEST mknod source_deletions_heal/block b 4 5
+TEST mknod source_deletions_heal/char c 1 5
+TEST touch source_deletions_heal/file
+TEST ln -s source_deletions_heal/file source_deletions_heal/slink
+TEST mkdir source_deletions_heal/dir1
+TEST mkdir source_deletions_heal/dir1/dir2
+
+TEST mkfifo source_deletions_me/fifo
+TEST mknod source_deletions_me/block b 4 5
+TEST mknod source_deletions_me/char c 1 5
+TEST touch source_deletions_me/file
+TEST ln -s source_deletions_me/file source_deletions_me/slink
+TEST mkdir source_deletions_me/dir1
+TEST mkdir source_deletions_me/dir1/dir2
+
+TEST mkfifo source_self_accusing/fifo
+TEST mknod source_self_accusing/block b 4 5
+TEST mknod source_self_accusing/char c 1 5
+TEST touch source_self_accusing/file
+TEST ln -s source_self_accusing/file source_self_accusing/slink
+TEST mkdir source_self_accusing/dir1
+TEST mkdir source_self_accusing/dir1/dir2
+
+TEST kill_brick $V0 $H0 $B0/${V0}0
+
+TEST touch spb_heal/0 spb/0 spb_me_heal/0 spb_me/0 fool_heal/0 fool_me/0 v1_fool_heal/0 v1_fool_me/0 v1_dirty_heal/0 v1_dirty_me/0
+TEST rm -rf source_deletions_heal/fifo source_deletions_heal/block source_deletions_heal/char source_deletions_heal/file source_deletions_heal/slink source_deletions_heal/dir1
+TEST rm -rf source_deletions_me/fifo source_deletions_me/block source_deletions_me/char source_deletions_me/file source_deletions_me/slink source_deletions_me/dir1
+TEST rm -rf source_self_accusing/fifo source_self_accusing/block source_self_accusing/char source_self_accusing/file source_self_accusing/slink source_self_accusing/dir1
+
+#Test that the files are deleted
+TEST ! stat $B0/${V0}1/source_deletions_heal/fifo
+TEST ! stat $B0/${V0}1/source_deletions_heal/block
+TEST ! stat $B0/${V0}1/source_deletions_heal/char
+TEST ! stat $B0/${V0}1/source_deletions_heal/file
+TEST ! stat $B0/${V0}1/source_deletions_heal/slink
+TEST ! stat $B0/${V0}1/source_deletions_heal/dir1
+TEST ! stat $B0/${V0}1/source_deletions_me/fifo
+TEST ! stat $B0/${V0}1/source_deletions_me/block
+TEST ! stat $B0/${V0}1/source_deletions_me/char
+TEST ! stat $B0/${V0}1/source_deletions_me/file
+TEST ! stat $B0/${V0}1/source_deletions_me/slink
+TEST ! stat $B0/${V0}1/source_deletions_me/dir1
+TEST ! stat $B0/${V0}1/source_self_accusing/fifo
+TEST ! stat $B0/${V0}1/source_self_accusing/block
+TEST ! stat $B0/${V0}1/source_self_accusing/char
+TEST ! stat $B0/${V0}1/source_self_accusing/file
+TEST ! stat $B0/${V0}1/source_self_accusing/slink
+TEST ! stat $B0/${V0}1/source_self_accusing/dir1
+
+
+TEST mkfifo source_creations_heal/fifo
+TEST mknod source_creations_heal/block b 4 5
+TEST mknod source_creations_heal/char c 1 5
+TEST touch source_creations_heal/file
+TEST ln -s source_creations_heal/file source_creations_heal/slink
+TEST mkdir source_creations_heal/dir1
+TEST mkdir source_creations_heal/dir1/dir2
+
+TEST mkfifo source_creations_me/fifo
+TEST mknod source_creations_me/block b 4 5
+TEST mknod source_creations_me/char c 1 5
+TEST touch source_creations_me/file
+TEST ln -s source_creations_me/file source_creations_me/slink
+TEST mkdir source_creations_me/dir1
+TEST mkdir source_creations_me/dir1/dir2
+
+$CLI volume stop $V0
+
+#simulate fool fool scenario for fool_* dirs
+setfattr -x trusted.afr.$V0-client-0 $B0/${V0}1/{fool_heal,fool_me}
+setfattr -n trusted.afr.dirty -v 0x000000000000000000000001 $B0/${V0}1/{fool_heal,fool_me}
+setfattr -n trusted.afr.$V0-client-1 -v 0x000000000000000000000001 $B0/${V0}1/{v1_fool_heal,v1_fool_me}
+
+#Simulate v1-dirty(self-accusing but no pending ops on others) scenario for v1-dirty
+setfattr -x trusted.afr.$V0-client-0 $B0/${V0}1/v1_dirty_{heal,me}
+setfattr -n trusted.afr.$V0-client-1 -v 0x000000000000000000000001 $B0/${V0}1/v1_dirty_{heal,me}
+
+$CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 1
+TEST kill_brick $V0 $H0 $B0/${V0}1
+
+TEST touch spb_heal/1 spb/0 spb_me_heal/1 spb_me/0 fool_heal/1 fool_me/1 v1_fool_heal/1 v1_fool_me/1
+
+$CLI volume stop $V0
+
+#simulate fool fool scenario for fool_* dirs
+setfattr -x trusted.afr.$V0-client-1 $B0/${V0}0/{fool_heal,fool_me}
+setfattr -n trusted.afr.dirty -v 0x000000000000000000000001 $B0/${V0}1/{fool_heal,fool_me}
+setfattr -n trusted.afr.$V0-client-0 -v 0x000000000000000000000001 $B0/${V0}1/{v1_fool_heal,v1_fool_me}
+
+#simulate self-accusing for source_self_accusing
+TEST setfattr -n trusted.afr.$V0-client-0 -v 0x000000000000000000000006 $B0/${V0}0/source_self_accusing
+
+$CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+
+# Check if conservative merges happened correctly on _me_ dirs
+TEST stat spb_me_heal/1
+TEST stat $B0/${V0}0/spb_me_heal/1
+TEST stat $B0/${V0}1/spb_me_heal/1
+
+TEST stat spb_me_heal/0
+TEST stat $B0/${V0}0/spb_me_heal/0
+TEST stat $B0/${V0}1/spb_me_heal/0
+
+TEST stat fool_me/1
+TEST stat $B0/${V0}0/fool_me/1
+TEST stat $B0/${V0}1/fool_me/1
+
+TEST stat fool_me/0
+TEST stat $B0/${V0}0/fool_me/0
+TEST stat $B0/${V0}1/fool_me/0
+
+TEST stat v1_fool_me/0
+TEST stat $B0/${V0}0/v1_fool_me/0
+TEST stat $B0/${V0}1/v1_fool_me/0
+
+TEST stat v1_fool_me/1
+TEST stat $B0/${V0}0/v1_fool_me/1
+TEST stat $B0/${V0}1/v1_fool_me/1
+
+TEST stat v1_dirty_me/0
+TEST stat $B0/${V0}0/v1_dirty_me/0
+TEST stat $B0/${V0}1/v1_dirty_me/0
+
+#Check if files that have gfid-mismatches in _me_ are giving EIO
+TEST ! stat spb_me/0
+
+#Check if stale files are deleted on access
+TEST ! stat source_deletions_me/fifo
+TEST ! stat $B0/${V0}0/source_deletions_me/fifo
+TEST ! stat $B0/${V0}1/source_deletions_me/fifo
+TEST ! stat source_deletions_me/block
+TEST ! stat $B0/${V0}0/source_deletions_me/block
+TEST ! stat $B0/${V0}1/source_deletions_me/block
+TEST ! stat source_deletions_me/char
+TEST ! stat $B0/${V0}0/source_deletions_me/char
+TEST ! stat $B0/${V0}1/source_deletions_me/char
+TEST ! stat source_deletions_me/file
+TEST ! stat $B0/${V0}0/source_deletions_me/file
+TEST ! stat $B0/${V0}1/source_deletions_me/file
+TEST ! stat source_deletions_me/file
+TEST ! stat $B0/${V0}0/source_deletions_me/file
+TEST ! stat $B0/${V0}1/source_deletions_me/file
+TEST ! stat source_deletions_me/dir1/dir2
+TEST ! stat $B0/${V0}0/source_deletions_me/dir1/dir2
+TEST ! stat $B0/${V0}1/source_deletions_me/dir1/dir2
+TEST ! stat source_deletions_me/dir1
+TEST ! stat $B0/${V0}0/source_deletions_me/dir1
+TEST ! stat $B0/${V0}1/source_deletions_me/dir1
+
+#Test if the files created as part of access are healed correctly
+r=$(get_file_type source_creations_me/fifo)
+EXPECT "$r" get_file_type $B0/${V0}0/source_creations_me/fifo
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_me/fifo
+TEST [ -p source_creations_me/fifo ]
+
+r=$(get_file_type source_creations_me/block)
+EXPECT "$r" get_file_type $B0/${V0}0/source_creations_me/block
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_me/block
+EXPECT "^4 5$" stat -c "%t %T" $B0/${V0}1/source_creations_me/block
+EXPECT "^4 5$" stat -c "%t %T" $B0/${V0}0/source_creations_me/block
+TEST [ -b source_creations_me/block ]
+
+r=$(get_file_type source_creations_me/char)
+EXPECT "$r" get_file_type $B0/${V0}0/source_creations_me/char
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_me/char
+EXPECT "^1 5$" stat -c "%t %T" $B0/${V0}1/source_creations_me/char
+EXPECT "^1 5$" stat -c "%t %T" $B0/${V0}0/source_creations_me/char
+TEST [ -c source_creations_me/char ]
+
+r=$(get_file_type source_creations_me/file)
+EXPECT "$r" get_file_type $B0/${V0}0/source_creations_me/file
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_me/file
+TEST [ -f source_creations_me/file ]
+
+r=$(get_file_type source_creations_me/slink)
+EXPECT "$r" get_file_type $B0/${V0}0/source_creations_me/slink
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_me/slink
+TEST [ -h source_creations_me/slink ]
+
+r=$(get_file_type source_creations_me/dir1/dir2)
+EXPECT "$r" get_file_type $B0/${V0}0/source_creations_me/dir1/dir2
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_me/dir1/dir2
+TEST [ -d source_creations_me/dir1/dir2 ]
+
+r=$(get_file_type source_creations_me/dir1)
+EXPECT "$r" get_file_type $B0/${V0}0/source_creations_me/dir1
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_me/dir1
+TEST [ -d source_creations_me/dir1 ]
+
+#Trigger heal and check _heal dirs are healed properly
+#Trigger change in event generation number. That way inodes would get refreshed during lookup
+TEST kill_brick $V0 $H0 $B0/${V0}1
+$CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+
+TEST stat spb_heal
+TEST stat spb_me_heal
+TEST stat fool_heal
+TEST stat fool_me
+TEST stat v1_fool_heal
+TEST stat v1_fool_me
+TEST stat source_deletions_heal
+TEST stat source_deletions_me
+TEST stat source_self_accusing
+TEST stat source_creations_heal
+TEST stat source_creations_me
+TEST stat v1_dirty_heal
+TEST stat v1_dirty_me
+TEST $CLI volume stop $V0
+TEST rm -rf $B0/${V0}{0,1}/.glusterfs/indices/xattrop/*
+
+$CLI volume start $V0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+
+#Create base entry in indices/xattrop
+echo "Data" > $M0/FILE
+rm -f $M0/FILE
+EXPECT "1" count_index_entries $B0/${V0}0
+EXPECT "1" count_index_entries $B0/${V0}1
+
+TEST $CLI volume stop $V0;
+
+#Create entries for fool_heal and fool_me to ensure they are fully healed and dirty xattrs erased, before triggering index heal
+create_brick_xattrop_entry $B0/${V0}0 fool_heal fool_me source_creations_heal/dir1
+
+$CLI volume start $V0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+
+$CLI volume heal $V0 enable
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" glustershd_up_status
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 1
+
+TEST $CLI volume heal $V0;
+EXPECT_WITHIN $HEAL_TIMEOUT "~" print_pending_heals spb_heal spb_me_heal fool_heal fool_me v1_fool_heal v1_fool_me source_deletions_heal source_deletions_me source_creations_heal source_creations_me v1_dirty_heal v1_dirty_me source_self_accusing
+
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 spb_heal
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 spb_me_heal
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 fool_heal
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 fool_me
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 v1_fool_heal
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 v1_fool_me
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 source_deletions_heal
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 source_deletions_me
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 source_self_accusing
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 source_creations_heal
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 source_creations_me
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 v1_dirty_heal
+EXPECT "Y${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}${zero_xattr}" heal_status $B0/${V0}0 $B0/${V0}1 v1_dirty_me
+
+#Don't access the files/dirs from mount point as that may cause self-heals
+# Check if conservative merges happened correctly on heal dirs
+TEST stat $B0/${V0}0/spb_heal/1
+TEST stat $B0/${V0}1/spb_heal/1
+
+TEST stat $B0/${V0}0/spb_heal/0
+TEST stat $B0/${V0}1/spb_heal/0
+
+TEST stat $B0/${V0}0/fool_heal/1
+TEST stat $B0/${V0}1/fool_heal/1
+
+TEST stat $B0/${V0}0/fool_heal/0
+TEST stat $B0/${V0}1/fool_heal/0
+
+TEST stat $B0/${V0}0/v1_fool_heal/0
+TEST stat $B0/${V0}1/v1_fool_heal/0
+
+TEST stat $B0/${V0}0/v1_fool_heal/1
+TEST stat $B0/${V0}1/v1_fool_heal/1
+
+TEST stat $B0/${V0}0/v1_dirty_heal/0
+TEST stat $B0/${V0}1/v1_dirty_heal/0
+
+#Check if files that have gfid-mismatches in spb are giving EIO
+TEST ! stat spb/0
+
+#Check if stale files are deleted on access
+TEST ! stat $B0/${V0}0/source_deletions_heal/fifo
+TEST ! stat $B0/${V0}1/source_deletions_heal/fifo
+TEST ! stat $B0/${V0}0/source_deletions_heal/block
+TEST ! stat $B0/${V0}1/source_deletions_heal/block
+TEST ! stat $B0/${V0}0/source_deletions_heal/char
+TEST ! stat $B0/${V0}1/source_deletions_heal/char
+TEST ! stat $B0/${V0}0/source_deletions_heal/file
+TEST ! stat $B0/${V0}1/source_deletions_heal/file
+TEST ! stat $B0/${V0}0/source_deletions_heal/file
+TEST ! stat $B0/${V0}1/source_deletions_heal/file
+TEST ! stat $B0/${V0}0/source_deletions_heal/dir1/dir2
+TEST ! stat $B0/${V0}1/source_deletions_heal/dir1/dir2
+TEST ! stat $B0/${V0}0/source_deletions_heal/dir1
+TEST ! stat $B0/${V0}1/source_deletions_heal/dir1
+
+#Check if stale files are deleted on access
+TEST ! stat $B0/${V0}0/source_self_accusing/fifo
+TEST ! stat $B0/${V0}1/source_self_accusing/fifo
+TEST ! stat $B0/${V0}0/source_self_accusing/block
+TEST ! stat $B0/${V0}1/source_self_accusing/block
+TEST ! stat $B0/${V0}0/source_self_accusing/char
+TEST ! stat $B0/${V0}1/source_self_accusing/char
+TEST ! stat $B0/${V0}0/source_self_accusing/file
+TEST ! stat $B0/${V0}1/source_self_accusing/file
+TEST ! stat $B0/${V0}0/source_self_accusing/file
+TEST ! stat $B0/${V0}1/source_self_accusing/file
+TEST ! stat $B0/${V0}0/source_self_accusing/dir1/dir2
+TEST ! stat $B0/${V0}1/source_self_accusing/dir1/dir2
+TEST ! stat $B0/${V0}0/source_self_accusing/dir1
+TEST ! stat $B0/${V0}1/source_self_accusing/dir1
+
+#Test if the files created as part of full self-heal correctly
+r=$(get_file_type $B0/${V0}0/source_creations_heal/fifo)
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_heal/fifo
+TEST [ -p $B0/${V0}0/source_creations_heal/fifo ]
+EXPECT "^4 5$" stat -c "%t %T" $B0/${V0}1/source_creations_heal/block
+EXPECT "^4 5$" stat -c "%t %T" $B0/${V0}0/source_creations_heal/block
+
+r=$(get_file_type $B0/${V0}0/source_creations_heal/block)
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_heal/block
+
+r=$(get_file_type $B0/${V0}0/source_creations_heal/char)
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_heal/char
+EXPECT "^1 5$" stat -c "%t %T" $B0/${V0}1/source_creations_heal/char
+EXPECT "^1 5$" stat -c "%t %T" $B0/${V0}0/source_creations_heal/char
+
+r=$(get_file_type $B0/${V0}0/source_creations_heal/file)
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_heal/file
+TEST [ -f $B0/${V0}0/source_creations_heal/file ]
+
+r=$(get_file_type source_creations_heal/file $B0/${V0}0/slink)
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_heal/file slink
+TEST [ -h $B0/${V0}0/source_creations_heal/slink ]
+
+r=$(get_file_type $B0/${V0}0/source_creations_heal/dir1/dir2)
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_heal/dir1/dir2
+TEST [ -d $B0/${V0}0/source_creations_heal/dir1/dir2 ]
+
+r=$(get_file_type $B0/${V0}0/source_creations_heal/dir1)
+EXPECT "$r" get_file_type $B0/${V0}1/source_creations_heal/dir1
+TEST [ -d $B0/${V0}0/source_creations_heal/dir1 ]
+
+cd -
+
+#Anonymous directory shouldn't be created
+TEST mkdir $M0/rename-dir
+before_rename=$(STAT_INO $B0/${V0}1/rename-dir)
+TEST kill_brick $V0 $H0 $B0/${V0}1
+TEST mv $M0/rename-dir $M0/new-name
+TEST $CLI volume start $V0 force
+#'spb' is in split-brain so pending-heal-count will be 2
+EXPECT_WITHIN $HEAL_TIMEOUT "^2$" get_pending_heal_count $V0
+after_rename=$(STAT_INO $B0/${V0}1/new-name)
+EXPECT "0" echo $(ls -a $B0/${V0}0/ | grep anonymous-inode | wc -l)
+EXPECT "0" echo $(ls -a $B0/${V0}1/ | grep anonymous-inode | wc -l)
+EXPECT_NOT "$before_rename" echo $after_rename
+cleanup
diff --git a/tests/basic/afr/gfid-mismatch-resolution-with-fav-child-policy.t b/tests/basic/afr/gfid-mismatch-resolution-with-fav-child-policy.t
index 12af0c85461..35e295dc170 100644
--- a/tests/basic/afr/gfid-mismatch-resolution-with-fav-child-policy.t
+++ b/tests/basic/afr/gfid-mismatch-resolution-with-fav-child-policy.t
@@ -227,4 +227,3 @@ HEALED_MD5=$(md5sum $B0/${V0}2/f4 | cut -d\ -f1)
TEST [ "$MAJORITY_MD5" == "$HEALED_MD5" ]
cleanup;
-#G_TESTDEF_TEST_STATUS_NETBSD7=1501390
diff --git a/tests/basic/afr/rename-data-loss.t b/tests/basic/afr/rename-data-loss.t
new file mode 100644
index 00000000000..256ee2aafce
--- /dev/null
+++ b/tests/basic/afr/rename-data-loss.t
@@ -0,0 +1,72 @@
+#!/bin/bash
+#Self-heal tests
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../afr.rc
+
+cleanup;
+
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 replica 2 $H0:$B0/brick{0,1}
+TEST $CLI volume set $V0 write-behind off
+TEST $CLI volume set $V0 self-heal-daemon off
+TEST $CLI volume set $V0 data-self-heal off
+TEST $CLI volume set $V0 metadata-self-heal off
+TEST $CLI volume set $V0 entry-self-heal off
+TEST $CLI volume start $V0
+EXPECT 'Started' volinfo_field $V0 'Status'
+TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 $M0;
+
+cd $M0
+TEST `echo "line1" >> file1`
+TEST mkdir dir1
+TEST mkdir dir2
+TEST mkdir -p dir1/dira/dirb
+TEST `echo "line1">>dir1/dira/dirb/file1`
+TEST mkdir delete_me
+TEST `echo "line1" >> delete_me/file1`
+
+#brick0 has witnessed the second write while brick1 is down.
+TEST kill_brick $V0 $H0 $B0/brick1
+TEST `echo "line2" >> file1`
+TEST `echo "line2" >> dir1/dira/dirb/file1`
+TEST `echo "line2" >> delete_me/file1`
+
+#Toggle the bricks that are up/down.
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status $V0 1
+TEST kill_brick $V0 $H0 $B0/brick0
+
+#Rename when the 'source' brick0 for data-selfheals is down.
+mv file1 file2
+mv dir1/dira dir2
+
+#Delete a dir when brick0 is down.
+rm -rf delete_me
+cd -
+
+#Bring everything up and trigger heal
+TEST $CLI volume set $V0 self-heal-daemon on
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status $V0 0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" glustershd_up_status
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 1
+TEST $CLI volume heal $V0
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" get_pending_heal_count $V0
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" afr_anon_entry_count $B0/brick0
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" afr_anon_entry_count $B0/brick1
+
+#Remount to avoid reading from caches
+EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0
+TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 $M0;
+EXPECT "line2" tail -1 $M0/file2
+EXPECT "line2" tail -1 $M0/dir2/dira/dirb/file1
+TEST ! stat $M0/delete_me/file1
+TEST ! stat $M0/delete_me
+
+anon_inode_name=$(ls -a $B0/brick0 | grep glusterfs-anonymous-inode)
+TEST [[ -d $B0/brick0/$anon_inode_name ]]
+TEST [[ -d $B0/brick1/$anon_inode_name ]]
+cleanup
diff --git a/tests/basic/afr/split-brain-favorite-child-policy-client-side-healing.t b/tests/basic/afr/split-brain-favorite-child-policy-client-side-healing.t
new file mode 100644
index 00000000000..7c249c4bcbd
--- /dev/null
+++ b/tests/basic/afr/split-brain-favorite-child-policy-client-side-healing.t
@@ -0,0 +1,124 @@
+#!/bin/bash
+
+#Test the client side split-brain resolution
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+
+cleanup;
+
+GET_MDATA_PATH=$(dirname $0)/../../utils
+build_tester $GET_MDATA_PATH/get-mdata-xattr.c
+
+TEST glusterd
+TEST pidof glusterd
+
+count_files () {
+ ls $1 | wc -l
+}
+
+#Create replica 2 volume
+TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{0,1}
+TEST $CLI volume set $V0 performance.write-behind off
+TEST $CLI volume heal $V0 disable
+TEST $CLI volume set $V0 cluster.quorum-type fixed
+TEST $CLI volume set $V0 cluster.quorum-count 1
+TEST $CLI volume set $V0 cluster.metadata-self-heal on
+TEST $CLI volume set $V0 cluster.data-self-heal on
+TEST $CLI volume set $V0 cluster.entry-self-heal on
+
+TEST $CLI volume start $V0
+TEST $GFS --volfile-id=/$V0 --volfile-server=$H0 $M0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status $V0 0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status $V0 1
+
+TEST mkdir $M0/data
+TEST touch $M0/data/file
+
+
+############ Client side healing using favorite-child-policy = mtime #################
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST dd if=/dev/urandom of=$M0/data/file bs=1024 count=1024
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+TEST kill_brick $V0 $H0 $B0/${V0}1
+TEST dd if=/dev/urandom of=$M0/data/file bs=1024 count=1024
+
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 1
+
+mtime1=$(get_mtime $B0/${V0}0/data/file)
+mtime2=$(get_mtime $B0/${V0}1/data/file)
+if (( $(echo "$mtime1 > $mtime2" | bc -l) )); then
+ LATEST_MTIME_MD5=$(md5sum $B0/${V0}0/data/file | cut -d\ -f1)
+else
+ LATEST_MTIME_MD5=$(md5sum $B0/${V0}1/data/file | cut -d\ -f1)
+fi
+
+#file will be in split-brain
+cat $M0/data/file > /dev/null
+EXPECT "1" echo $?
+
+TEST $CLI volume set $V0 cluster.favorite-child-policy mtime
+TEST $CLI volume start $V0 force
+
+EXPECT_WITHIN $HEAL_TIMEOUT "^2$" afr_get_split_brain_count $V0
+cat $M0/data/file > /dev/null
+EXPECT "0" echo $?
+M0_MD5=$(md5sum $M0/data/file | cut -d\ -f1)
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" afr_get_split_brain_count $V0
+TEST [ "$LATEST_MTIME_MD5" == "$M0_MD5" ]
+
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" get_pending_heal_count $V0
+
+B0_MD5=$(md5sum $B0/${V0}0/data/file | cut -d\ -f1)
+B1_MD5=$(md5sum $B0/${V0}1/data/file | cut -d\ -f1)
+TEST [ "$LATEST_MTIME_MD5" == "$B0_MD5" ]
+TEST [ "$LATEST_MTIME_MD5" == "$B1_MD5" ]
+
+############ Client side directory conservative merge #################
+TEST $CLI volume reset $V0 cluster.favorite-child-policy
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST touch $M0/data/test
+files=$(count_files $M0/data)
+EXPECT "2" echo $files
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+TEST kill_brick $V0 $H0 $B0/${V0}1
+TEST touch $M0/data/test1
+files=$(count_files $M0/data)
+EXPECT "2" echo $files
+
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" brick_up_status $V0 $H0 $B0/${V0}1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 1
+
+#data dir will be in entry split-brain
+ls $M0/data > /dev/null
+EXPECT "2" echo $?
+
+TEST $CLI volume set $V0 cluster.favorite-child-policy mtime
+
+EXPECT_WITHIN $HEAL_TIMEOUT "^2$" afr_get_split_brain_count $V0
+
+
+ls $M0/data > /dev/null
+EXPECT "0" echo $?
+
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" afr_get_split_brain_count $V0
+#Entry Split-brain is gone, but data self-heal is pending on the files
+EXPECT_WITHIN $HEAL_TIMEOUT "^2$" get_pending_heal_count $V0
+
+cat $M0/data/test > /dev/null
+cat $M0/data/test1 > /dev/null
+
+EXPECT_WITHIN $HEAL_TIMEOUT "^0$" get_pending_heal_count $V0
+files=$(count_files $M0/data)
+EXPECT "3" echo $files
+
+TEST force_umount $M0
+TEST rm $GET_MDATA_PATH/get-mdata-xattr
+
+cleanup
diff --git a/tests/basic/distribute/spare_file_rebalance.t b/tests/basic/distribute/spare_file_rebalance.t
new file mode 100644
index 00000000000..061c02f7392
--- /dev/null
+++ b/tests/basic/distribute/spare_file_rebalance.t
@@ -0,0 +1,51 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../dht.rc
+
+# Initialize
+#------------------------------------------------------------
+cleanup;
+
+# Start glusterd
+TEST glusterd;
+TEST pidof glusterd;
+TEST $CLI volume info;
+
+# Create a volume
+TEST $CLI volume create $V0 $H0:$B0/${V0}{1,2};
+
+# Verify volume creation
+EXPECT "$V0" volinfo_field $V0 'Volume Name';
+EXPECT 'Created' volinfo_field $V0 'Status';
+
+# Start volume and verify successful start
+TEST $CLI volume start $V0;
+EXPECT 'Started' volinfo_field $V0 'Status';
+TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 --entry-timeout=0 $M0;
+
+#------------------------------------------------------------
+
+# Test case - Create sparse files on MP and verify
+# file info after rebalance
+#------------------------------------------------------------
+
+# Create some sparse files and get their size
+TEST cd $M0;
+dd if=/dev/urandom of=sparse_file bs=10k count=1 seek=2M
+cp --sparse=always sparse_file sparse_file_3;
+
+# Add a 3rd brick
+TEST $CLI volume add-brick $V0 $H0:$B0/${V0}3;
+
+# Trigger rebalance
+TEST $CLI volume rebalance $V0 start force;
+EXPECT_WITHIN $REBALANCE_TIMEOUT "0" rebalance_completed;
+
+# Compare original and rebalanced files
+TEST cd $B0/${V0}2
+TEST cmp sparse_file $B0/${V0}3/sparse_file_3
+EXPECT_WITHIN 30 "";
+
+cleanup;
diff --git a/tests/basic/ec/ec-quorum-count.t b/tests/basic/ec/ec-quorum-count.t
index 56b5329c411..9310ebbb8f2 100644
--- a/tests/basic/ec/ec-quorum-count.t
+++ b/tests/basic/ec/ec-quorum-count.t
@@ -11,6 +11,8 @@ TEST $CLI volume create $V0 disperse 6 redundancy 2 $H0:$B0/${V0}{0..5}
TEST $CLI volume create $V1 $H0:$B0/${V1}{0..5}
TEST $CLI volume set $V0 disperse.eager-lock-timeout 5
TEST $CLI volume set $V0 performance.flush-behind off
+TEST $CLI volume set $V0 disperse.background-heals 0
+TEST $CLI volume set $V0 disperse.heal-wait-qlength 0
#Should fail on non-disperse volume
TEST ! $CLI volume set $V1 disperse.quorum-count 5
diff --git a/tests/basic/fencing/afr-lock-heal-basic.t b/tests/basic/fencing/afr-lock-heal-basic.t
index c5d7d6fe8fd..69131af085d 100644
--- a/tests/basic/fencing/afr-lock-heal-basic.t
+++ b/tests/basic/fencing/afr-lock-heal-basic.t
@@ -56,8 +56,8 @@ EXPECT "Y" is_gfapi_program_alive $client1_pid
# Check lock is present on brick-1 and brick-2
b1_sdump=$(generate_brick_statedump $V0 $H0 $B0/${V0}0)
-b2_sdump=$(generate_brick_statedump $V0 $H0 $B0/${V0}1)
c1_lock_on_b1="$(egrep "$inode" $b1_sdump -A3| egrep 'ACTIVE.*client-0'| uniq| awk '{print $1,$2,$3,S4,$5,$6,$7,$8}'|tr -d '(,), ,')"
+b2_sdump=$(generate_brick_statedump $V0 $H0 $B0/${V0}1)
c1_lock_on_b2="$(egrep "$inode" $b2_sdump -A3| egrep 'ACTIVE.*client-1'| uniq| awk '{print $1,$2,$3,S4,$5,$6,$7,$8}'|tr -d '(,), ,')"
TEST [ "$c1_lock_on_b1" == "$c1_lock_on_b2" ]
@@ -83,10 +83,10 @@ TEST sleep 10 #Needed for client to re-open fd? Otherwise client_pre_lk_v2() fai
# Check that all bricks now have locks from client 2 only.
b1_sdump=$(generate_brick_statedump $V0 $H0 $B0/${V0}0)
-b2_sdump=$(generate_brick_statedump $V0 $H0 $B0/${V0}1)
-b3_sdump=$(generate_brick_statedump $V0 $H0 $B0/${V0}2)
c2_lock_on_b1="$(egrep "$inode" $b1_sdump -A3| egrep 'ACTIVE.*client-0'| uniq| awk '{print $1,$2,$3,S4,$5,$6,$7,$8}'|tr -d '(,), ,')"
+b2_sdump=$(generate_brick_statedump $V0 $H0 $B0/${V0}1)
c2_lock_on_b2="$(egrep "$inode" $b2_sdump -A3| egrep 'ACTIVE.*client-1'| uniq| awk '{print $1,$2,$3,S4,$5,$6,$7,$8}'|tr -d '(,), ,')"
+b3_sdump=$(generate_brick_statedump $V0 $H0 $B0/${V0}2)
c2_lock_on_b3="$(egrep "$inode" $b3_sdump -A3| egrep 'ACTIVE.*client-2'| uniq| awk '{print $1,$2,$3,S4,$5,$6,$7,$8}'|tr -d '(,), ,')"
TEST [ "$c2_lock_on_b1" == "$c2_lock_on_b2" ]
TEST [ "$c2_lock_on_b1" == "$c2_lock_on_b3" ]
diff --git a/tests/basic/fuse/active-io-graph-switch.t b/tests/basic/fuse/active-io-graph-switch.t
new file mode 100644
index 00000000000..6ec3e1fcbfa
--- /dev/null
+++ b/tests/basic/fuse/active-io-graph-switch.t
@@ -0,0 +1,65 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+
+TESTS_EXPECTED_IN_LOOP=12
+
+function perform_io_on_mount {
+ local m="$1"
+ local f="$2"
+ local lockfile="$3"
+ while [ -f "$m/$lockfile" ];
+ do
+ dd if=/dev/zero of=$m/$f bs=1M count=1
+ done
+}
+
+function perform_graph_switch {
+ for i in {1..3}
+ do
+ TEST_IN_LOOP $CLI volume set $V0 performance.stat-prefetch off
+ sleep 3
+ TEST_IN_LOOP $CLI volume set $V0 performance.stat-prefetch on
+ sleep 3
+ done
+}
+
+function count_files {
+ ls $M0 | wc -l
+}
+
+cleanup;
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 replica 3 $H0:$B0/${V0}{0,1,2}
+TEST $CLI volume set $V0 flush-behind off
+TEST $CLI volume start $V0
+TEST $GFS --volfile-id=/$V0 --volfile-server=$H0 $M0
+TEST touch $M0/lock
+for i in {1..100}; do perform_io_on_mount $M0 $i lock & done
+EXPECT_WITHIN 5 "101" count_files
+
+perform_graph_switch
+TEST rm -f $M0/lock
+wait
+EXPECT "100" count_files
+TEST rm -f $M0/{1..100}
+EXPECT "0" count_files
+
+EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0
+
+#Repeat the tests with reader-thread-count
+TEST $GFS --reader-thread-count=10 --volfile-id=/$V0 --volfile-server=$H0 $M0
+TEST touch $M0/lock
+for i in {1..100}; do perform_io_on_mount $M0 $i lock & done
+EXPECT_WITHIN 5 "101" count_files
+
+perform_graph_switch
+TEST rm -f $M0/lock
+wait
+EXPECT "100" count_files
+TEST rm -f $M0/{1..100}
+EXPECT "0" count_files
+
+cleanup
diff --git a/tests/basic/gfapi/gfapi-ssl-load-volfile-test.c b/tests/basic/gfapi/gfapi-ssl-load-volfile-test.c
new file mode 100644
index 00000000000..7beb8dd1fe4
--- /dev/null
+++ b/tests/basic/gfapi/gfapi-ssl-load-volfile-test.c
@@ -0,0 +1,127 @@
+#include <fcntl.h>
+#include <unistd.h>
+#include <time.h>
+#include <limits.h>
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <errno.h>
+#include <glusterfs/api/glfs.h>
+#include <glusterfs/api/glfs-handles.h>
+
+#define LOG_ERR(msg) \
+ do { \
+ fprintf(stderr, "%s : Error (%s)\n", msg, strerror(errno)); \
+ } while (0)
+
+glfs_t *
+init_glfs(const char *hostname, const char *volname, const char *volfile,
+ const char *logfile)
+{
+ int ret = -1;
+ glfs_t *fs = NULL;
+
+ fs = glfs_new(volname);
+ if (!fs) {
+ LOG_ERR("glfs_new failed");
+ return NULL;
+ }
+
+ ret = glfs_set_volfile(fs, volfile);
+ if (ret < 0) {
+ LOG_ERR("glfs_set_volfile failed");
+ goto out;
+ }
+
+ ret = glfs_set_logging(fs, logfile, 7);
+ if (ret < 0) {
+ LOG_ERR("glfs_set_logging failed");
+ goto out;
+ }
+
+ ret = glfs_init(fs);
+ if (ret < 0) {
+ LOG_ERR("glfs_init failed");
+ goto out;
+ }
+
+ ret = 0;
+out:
+ if (ret) {
+ glfs_fini(fs);
+ fs = NULL;
+ }
+
+ return fs;
+}
+
+int
+glfs_test_function(const char *hostname, const char *volname,
+ const char *volfile, const char *logfile)
+{
+ int ret = -1;
+ int flags = O_CREAT | O_RDWR;
+ glfs_t *fs = NULL;
+ glfs_fd_t *glfd = NULL;
+ const char *buff = "This is from my prog\n";
+ const char *filename = "glfs_test.txt";
+
+ fs = init_glfs(hostname, volname, volfile, logfile);
+ if (fs == NULL) {
+ LOG_ERR("init_glfs failed");
+ return -1;
+ }
+
+ glfd = glfs_creat(fs, filename, flags, 0644);
+ if (glfd == NULL) {
+ LOG_ERR("glfs_creat failed");
+ goto out;
+ }
+
+ ret = glfs_write(glfd, buff, strlen(buff), flags);
+ if (ret < 0) {
+ LOG_ERR("glfs_write failed");
+ goto out;
+ }
+
+ ret = glfs_close(glfd);
+ if (ret < 0) {
+ LOG_ERR("glfs_write failed");
+ goto out;
+ }
+
+out:
+ ret = glfs_fini(fs);
+ if (ret) {
+ LOG_ERR("glfs_fini failed");
+ }
+
+ return ret;
+}
+
+int
+main(int argc, char *argv[])
+{
+ int ret = 0;
+ char *hostname = NULL;
+ char *volname = NULL;
+ char *volfile = NULL;
+ char *logfile = NULL;
+
+ if (argc != 5) {
+ fprintf(stderr, "Invalid argument\n");
+ exit(1);
+ }
+
+ hostname = argv[1];
+ volname = argv[2];
+ volfile = argv[3];
+ logfile = argv[4];
+
+ ret = glfs_test_function(hostname, volname, volfile, logfile);
+ if (ret) {
+ LOG_ERR("glfs_test_function failed");
+ }
+
+ return ret;
+}
diff --git a/tests/basic/gfapi/gfapi-ssl-load-volfile-test.t b/tests/basic/gfapi/gfapi-ssl-load-volfile-test.t
new file mode 100755
index 00000000000..8e94df9d321
--- /dev/null
+++ b/tests/basic/gfapi/gfapi-ssl-load-volfile-test.t
@@ -0,0 +1,76 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../traps.rc
+. $(dirname $0)/../../ssl.rc
+
+cleanup;
+
+sed -e "s,@@HOSTNAME@@,${H0},g" -e "s,@@BRICKPATH@@,${B0}/brick1,g" \
+ -e "s,@@SSL@@,off,g" \
+ $(dirname ${0})/protocol-client-ssl.vol.in \
+ > $(dirname ${0})/protocol-client-ssl.vol
+
+TEST create_self_signed_certs
+
+TEST glusterd
+
+TEST $CLI volume create $V0 $H0:$B0/brick1;
+EXPECT 'Created' volinfo_field $V0 'Status';
+
+TEST $CLI volume start $V0;
+EXPECT 'Started' volinfo_field $V0 'Status';
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" online_brick_count
+
+logdir=`gluster --print-logdir`
+
+TEST build_tester $(dirname $0)/gfapi-ssl-load-volfile-test.c -lgfapi
+
+# Run test without I/O or management encryption
+TEST $(dirname $0)/gfapi-ssl-load-volfile-test $H0 $V0 \
+ $(dirname ${0})/protocol-client-ssl.vol \
+ $logdir/gfapi-ssl-load-volfile-test.log
+
+# Enable management encryption
+touch $GLUSTERD_WORKDIR/secure-access
+
+killall_gluster
+
+TEST glusterd
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" online_brick_count
+
+# Run test with management encryption (No I/O encryption)
+TEST $(dirname $0)/gfapi-ssl-load-volfile-test $H0 $V0 \
+ $(dirname ${0})/protocol-client-ssl.vol \
+ $logdir/gfapi-ssl-load-volfile-test.log
+
+# Enable I/O encryption
+TEST $CLI volume set $V0 server.ssl on
+
+killall_gluster
+
+sed -e "s,@@HOSTNAME@@,${H0},g" -e "s,@@BRICKPATH@@,${B0}/brick1,g" \
+ -e "s,@@SSL@@,on,g" \
+ $(dirname ${0})/protocol-client-ssl.vol.in \
+ > $(dirname ${0})/protocol-client-ssl.vol
+
+TEST glusterd
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" online_brick_count
+
+# Run test without I/O or management encryption
+TEST $(dirname $0)/gfapi-ssl-load-volfile-test $H0 $V0 \
+ $(dirname ${0})/protocol-client-ssl.vol \
+ $logdir/gfapi-ssl-load-volfile-test.log
+
+cleanup_tester $(dirname $0)/gfapi-ssl-load-volfile-test
+
+TEST $CLI volume stop $V0
+TEST $CLI volume delete $V0
+
+cleanup;
+
+# NetBSD build scripts are not up to date therefore this test
+# is failing in NetBSD. Therefore skipping the test in NetBSD
+# as of now.
+#G_TESTDEF_TEST_STATUS_NETBSD7=KNOWN_ISSUE,BUG=000000
diff --git a/tests/basic/gfapi/protocol-client-ssl.vol.in b/tests/basic/gfapi/protocol-client-ssl.vol.in
new file mode 100644
index 00000000000..cdc0c9d0671
--- /dev/null
+++ b/tests/basic/gfapi/protocol-client-ssl.vol.in
@@ -0,0 +1,15 @@
+#
+# This .vol file expects that there is
+#
+# 1. GlusterD listening on @@HOSTNAME@@
+# 2. a volume that provides a brick on @@BRICKPATH@@
+# 3. the volume with the brick has been started
+#
+volume test
+ type protocol/client
+ option remote-host @@HOSTNAME@@
+ option remote-subvolume @@BRICKPATH@@
+ option transport-type socket
+ option transport.socket.ssl-enabled @@SSL@@
+end-volume
+
diff --git a/tests/basic/metadisp/fsyncdir.c b/tests/basic/metadisp/fsyncdir.c
new file mode 100644
index 00000000000..62b532b9ce4
--- /dev/null
+++ b/tests/basic/metadisp/fsyncdir.c
@@ -0,0 +1,29 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+
+#include <fcntl.h>
+
+int
+main(int argc, char **argv)
+{
+ int pfd;
+
+ pfd = open(argv[1], O_RDONLY | O_DIRECTORY);
+ if (pfd == (-1)) {
+ perror("open");
+ return EXIT_FAILURE;
+ }
+
+ if (rename(argv[2], argv[3]) == (-1)) {
+ perror("rename");
+ return EXIT_FAILURE;
+ }
+
+ if (fsync(pfd) == (-1)) {
+ perror("fsync");
+ return EXIT_FAILURE;
+ }
+
+ return EXIT_SUCCESS;
+}
diff --git a/tests/basic/metadisp/ftruncate.c b/tests/basic/metadisp/ftruncate.c
new file mode 100644
index 00000000000..c9185212c31
--- /dev/null
+++ b/tests/basic/metadisp/ftruncate.c
@@ -0,0 +1,34 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+
+#include <fcntl.h>
+
+int
+main(int argc, char **argv)
+{
+ int pfd;
+
+ pfd = open(argv[1], O_RDWR);
+ if (pfd == (-1)) {
+ perror("open");
+ return EXIT_FAILURE;
+ }
+
+ if (ftruncate(pfd, 0) == (-1)) {
+ perror("ftruncate");
+ return EXIT_FAILURE;
+ }
+
+ if (write(pfd, "hello", 5) == (-1)) {
+ perror("write");
+ return EXIT_FAILURE;
+ }
+
+ if (fsync(pfd) == (-1)) {
+ perror("fsync");
+ return EXIT_FAILURE;
+ }
+
+ return EXIT_SUCCESS;
+}
diff --git a/tests/basic/metadisp/fxattr.c b/tests/basic/metadisp/fxattr.c
new file mode 100644
index 00000000000..e552057778a
--- /dev/null
+++ b/tests/basic/metadisp/fxattr.c
@@ -0,0 +1,107 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+
+#include <fcntl.h>
+#include <string.h>
+#include <sys/types.h>
+#include <sys/xattr.h>
+
+static char MY_XATTR[] = "user.fxtest";
+static char *PROGRAM;
+#define CONSUME(v) \
+ do { \
+ if (!argc) { \
+ fprintf(stderr, "missing argument\n"); \
+ return EXIT_FAILURE; \
+ } \
+ v = argv[0]; \
+ ++argv; \
+ --argc; \
+ } while (0)
+
+static int
+do_get(int argc, char **argv, int fd)
+{
+ char *value;
+ int ret;
+ char buf[1024];
+
+ CONSUME(value);
+
+ ret = fgetxattr(fd, MY_XATTR, buf, sizeof(buf));
+ if (ret == (-1)) {
+ perror("fgetxattr");
+ return EXIT_FAILURE;
+ }
+
+ if (strncmp(buf, value, ret) != 0) {
+ fprintf(stderr, "data mismatch\n");
+ return EXIT_FAILURE;
+ }
+
+ return EXIT_SUCCESS;
+}
+
+static int
+do_set(int argc, char **argv, int fd)
+{
+ char *value;
+ int ret;
+
+ CONSUME(value);
+
+ ret = fsetxattr(fd, MY_XATTR, value, strlen(value), 0);
+ if (ret == (-1)) {
+ perror("fsetxattr");
+ return EXIT_FAILURE;
+ }
+
+ return EXIT_SUCCESS;
+}
+
+static int
+do_remove(int argc, char **argv, int fd)
+{
+ int ret;
+
+ ret = fremovexattr(fd, MY_XATTR);
+ if (ret == (-1)) {
+ perror("femovexattr");
+ return EXIT_FAILURE;
+ }
+
+ return EXIT_SUCCESS;
+}
+
+int
+main(int argc, char **argv)
+{
+ int fd;
+ char *path;
+ char *cmd;
+
+ CONSUME(PROGRAM);
+ CONSUME(path);
+ CONSUME(cmd);
+
+ fd = open(path, O_RDWR);
+ if (fd == (-1)) {
+ perror("open");
+ return EXIT_FAILURE;
+ }
+
+ if (strcmp(cmd, "get") == 0) {
+ return do_get(argc, argv, fd);
+ }
+
+ if (strcmp(cmd, "set") == 0) {
+ return do_set(argc, argv, fd);
+ }
+
+ if (strcmp(cmd, "remove") == 0) {
+ return do_remove(argc, argv, fd);
+ }
+
+ return EXIT_SUCCESS;
+}
diff --git a/tests/basic/metadisp/gfs-fsetxattr.c b/tests/basic/metadisp/gfs-fsetxattr.c
new file mode 100644
index 00000000000..63578bc528f
--- /dev/null
+++ b/tests/basic/metadisp/gfs-fsetxattr.c
@@ -0,0 +1,141 @@
+#include <glusterfs/api/glfs.h>
+#include <glusterfs/api/glfs-handles.h>
+#include <errno.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+int gfapi = 1;
+
+int
+main(int argc, char *argv[])
+{
+ glfs_t *fs = NULL;
+ int ret = 0;
+ int i = 0;
+ glfs_fd_t *fd = NULL;
+ char *topdir = "topdir", *filename = "file1";
+ char *buf = NULL;
+ char *logfile = NULL;
+ char *hostname = NULL;
+ char *basename = NULL;
+ char *dir1 = NULL, *dir2 = NULL, *filename1 = NULL, *filename2 = NULL;
+ struct stat sb = {
+ 0,
+ };
+
+ if (argc != 5) {
+ fprintf(
+ stderr,
+ "Expect following args %s <hostname> <Vol> <log file> <basename>\n",
+ argv[0]);
+ return -1;
+ }
+
+ hostname = argv[1];
+ logfile = argv[3];
+ basename = argv[4];
+
+ fs = glfs_new(argv[2]);
+ if (!fs) {
+ fprintf(stderr, "glfs_new: returned NULL (%s)\n", strerror(errno));
+ return -1;
+ }
+
+ ret = glfs_set_volfile_server(fs, "tcp", hostname, 24007);
+ if (ret < 0) {
+ fprintf(stderr, "glfs_set_volfile_server failed ret:%d (%s)\n", ret,
+ strerror(errno));
+ return -1;
+ }
+
+ ret = glfs_set_logging(fs, logfile, 7);
+ if (ret < 0) {
+ fprintf(stderr, "glfs_set_logging failed with ret: %d (%s)\n", ret,
+ strerror(errno));
+ return -1;
+ }
+
+ ret = glfs_init(fs);
+ if (ret < 0) {
+ fprintf(stderr, "glfs_init failed with ret: %d (%s)\n", ret,
+ strerror(errno));
+ return -1;
+ }
+
+ ret = asprintf(&dir1, "%s-dir", basename);
+ if (ret < 0) {
+ fprintf(stderr, "cannot construct filename (%s)", strerror(errno));
+ return ret;
+ }
+
+ ret = glfs_mkdir(fs, dir1, 0755);
+ if (ret < 0) {
+ fprintf(stderr, "mkdir(%s): %s\n", dir1, strerror(errno));
+ return -1;
+ }
+
+ fd = glfs_opendir(fs, dir1);
+ if (!fd) {
+ fprintf(stderr, "/: %s\n", strerror(errno));
+ return -1;
+ }
+
+ ret = glfs_fsetxattr(fd, "user.dirfattr", "fsetxattr", 9, 0);
+ if (ret < 0) {
+ fprintf(stderr, "fsetxattr(%s): %d (%s)\n", dir1, ret, strerror(errno));
+ return -1;
+ }
+
+ ret = glfs_closedir(fd);
+ if (ret < 0) {
+ fprintf(stderr, "glfs_closedir failed with ret: %d (%s)\n", ret,
+ strerror(errno));
+ return -1;
+ }
+
+ ret = asprintf(&filename1, "%s-file", basename);
+ if (ret < 0) {
+ fprintf(stderr, "cannot construct filename (%s)", strerror(errno));
+ return ret;
+ }
+
+ ret = asprintf(&filename2, "%s-file-renamed", basename);
+ if (ret < 0) {
+ fprintf(stderr, "cannot construct filename (%s)", strerror(errno));
+ return ret;
+ }
+
+ fd = glfs_creat(fs, filename1, O_RDWR, 0644);
+ if (!fd) {
+ fprintf(stderr, "%s: (%p) %s\n", filename1, fd, strerror(errno));
+ return -1;
+ }
+
+ ret = glfs_rename(fs, filename1, filename2);
+ if (ret < 0) {
+ fprintf(stderr, "glfs_rename failed with ret: %d (%s)\n", ret,
+ strerror(errno));
+ return -1;
+ }
+
+ ret = glfs_lstat(fs, filename2, &sb);
+ if (ret < 0) {
+ fprintf(stderr, "glfs_lstat failed with ret: %d (%s)\n", ret,
+ strerror(errno));
+ return -1;
+ }
+
+ ret = glfs_fsetxattr(fd, "user.filefattr", "fsetxattr", 9, 0);
+ if (ret < 0) {
+ fprintf(stderr, "fsetxattr(%s): %d (%s)\n", dir1, ret, strerror(errno));
+ return -1;
+ }
+
+ ret = glfs_close(fd);
+ if (ret < 0) {
+ fprintf(stderr, "glfs_close failed with ret: %d (%s)\n", ret,
+ strerror(errno));
+ return -1;
+ }
+}
diff --git a/tests/basic/metadisp/metadisp.t b/tests/basic/metadisp/metadisp.t
new file mode 100644
index 00000000000..894ffe07226
--- /dev/null
+++ b/tests/basic/metadisp/metadisp.t
@@ -0,0 +1,316 @@
+#!/usr/bin/env bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+
+
+# Considering `--enable-metadisp` is an option for `./configure`,
+# which is disabled by default, this test will never pass regression.
+# But to see the value of this test, run below after configuring
+# with above option :
+# `prove -vmfe '/bin/bash' tests/basic/metadisp/metadisp.t`
+
+#G_TESTDEF_TEST_STATUS_CENTOS6=BAD_TEST
+
+cleanup;
+
+TEST mkdir -p $B0/b0/{0,1}
+
+TEST setfattr -n trusted.glusterfs.volume-id -v 0xddab9eece7b64a95b07351a1f748f56f ${B0}/b0/0
+TEST setfattr -n trusted.glusterfs.volume-id -v 0xddab9eece7b64a95b07351a1f748f56f ${B0}/b0/1
+
+TEST $GFS --volfile=$(dirname $0)/metadisp.vol --volfile-id=$V0 $M0;
+
+NUM_FILES=40
+TEST touch $M0/{1..${NUM_FILES}}
+
+# each drive should get 40 files
+TEST [ $(dir -1 $B0/b0/0/ | wc -l) -eq $NUM_FILES ]
+TEST [ $(dir -1 $B0/b0/1/ | wc -l) -eq $NUM_FILES ]
+
+# now write some data to a file
+echo "hello" > $M0/3
+filename=$$
+echo "hello" > /tmp/metadisp-write-${filename}
+checksum=$(md5sum /tmp/metadisp-write-${filename} | awk '{print $1}')
+TEST [ "$(md5sum $M0/3 | awk '{print $1}')" == "$checksum" ]
+
+# check that the backend file exists on b1
+gfid=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/b0/*/3))
+TEST [ $(dir -1 $B0/b0/1/$gfid | wc -l) -eq 1 ]
+
+# check that the backend file matches the frontend
+TEST [ "$(md5sum $B0/b0/1/$gfid | awk '{print $1}')" == "$checksum" ]
+
+# delete the file
+TEST rm $M0/3
+
+# ensure the frontend and backend files are cleaned up
+TEST ! -e $M0/3
+TEST ! [ stat $B0/b*/*/$gfid ]
+
+# Test TRUNCATE + WRITE flow
+echo "hello" | tee $M0/4
+echo "goo" | tee $M0/4
+filename=$$
+echo "goo" | tee /tmp/metadisp-truncate-${filename}
+checksum=$(md5sum /tmp/metadisp-truncate-${filename} | awk '{print $1}')
+TEST [ "$(md5sum $M0/4 | awk '{print $1}')" == "$checksum" ]
+
+# Test mkdir + rmdir.
+TEST mkdir $M0/rmdir_me
+nfiles=$(ls -d $B0/b*/*/rmdir_me 2> /dev/null | wc -l)
+TEST [ "$nfiles" = "1" ]
+TEST rmdir $M0/rmdir_me
+nfiles=$(ls -d $B0/b*/*/rmdir_me 2> /dev/null | wc -l)
+TEST [ "$nfiles" = "0" ]
+
+# Test rename.
+TEST touch $M0/rename_me
+nfiles=$(ls $B0/b*/*/rename_me 2> /dev/null | wc -l)
+TEST [ "$nfiles" = "1" ]
+nfiles=$(ls $B0/b*/*/such_rename 2> /dev/null | wc -l)
+TEST [ "$nfiles" = "0" ]
+TEST mv $M0/rename_me $M0/such_rename
+nfiles=$(ls $B0/b*/*/rename_me 2> /dev/null | wc -l)
+TEST [ "$nfiles" = "0" ]
+nfiles=$(ls $B0/b*/*/such_rename 2> /dev/null | wc -l)
+TEST [ "$nfiles" = "1" ]
+
+# Test rename of a file that doesn't exist.
+TEST ! mv $M0/does-not-exist $M0/neither-does-this
+
+
+# cleanup all the other files.
+TEST rm -v $M0/1 $M0/2 $M0/{4..${NUM_FILES}}
+TEST rm $M0/such_rename
+TEST [ $(ls /d/backends/b0/0/ | wc -l) -eq 0 ]
+TEST [ $(ls /d/backends/b0/1/ | wc -l) -eq 0 ]
+
+# Test CREATE flow
+NUM_FILES=40
+TEST touch $M0/{1..${NUM_FILES}}
+TEST [ $(ls /d/backends/b0/0/ | wc -l) -eq $NUM_FILES ]
+TEST [ $(ls /d/backends/b0/1/ | wc -l) -eq $NUM_FILES ]
+
+# Test UNLINK flow
+# No drives should have any files
+TEST rm -v $M0/{1..${NUM_FILES}}
+TEST [ $(ls /d/backends/b0/0/ | wc -l) -eq 0 ]
+TEST [ $(ls /d/backends/b0/1/ | wc -l) -eq 0 ]
+
+# Test CREATE + WRITE + READ flow
+filename=$$
+dd if=/dev/urandom of=/tmp/${filename} bs=1M count=10
+checksum=$(md5sum /tmp/${filename} | awk '{print $1}')
+TEST cp -v /tmp/${filename} $M0/1
+TEST cp -v /tmp/${filename} $M0/2
+TEST cp -v /tmp/${filename} $M0/3
+TEST cp -v /tmp/${filename} $M0/4
+TEST [ "$(md5sum $M0/1 | awk '{print $1}')" == "$checksum" ]
+TEST [ "$(md5sum $M0/2 | awk '{print $1}')" == "$checksum" ]
+TEST [ "$(md5sum $M0/3 | awk '{print $1}')" == "$checksum" ]
+TEST [ "$(md5sum $M0/4 | awk '{print $1}')" == "$checksum" ]
+
+# Test TRUNCATE + WRITE flow
+TEST dd if=/dev/zero of=$M0/1 bs=1M count=20
+
+# Check that readdir stats the files properly and we get the correct sizes
+TEST [ $(find $M0 -size +9M | wc -l) -eq 4 ];
+
+# Test mkdir + rmdir.
+TEST mkdir $M0/rmdir_me
+nfiles=$(ls -d $B0/b*/*/rmdir_me 2> /dev/null | wc -l)
+TEST [ "$nfiles" = "1" ]
+TEST rmdir $M0/rmdir_me
+nfiles=$(ls -d $B0/b*/*/rmdir_me 2> /dev/null | wc -l)
+TEST [ "$nfiles" = "0" ]
+
+# Test rename.
+# Still flaky, so disabled until it can be debugged.
+TEST touch $M0/rename_me
+nfiles=$(ls $B0/b*/*/rename_me 2> /dev/null | wc -l)
+TEST [ "$nfiles" = "1" ]
+nfiles=$(ls $B0/b*/*/such_rename 2> /dev/null | wc -l)
+TEST [ "$nfiles" = "0" ]
+TEST mv $M0/rename_me $M0/such_rename
+nfiles=$(ls $B0/b*/*/rename_me 2> /dev/null | wc -l)
+TEST [ "$nfiles" = "0" ]
+nfiles=$(ls $B0/b*/*/such_rename 2> /dev/null | wc -l)
+TEST [ "$nfiles" = "1" ]
+
+# Test rename of a file that doesn't exist.
+TEST ! mv $M0/does-not-exist $M0/neither-does-this
+
+# Test rename over an existing file.
+ok=yes
+for i in $(seq 0 9); do
+ echo foo > $M0/src$i
+ echo bar > $M0/dst$i
+done
+for i in $(seq 0 9); do
+ mv $M0/src$i $M0/dst$i
+done
+for i in $(seq 0 9); do
+ nfiles=$(cat $B0/b0/*/dst$i | wc -l)
+ if [ "$nfiles" = "2" ]; then
+ echo "COLLISION on dst$i"
+ (ls -l $B0/b0/*/dst$i; cat $B0/b0/*/dst$i) | sed "/^/s// /"
+ ok=no
+ fi
+done
+EXPECT "yes" echo $ok
+
+# Test rename of a directory.
+count_copies () {
+ ls -d $B0/b?/?/$1 2> /dev/null | wc -l
+}
+TEST mkdir $M0/foo_dir
+EXPECT 1 count_copies foo_dir
+EXPECT 0 count_copies bar_dir
+TEST mv $M0/foo_dir $M0/bar_dir
+EXPECT 0 count_copies foo_dir
+EXPECT 1 count_copies bar_dir
+
+for x in $(seq 0 99); do
+ touch $M0/target$x
+ ln -s $M0/target$x $M0/link$x
+done
+on_0=$(ls $B0/b*/0/link* | wc -l)
+on_1=$(ls $B0/b*/1/link* | wc -l)
+TEST [ "$on_0" -eq 100 ]
+TEST [ "$on_1" -eq 0 ]
+TEST [ "$(ls -l $M0/link* | wc -l)" = 100 ]
+
+# Test (hard) link.
+_test_hardlink () {
+ local b
+ local has_src
+ local has_dst
+ local src_inum
+ local dst_inum
+ touch $M0/hardsrc$1
+ ln $M0/hardsrc$1 $M0/harddst$1
+ for b in $B0/b{0}/{0,1}; do
+ [ -f $b/hardsrc$1 ]; has_src=$?
+ [ -f $b/harddst$1 ]; has_dst=$?
+ if [ "$has_src" != "$has_dst" ]; then
+ echo "MISSING $b/hardxxx$1 $has_src $has_dst"
+ return
+ fi
+ if [ "$has_src$has_dst" = "00" ]; then
+ src_inum=$(stat -c '%i' $b/hardsrc$1)
+ dst_inum=$(stat -c '%i' $b/harddst$1)
+ if [ "$dst_inum" != "$src_inum" ]; then
+ echo "MISMATCH $b/hardxx$i $src_inum $dst_inum"
+ return
+ fi
+ fi
+ done
+ echo "OK"
+}
+
+test_hardlink () {
+ local result=$(_test_hardlink $*)
+ # [ "$result" = "OK" ] || echo $result > /dev/tty
+ echo $result
+}
+
+# Do this multiple times to make sure colocation isn't a fluke.
+EXPECT "OK" test_hardlink 0
+EXPECT "OK" test_hardlink 1
+EXPECT "OK" test_hardlink 2
+EXPECT "OK" test_hardlink 3
+EXPECT "OK" test_hardlink 4
+EXPECT "OK" test_hardlink 5
+EXPECT "OK" test_hardlink 6
+EXPECT "OK" test_hardlink 7
+EXPECT "OK" test_hardlink 8
+EXPECT "OK" test_hardlink 9
+
+# Test remove hardlink source. ensure deleting one file
+# doesn't delete the data unless link-count is 1
+TEST mkdir $M0/hardlink
+TEST touch $M0/hardlink/fileA
+echo "data" >> $M0/hardlink/fileA
+checksum=$(md5sum $M0/hardlink/fileA | awk '{print $1}')
+TEST ln $M0/hardlink/fileA $M0/hardlink/fileB
+TEST [ $(dir -1 $M0/hardlink/ | wc -l) -eq 2 ]
+TEST rm $M0/hardlink/fileA
+TEST [ $(dir -1 $M0/hardlink/ | wc -l) -eq 1 ]
+TEST [ "$(md5sum $M0/hardlink/fileB | awk '{print $1}')" == "$checksum" ]
+
+#
+# FIXME: statfs values look ok but the test is bad
+#
+# Test statfs. If we're doing it right, the numbers for the mountpoint should be
+# double those for the brick filesystem times the number of bricks,
+# but unless we're on a completely idle
+# system (which never happens) the numbers can change even while this function
+# runs and that would trip us up. Do a sloppy comparison to deal with that.
+#compare_fields () {
+# val1=$(df $1 | grep / | awk "{print \$$3}")
+# val2=$(df $2 | grep / | awk "{print \$$3}")
+# [ "$val2" -gt "$(((val1/(29/10))*19/10))" -a "$val2" -lt "$(((val1/(31/10))*21/10))" ]
+#}
+
+#brick_df=$(df $B0 | grep /)
+#mount_df=$(df $M0 | grep /)
+#TEST compare_fields $B0 $M0 2 # Total blocks
+#TEST compare_fields $B0 $M0 3 # Used
+#TEST compare_fields $B0 $M0 4 # Available
+
+# Test removexattr.
+#RXATTR_FILE=$(get_file_not_on_disk0 rxtest)
+#TEST setfattr -n user.foo -v bar $M0/$RXATTR_FILE
+#TEST getfattr -n user.foo $B0/b0/1/$RXATTR_FILE
+#TEST setfattr -x user.foo $M0/$RXATTR_FILE
+#TEST ! getfattr -n user.foo $B0/b0/1/$RXATTR_FILE
+
+# Test fsyncdir. We can't really test whether it's doing the right thing,
+# but we can test that it doesn't fail and we can hand-check that it's calling
+# down to all of the disks instead of just one.
+#
+# P.S. There's no fsyncdir test in the rest of Gluster, so who even knows if
+# other translators are handling it correctly?
+
+#FSYNCDIR_EXE=$(dirname $0)/fsyncdir
+#build_tester ${FSYNCDIR_EXE}.c
+#TEST touch $M0/fsyncdir_src
+#TEST $FSYNCDIR_EXE $M0 $M0/fsyncdir_src $M0/fsyncdir_dst
+#TEST rm -f $FSYNCDIR_EXE
+
+# Test fsetxattr, fgetxattr, fremovexattr (in that order).
+FXATTR_FILE=$M0/fxfile1
+TEST touch $FXATTR_FILE
+FXATTR_EXE=$(dirname $0)/fxattr
+build_tester ${FXATTR_EXE}.c
+TEST ! getfattr -n user.fxtest $FXATTR_FILE
+TEST $FXATTR_EXE $FXATTR_FILE set value1
+TEST getfattr -n user.fxtest $FXATTR_FILE
+TEST setfattr -n user.fxtest -v value2 $FXATTR_FILE
+TEST $FXATTR_EXE $FXATTR_FILE get value2
+TEST $FXATTR_EXE $FXATTR_FILE remove
+TEST ! getfattr -n user.fxtest $FXATTR_FILE
+TEST rm -f $FXATTR_EXE
+
+# Test ftruncate
+FTRUNCATE_EXE=$(dirname $0)/ftruncate
+build_tester ${FTRUNCATE_EXE}.c
+FTRUNCATE_FILE=$M0/ftfile1
+TEST dd if=/dev/urandom of=$FTRUNCATE_FILE count=1 bs=1MB
+TEST $FTRUNCATE_EXE $FTRUNCATE_FILE
+#gfid=$(gf_gfid_xattr_to_str $(gf_get_gfid_xattr $B0/b0/*/ftfile1))
+
+# Test fallocate, discard, zerofill. Actually we don't so much check that these
+# *work* as that they don't throw any errors (especially ENOENT because the
+# file's not on disk zero).
+FALLOC_FILE=fatest1
+TEST touch $M0/$FALLOC_FILE
+TEST fallocate -l $((4096*5)) $M0/$FALLOC_FILE
+TEST fallocate -p -o 4096 -l 4096 $M0/$FALLOC_FILE
+# This actually fails with "operation not supported" on most filesystems, so
+# don't leave it enabled except to test changes.
+#TEST fallocate -z -o $((4096*3)) -l 4096 $M0/$FALLOC_FILE
+
+#cleanup;
diff --git a/tests/basic/metadisp/metadisp.vol b/tests/basic/metadisp/metadisp.vol
new file mode 100644
index 00000000000..58ae2f6f2a8
--- /dev/null
+++ b/tests/basic/metadisp/metadisp.vol
@@ -0,0 +1,14 @@
+volume posix-0
+ type storage/posix
+ option directory /d/backends/b0/0
+end-volume
+
+volume posix-1
+ type storage/posix
+ option directory /d/backends/b0/1
+end-volume
+
+volume metadisp-0
+ type features/metadisp
+ subvolumes posix-0 posix-1
+end-volume
diff --git a/tests/basic/quick-read-with-upcall.t b/tests/basic/quick-read-with-upcall.t
index 2012053d125..dfb751dfcdb 100644
--- a/tests/basic/quick-read-with-upcall.t
+++ b/tests/basic/quick-read-with-upcall.t
@@ -15,8 +15,8 @@ TEST $CLI volume create $V0 $H0:$B0/${V0}{1..2};
TEST $CLI volume start $V0
# Mount FUSE without selinux:
-TEST glusterfs -s $H0 --volfile-id $V0 $M0;
-TEST glusterfs -s $H0 --volfile-id $V0 $M1;
+TEST glusterfs -s $H0 --volfile-id $V0 --direct-io-mode=enable $M0;
+TEST glusterfs -s $H0 --volfile-id $V0 --direct-io-mode=enable $M1;
D0="test-message0";
D1="test-message1";
@@ -37,16 +37,13 @@ TEST write_to "$M0/test.txt" "$D1"
EXPECT "$D1" cat $M0/test.txt
EXPECT "$D0" cat $M1/test.txt
-# This is 3.7 for no good reason. We could have kept this to
-# any number above 2 seconds. Noticed that when it is 2 seconds, or
-# less, there is a possibility of not getting a lookup on the same inode.
-sleep 3.7
+sleep 1
# TODO: This line normally fails
EXPECT "$D1" cat $M1/test.txt
TEST $CLI volume set $V0 features.cache-invalidation on
-TEST $CLI volume set $V0 performance.qr-cache-timeout 15
+TEST $CLI volume set $V0 performance.quick-read-cache-timeout 15
TEST $CLI volume set $V0 performance.md-cache-timeout 15
TEST write_to "$M0/test1.txt" "$D0"
@@ -73,7 +70,3 @@ EXPECT "$D0" cat $M1/test2.txt
TEST write_to "$M0/test2.txt" "$D1"
EXPECT "$D1" cat $M0/test2.txt
EXPECT "$D1" cat $M1/test2.txt
-
-
-#G_TESTDEF_TEST_STATUS_CENTOS6=BAD_TEST,BUG=1718191
-#G_TESTDEF_TEST_STATUS_NETBSD7=BAD_TEST,BUG=1718191
diff --git a/tests/bugs/distribute/bug-1600379.t b/tests/bugs/distribute/bug-1600379.t
new file mode 100644
index 00000000000..8d2f6154100
--- /dev/null
+++ b/tests/bugs/distribute/bug-1600379.t
@@ -0,0 +1,54 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+
+# Initialize
+#------------------------------------------------------------
+cleanup;
+
+# Start glusterd
+TEST glusterd;
+TEST pidof glusterd;
+TEST $CLI volume info;
+
+# Create a volume
+TEST $CLI volume create $V0 $H0:$B0/${V0}{1,2}
+
+# Verify volume creation
+EXPECT "$V0" volinfo_field $V0 'Volume Name';
+EXPECT 'Created' volinfo_field $V0 'Status';
+
+# Start volume and verify successful start
+TEST $CLI volume start $V0;
+EXPECT 'Started' volinfo_field $V0 'Status';
+TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 --entry-timeout=0 $M0;
+#------------------------------------------------------------
+
+# Test case - Remove xattr from killed brick on lookup
+#------------------------------------------------------------
+# Create a dir and set custom xattr
+TEST mkdir $M0/testdir
+TEST setfattr -n user.attr -v val $M0/testdir
+xattr_val=`getfattr -d $B0/${V0}2/testdir | awk '{print $1}'`;
+TEST ${xattr_val}='user.attr="val"';
+
+# Kill 2nd brick process
+TEST kill_brick $V0 $H0 $B0/${V0}2
+EXPECT_WITHIN ${PROCESS_UP_TIMEOUT} "1" online_brick_count
+
+# Remove custom xattr
+TEST setfattr -x user.attr $M0/testdir
+
+# Bring up the killed brick process
+TEST $CLI volume start $V0 force
+
+# Perform lookup
+sleep 5
+TEST ls $M0/testdir
+
+# Check brick xattrs
+xattr_val_2=`getfattr -d $B0/${V0}2/testdir`;
+TEST [ ${xattr_val_2} = ''] ;
+
+cleanup;
diff --git a/tests/bugs/distribute/issue-1327.t b/tests/bugs/distribute/issue-1327.t
new file mode 100755
index 00000000000..acd8c8c6614
--- /dev/null
+++ b/tests/bugs/distribute/issue-1327.t
@@ -0,0 +1,33 @@
+#!/bin/bash
+
+SCRIPT_TIMEOUT=250
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../dht.rc
+
+cleanup
+
+TEST glusterd
+TEST pidof glusterd
+
+BRICK1=$B0/${V0}-0
+BRICK2=$B0/${V0}-1
+
+TEST $CLI volume create $V0 $H0:$BRICK1 $H0:$BRICK2
+TEST $CLI volume start $V0
+
+TEST glusterfs -s $H0 --volfile-id $V0 $M0
+TEST mkdir $M0/dir
+
+#remove dir from one of the brick
+TEST rmdir $BRICK2/dir
+
+#safe cache timeout for lookup to be triggered
+sleep 2
+
+TEST ls $M0/dir
+
+TEST stat $BRICK2/dir
+
+cleanup
diff --git a/tests/bugs/fuse/many-groups-for-acl.t b/tests/bugs/fuse/many-groups-for-acl.t
index d959f750ee0..a51b1bc7267 100755
--- a/tests/bugs/fuse/many-groups-for-acl.t
+++ b/tests/bugs/fuse/many-groups-for-acl.t
@@ -38,6 +38,13 @@ do
done
TEST useradd -o -M -u ${NEW_UID} -g ${NEW_GID} -G ${NEW_USER}-${NEW_GIDS} ${NEW_USER}
+# Linux < 3.8 exports only first 32 gids of pid to userspace
+kernel_exports_few_gids=0
+if [ "$OSTYPE" = Linux ] && \
+ su -m ${NEW_USER} -c "grep ^Groups: /proc/self/status | wc -w | xargs -I@ expr @ - 1 '<' $LAST_GID - $NEW_GID + 1" > /dev/null; then
+ kernel_exports_few_gids=1
+fi
+
# preparation done, start the tests
TEST glusterd
@@ -48,6 +55,8 @@ TEST $CLI volume set $V0 nfs.disable off
TEST $CLI volume set ${V0} server.manage-gids off
TEST $CLI volume start ${V0}
+# This is just a synchronization hack to make sure the bricks are
+# up before going on.
EXPECT_WITHIN ${NFS_EXPORT_TIMEOUT} "1" is_nfs_export_available
# mount the volume with POSIX ACL support, without --resolve-gids
@@ -69,8 +78,8 @@ TEST [ $? -eq 0 ]
su -m ${NEW_USER} -c "touch ${M0}/first-32-gids-2/success > /dev/null"
TEST [ $? -eq 0 ]
-su -m ${NEW_USER} -c "touch ${M0}/gid-64/failure > /dev/null"
-TEST [ $? -ne 0 ]
+su -m ${NEW_USER} -c "touch ${M0}/gid-64/success--if-all-gids-exported > /dev/null"
+TEST [ $? -eq $kernel_exports_few_gids ]
su -m ${NEW_USER} -c "touch ${M0}/gid-120/failure > /dev/null"
TEST [ $? -ne 0 ]
diff --git a/tests/bugs/glusterd/brick-mux-validation-in-cluster.t b/tests/bugs/glusterd/brick-mux-validation-in-cluster.t
index f088dbb426c..b6af487a791 100644
--- a/tests/bugs/glusterd/brick-mux-validation-in-cluster.t
+++ b/tests/bugs/glusterd/brick-mux-validation-in-cluster.t
@@ -100,10 +100,8 @@ $CLI_2 volume set $V0 performance.readdir-ahead on
$CLI_2 volume set $V1 performance.readdir-ahead on
TEST $glusterd_1;
+EXPECT_WITHIN $PROBE_TIMEOUT 2 peer_count
-sleep 10
-
-EXPECT 4 count_brick_processes
EXPECT_WITHIN $PROCESS_UP_TIMEOUT 4 count_brick_pids
EXPECT_WITHIN $PROCESS_UP_TIMEOUT 0 count_N/A_brick_pids
diff --git a/tests/bugs/glusterd/brick-mux.t b/tests/bugs/glusterd/brick-mux.t
index eeaa3ebfea8..927940534c1 100644
--- a/tests/bugs/glusterd/brick-mux.t
+++ b/tests/bugs/glusterd/brick-mux.t
@@ -39,7 +39,7 @@ TEST glusterd
EXPECT_WITHIN $PROCESS_UP_TIMEOUT 4 online_brick_count
EXPECT 1 count_brick_processes
-TEST $CLI volume set $V1 performance.cache-size 32MB
+TEST $CLI volume set $V1 performance.io-cache-size 32MB
TEST $CLI volume stop $V1
TEST $CLI volume start $V1
diff --git a/tests/bugs/glusterd/brick-order-check-add-brick.t b/tests/bugs/glusterd/brick-order-check-add-brick.t
new file mode 100644
index 00000000000..0be31dac768
--- /dev/null
+++ b/tests/bugs/glusterd/brick-order-check-add-brick.t
@@ -0,0 +1,61 @@
+#!/bin/bash
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+. $(dirname $0)/../../cluster.rc
+. $(dirname $0)/../../snapshot.rc
+
+cleanup;
+
+TEST verify_lvm_version;
+#Create cluster with 3 nodes
+TEST launch_cluster 3 -NO_DEBUG -NO_FORCE
+TEST setup_lvm 3
+
+TEST $CLI_1 peer probe $H2
+TEST $CLI_1 peer probe $H3
+EXPECT_WITHIN $PROBE_TIMEOUT 2 peer_count
+
+TEST $CLI_1 volume create $V0 replica 3 $H1:$L1/$V0 $H2:$L2/$V0 $H3:$L3/$V0
+EXPECT '1 x 3 = 3' volinfo_field $V0 'Number of Bricks'
+EXPECT 'Created' volinfo_field $V0 'Status'
+
+TEST $CLI_1 volume start $V0
+EXPECT 'Started' volinfo_field $V0 'Status'
+
+#add-brick with or without mentioning the replica count should not fail
+TEST $CLI_1 volume add-brick $V0 replica 3 $H1:$L1/${V0}_1 $H2:$L2/${V0}_1 $H3:$L3/${V0}_1
+EXPECT '2 x 3 = 6' volinfo_field $V0 'Number of Bricks'
+
+TEST $CLI_1 volume add-brick $V0 $H1:$L1/${V0}_2 $H2:$L2/${V0}_2 $H3:$L3/${V0}_2
+EXPECT '3 x 3 = 9' volinfo_field $V0 'Number of Bricks'
+
+#adding bricks from same host should fail the brick order check
+TEST ! $CLI_1 volume add-brick $V0 $H1:$L1/${V0}_3 $H1:$L1/${V0}_4 $H1:$L1/${V0}_5
+EXPECT '3 x 3 = 9' volinfo_field $V0 'Number of Bricks'
+
+#adding bricks from same host with force should succeed
+TEST $CLI_1 volume add-brick $V0 $H1:$L1/${V0}_3 $H1:$L1/${V0}_4 $H1:$L1/${V0}_5 force
+EXPECT '4 x 3 = 12' volinfo_field $V0 'Number of Bricks'
+
+TEST $CLI_1 volume stop $V0
+TEST $CLI_1 volume delete $V0
+
+TEST $CLI_1 volume create $V0 replica 2 $H1:$L1/${V0}1 $H2:$L2/${V0}1
+EXPECT '1 x 2 = 2' volinfo_field $V0 'Number of Bricks'
+EXPECT 'Created' volinfo_field $V0 'Status'
+
+TEST $CLI_1 volume start $V0
+EXPECT 'Started' volinfo_field $V0 'Status'
+
+#Add-brick with Increasing replica count
+TEST $CLI_1 volume add-brick $V0 replica 3 $H3:$L3/${V0}1
+EXPECT '1 x 3 = 3' volinfo_field $V0 'Number of Bricks'
+
+#Add-brick with Increasing replica count from same host should fail
+TEST ! $CLI_1 volume add-brick $V0 replica 5 $H1:$L1/${V0}2 $H1:$L1/${V0}3
+
+#adding multiple bricks from same host should fail the brick order check
+TEST ! $CLI_1 volume add-brick $V0 replica 3 $H1:$L1/${V0}{4..6} $H2:$L2/${V0}{7..9}
+EXPECT '1 x 3 = 3' volinfo_field $V0 'Number of Bricks'
+
+cleanup
diff --git a/tests/bugs/glusterd/check_elastic_server.t b/tests/bugs/glusterd/check_elastic_server.t
index 8e9e4e5b0eb..41d2140aa2b 100644
--- a/tests/bugs/glusterd/check_elastic_server.t
+++ b/tests/bugs/glusterd/check_elastic_server.t
@@ -23,6 +23,9 @@ EXPECT 'Created' cluster_volinfo_field 1 $V0 'Status';
$CLI_1 volume start $V0
EXPECT 'Started' cluster_volinfo_field 1 $V0 'Status';
+#Mount invalid volume
+TEST ! glusterfs -s $H1 --volfile-id=$V0_NA $M0;
+
#Mount FUSE
TEST glusterfs -s $H1 --volfile-id=$V0 $M0;
diff --git a/tests/bugs/glusterd/mgmt-handshake-and-volume-sync-post-glusterd-restart.t b/tests/bugs/glusterd/mgmt-handshake-and-volume-sync-post-glusterd-restart.t
index fdc0a73f60c..8001359e6b3 100644
--- a/tests/bugs/glusterd/mgmt-handshake-and-volume-sync-post-glusterd-restart.t
+++ b/tests/bugs/glusterd/mgmt-handshake-and-volume-sync-post-glusterd-restart.t
@@ -4,7 +4,7 @@
. $(dirname $0)/../../cluster.rc
function check_peers {
-$CLI_1 peer status | grep 'Peer in Cluster (Connected)' | wc -l
+eval \$CLI_$1 peer status | grep 'Peer in Cluster (Connected)' | wc -l
}
cleanup
@@ -36,23 +36,35 @@ TEST [[ $OP_VERS_ORIG == $OP_VERS_NEW ]]
#bug-948686 - volume sync after bringing up the killed node
TEST $CLI_1 peer probe $H3
-EXPECT_WITHIN $PROBE_TIMEOUT 2 check_peers;
+EXPECT_WITHIN $PROBE_TIMEOUT 2 check_peers 1
+EXPECT_WITHIN $PROBE_TIMEOUT 2 check_peers 2
+EXPECT_WITHIN $PROBE_TIMEOUT 2 check_peers 3
TEST $CLI_1 volume create $V0 replica 2 $H1:$B1/$V0 $H1:$B1/${V0}_1 $H2:$B2/$V0 $H3:$B3/$V0
TEST $CLI_1 volume start $V0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT 'Started' volinfo_field_1 $V0 'Status'
TEST glusterfs --volfile-server=$H1 --volfile-id=$V0 $M0
#kill a node
TEST kill_node 3
+EXPECT_WITHIN $PROBE_TIMEOUT 1 check_peers 1
+EXPECT_WITHIN $PROBE_TIMEOUT 1 check_peers 2
#modify volume config to see change in volume-sync
TEST $CLI_1 volume set $V0 write-behind off
#add some files to the volume to see effect of volume-heal cmd
TEST touch $M0/{1..100};
TEST $CLI_1 volume stop $V0;
+EXPECT_WITHIN $PROCESS_DOWN_TIMEOUT 'Stopped' volinfo_field_1 $V0 'Status'
+
TEST $glusterd_3;
-EXPECT_WITHIN $PROBE_TIMEOUT 2 check_peers
+EXPECT_WITHIN $PROBE_TIMEOUT 2 check_peers 1
+EXPECT_WITHIN $PROBE_TIMEOUT 2 check_peers 2
+EXPECT_WITHIN $PROBE_TIMEOUT 2 check_peers 3
+
+sleep 5
TEST $CLI_3 volume start $V0;
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT 'Started' volinfo_field_1 $V0 'Status'
TEST $CLI_2 volume stop $V0;
TEST $CLI_2 volume delete $V0;
diff --git a/tests/bugs/glusterd/enable-shared-storage-and-remove-brick-validation.t b/tests/bugs/glusterd/remove-brick-validation.t
index 11ed0d94d79..a0ff4ff6a24 100644
--- a/tests/bugs/glusterd/enable-shared-storage-and-remove-brick-validation.t
+++ b/tests/bugs/glusterd/remove-brick-validation.t
@@ -18,20 +18,6 @@ TEST $CLI_1 peer probe $H2;
EXPECT_WITHIN $PROBE_TIMEOUT 1 peer_count 1
-#test case for bug 1266818 - disabling enable-shared-storage option
-##should not delete user created volume with name glusterd_shared_storage
-
-## creating a volume with name glusterd_shared_storage
-TEST $CLI_1 volume create glusterd_shared_storage $H1:$B1/${V0}0 $H2:$B2/${V0}1
-TEST $CLI_1 volume start glusterd_shared_storage
-
-## disabling enable-shared-storage should not succeed and should not delete the
-## user created volume with name "glusterd_shared_storage"
-TEST ! $CLI_1 volume all enable-shared-storage disable
-
-## volume with name should exist
-TEST $CLI_1 volume info glusterd_shared_storage
-
#testcase: bug-1245045-remove-brick-validation
TEST $CLI_1 peer probe $H3;
diff --git a/tests/bugs/glusterd/validating-options-for-replicated-volume.t b/tests/bugs/glusterd/validating-options-for-replicated-volume.t
index ee231338ff1..ddc80b17870 100644
--- a/tests/bugs/glusterd/validating-options-for-replicated-volume.t
+++ b/tests/bugs/glusterd/validating-options-for-replicated-volume.t
@@ -61,10 +61,15 @@ TEST ! $CLI volume set $V0 background-self-heal-count " "
TEST $CLI volume set $V0 background-self-heal-count 10
EXPECT "10" volume_option $V0 cluster.background-self-heal-count
-TEST ! $CLI volume set $V0 cache-size ""
-TEST ! $CLI volume set $V0 cache-size " "
-TEST $CLI volume set $V0 cache-size 512MB
-EXPECT "512MB" volume_option $V0 performance.cache-size
+TEST ! $CLI volume set $V0 io-cache-size ""
+TEST ! $CLI volume set $V0 io-cache-size " "
+TEST $CLI volume set $V0 io-cache-size 64MB
+EXPECT "64MB" volume_option $V0 performance.io-cache-size
+
+TEST ! $CLI volume set $V0 quick-read-cache-size ""
+TEST ! $CLI volume set $V0 quick-read-cache-size " "
+TEST $CLI volume set $V0 quick-read-cache-size 512MB
+EXPECT "512MB" volume_option $V0 performance.quick-read-cache-size
TEST ! $CLI volume set $V0 self-heal-daemon ""
TEST ! $CLI volume set $V0 self-heal-daemon " "
diff --git a/tests/bugs/glusterfs-server/bug-873549.t b/tests/bugs/glusterfs-server/bug-873549.t
index a3b2f9c9bf7..8b5534728fd 100644
--- a/tests/bugs/glusterfs-server/bug-873549.t
+++ b/tests/bugs/glusterfs-server/bug-873549.t
@@ -10,7 +10,7 @@ TEST $CLI volume info;
TEST $CLI volume create $V0 $H0:$B0/${V0}{1,2};
-TEST $CLI volume set $V0 performance.cache-size 512MB
+TEST $CLI volume set $V0 performance.quick-read-cache-size 512MB
TEST $CLI volume start $V0
TEST $CLI volume statedump $V0 all
diff --git a/tests/bugs/logging/bug-823081.t b/tests/bugs/logging/bug-823081.t
index 0ed8f4c26c1..bd1965d2d49 100755
--- a/tests/bugs/logging/bug-823081.t
+++ b/tests/bugs/logging/bug-823081.t
@@ -22,20 +22,20 @@ function set_tail ()
set_tail $V0;
TEST $CLI volume create $V0 $H0:$B0/${V0}{1,2};
-tail=`tail -n 1 $logdir/$cmd_log_history | cut -d " " -f 5-`
+tail=`tail -n 1 $logdir/$cmd_log_history | cut -d " " -f 6-`
TEST [[ \"$tail\" == \"$tail_success\" ]]
TEST ! $CLI volume create $V0 $H0:$B0/${V0}{1,2};
-tail=`tail -n 1 $logdir/$cmd_log_history | cut -d " " -f 5-`
+tail=`tail -n 1 $logdir/$cmd_log_history | cut -d " " -f 6-`
TEST [[ \"$tail\" == \"$tail_failure\" ]]
set_tail $V1;
TEST gluster volume create $V1 $H0:$B0/${V1}{1,2} force;
-tail=`tail -n 1 $logdir/$cmd_log_history | cut -d " " -f 5-`
+tail=`tail -n 1 $logdir/$cmd_log_history | cut -d " " -f 6-`
TEST [[ \"$tail\" == \"$tail_success_force\" ]]
TEST ! gluster volume create $V1 $H0:$B0/${V1}{1,2} force;
-tail=`tail -n 1 $logdir/$cmd_log_history | cut -d " " -f 5-`
+tail=`tail -n 1 $logdir/$cmd_log_history | cut -d " " -f 6-`
TEST [[ \"$tail\" == \"$tail_failure_force\" ]]
cleanup;
diff --git a/tests/bugs/replicate/bug-1101647.t b/tests/bugs/replicate/bug-1101647.t
index 8f420eec012..708bc1a1e29 100644
--- a/tests/bugs/replicate/bug-1101647.t
+++ b/tests/bugs/replicate/bug-1101647.t
@@ -12,6 +12,8 @@ TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{1,2};
TEST $CLI volume start $V0;
TEST glusterfs --volfile-id=/$V0 --volfile-server=$H0 $M0 --attribute-timeout=0 --entry-timeout=0
EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" glustershd_up_status
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 0
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 1
#Create base entry in indices/xattrop
echo "Data">$M0/file
diff --git a/tests/bugs/replicate/bug-1744548-heal-timeout.t b/tests/bugs/replicate/bug-1744548-heal-timeout.t
index c208112c8b0..011535066f9 100644
--- a/tests/bugs/replicate/bug-1744548-heal-timeout.t
+++ b/tests/bugs/replicate/bug-1744548-heal-timeout.t
@@ -25,14 +25,14 @@ TEST ! $CLI volume heal $V0
TEST $CLI volume profile $V0 start
TEST $CLI volume profile $V0 info clear
TEST $CLI volume heal $V0 enable
-# Each brick does 3 opendirs, corresponding to dirty, xattrop and entry-changes
-EXPECT_WITHIN $HEAL_TIMEOUT "^333$" get_cumulative_opendir_count
+# Each brick does 4 opendirs, corresponding to dirty, xattrop and entry-changes, anonymous-inode
+EXPECT_WITHIN 4 "^444$" get_cumulative_opendir_count
# Check that a change in heal-timeout is honoured immediately.
TEST $CLI volume set $V0 cluster.heal-timeout 5
sleep 10
# Two crawls must have happened.
-EXPECT_WITHIN $HEAL_TIMEOUT "^999$" get_cumulative_opendir_count
+EXPECT_WITHIN $CHILD_UP_TIMEOUT "^121212$" get_cumulative_opendir_count
# shd must not heal if it is disabled and heal-timeout is changed.
TEST $CLI volume heal $V0 disable
diff --git a/tests/cluster.rc b/tests/cluster.rc
index 2d826d07e39..34f5b02398f 100644
--- a/tests/cluster.rc
+++ b/tests/cluster.rc
@@ -11,7 +11,7 @@ function launch_cluster() {
define_backends $count;
define_hosts $count;
define_glusterds $count $2;
- define_clis $count;
+ define_clis $count $3;
start_glusterds;
}
@@ -90,6 +90,20 @@ function kill_glusterd() {
kill `cat $pidfile`;
}
+function restart_glusterd() {
+ local index=$1
+ local b
+ local pidfile
+ local g
+
+ b="B$index"
+ pidfile="${!b}/glusterd.pid"
+
+ kill `cat $pidfile`
+
+ g="glusterd_${index}"
+ ${!g}
+}
function kill_node() {
local index=$1;
@@ -134,8 +148,13 @@ function define_clis() {
lopt1="--log-file=$logdir/$logfile1"
- eval "CLI_$i='$CLI --glusterd-sock=${!b}/glusterd/gd.sock $lopt'";
- eval "CLI$i='$CLI --glusterd-sock=${!b}/glusterd/gd.sock $lopt1'";
+ if [ "$2" == "-NO_FORCE" ]; then
+ eval "CLI_$i='$CLI_NO_FORCE --glusterd-sock=${!b}/glusterd/gd.sock $lopt'";
+ eval "CLI$i='$CLI_NO_FORCE --glusterd-sock=${!b}/glusterd/gd.sock $lopt1'";
+ else
+ eval "CLI_$i='$CLI --glusterd-sock=${!b}/glusterd/gd.sock $lopt'";
+ eval "CLI$i='$CLI --glusterd-sock=${!b}/glusterd/gd.sock $lopt1'";
+ fi
done
}
diff --git a/tests/features/flock_interrupt.t b/tests/features/flock_interrupt.t
index 964a4bc20ef..b8717e30dfb 100644
--- a/tests/features/flock_interrupt.t
+++ b/tests/features/flock_interrupt.t
@@ -28,6 +28,5 @@ flock $M0/testfile sleep 6 & { sleep 0.3; flock -w 2 $M0/testfile true; echo ok
EXPECT_WITHIN 4 ok cat got_lock;
## Finish up
-sleep 7;
rm -f got_lock;
cleanup;
diff --git a/tests/features/interrupt.t b/tests/features/interrupt.t
index bd70ff87545..067eb1b7486 100644
--- a/tests/features/interrupt.t
+++ b/tests/features/interrupt.t
@@ -43,18 +43,22 @@ function test_interrupt {
# If the test helper fails (which is considered a setup error, not failure of the test
# case itself), kill will be invoked without argument, and that will be the actual
# error which is caught.
- TEST "./$(dirname $0)/open_and_sleep $M0/testfile | { sleep 0.1; xargs -n1 kill -INT; }"
+ TEST "./$(dirname $0)/open_and_sleep $M0/testfile-$handlebool | { sleep 0.1; xargs -n1 kill -INT; }"
TEST "grep -E '$logpattern' $log_file"
# Basic sanity check, making sure filesystem has not crashed.
- TEST test -f $M0/testfile
+ TEST test -f $M0/testfile-$handlebool
}
# Theoretically FLUSH might finish before INTERRUPT is handled,
-# in which case we'd get the "no handler found" message (but it's unlikely).
-test_interrupt yes 'FLUSH.*interrupt handler triggered|INTERRUPT.*no handler found'
+# in which case we'd get the "no handler found" message instead of
+# "interrupt handler triggered" (but it's unlikely).
+# If that's observed, the pattern can be changed to
+# 'FLUSH.*interrupt handler triggered|[I]NTERRUPT.*no handler found'
+# to fix the test.
+test_interrupt yes '[F]LUSH.*interrupt handler triggered'
EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0
-test_interrupt no 'INTERRUPT.*no handler found'
+test_interrupt no '[I]NTERRUPT.*no handler found'
## Finish up
TEST $CLI volume stop $V0;
diff --git a/tests/features/ssl-ciphers.t b/tests/features/ssl-ciphers.t
index e4bcdf51095..b70fe360e02 100644
--- a/tests/features/ssl-ciphers.t
+++ b/tests/features/ssl-ciphers.t
@@ -110,8 +110,15 @@ EXPECT "N" openssl_connect -ssl2 -connect $H0:$BRICK_PORT
# Test SSLv3 protocol fails
EXPECT "N" openssl_connect -ssl3 -connect $H0:$BRICK_PORT
-# Test TLSv1 protocol fails
-EXPECT "N" openssl_connect -tls1 -connect $H0:$BRICK_PORT
+# Test TLSv1 protocol based on openssl version
+cmd="openssl version"
+ver=$(eval $cmd | awk -F " " '{print $2}' | grep "^1.1")
+if [ "x${ver}" = "x" ]; then
+ supp="N"
+else
+ supp="Y"
+fi
+EXPECT "${supp}" openssl_connect -tls1 -connect $H0:$BRICK_PORT
# Test a HIGH CBC cipher
cph=`check_cipher -cipher AES256-SHA -connect $H0:$BRICK_PORT`
diff --git a/tests/features/trash.t b/tests/features/trash.t
index 472e909e567..da5b50bc85a 100755
--- a/tests/features/trash.t
+++ b/tests/features/trash.t
@@ -94,105 +94,105 @@ wildcard_not_exists() {
if [ $? -eq 0 ]; then echo "Y"; else echo "N"; fi
}
-# testing glusterd [1-3]
+# testing glusterd
TEST glusterd
TEST pidof glusterd
TEST $CLI volume info
-# creating distributed volume [4]
+# creating distributed volume
TEST $CLI volume create $V0 $H0:$B0/${V0}{1,2}
-# checking volume status [5-7]
+# checking volume status
EXPECT "$V0" volinfo_field $V0 'Volume Name'
EXPECT 'Created' volinfo_field $V0 'Status'
EXPECT '2' brick_count $V0
-# test without enabling trash translator [8]
+# test without enabling trash translator
TEST start_vol $V0 $M0
-# test on enabling trash translator [9-10]
+# test on enabling trash translator
TEST $CLI volume set $V0 features.trash on
EXPECT 'on' volinfo_field $V0 'features.trash'
-# files directly under mount point [11]
+# files directly under mount point
create_files $M0/file1 $M0/file2
TEST file_exists $V0 file1 file2
-# perform unlink [12]
+# perform unlink
TEST unlink_op file1
-# perform truncate [13]
+# perform truncate
TEST truncate_op file2 4
-# create files directory hierarchy and check [14]
+# create files directory hierarchy and check
mkdir -p $M0/1/2/3
create_files $M0/1/2/3/foo1 $M0/1/2/3/foo2
TEST file_exists $V0 1/2/3/foo1 1/2/3/foo2
-# perform unlink [15]
+# perform unlink
TEST unlink_op 1/2/3/foo1
-# perform truncate [16]
+# perform truncate
TEST truncate_op 1/2/3/foo2 4
# create a directory for eliminate pattern
mkdir $M0/a
-# set the eliminate pattern [17-18]
+# set the eliminate pattern
TEST $CLI volume set $V0 features.trash-eliminate-path /a
EXPECT '/a' volinfo_field $V0 'features.trash-eliminate-path'
-# create two files and check [19]
+# create two files and check
create_files $M0/a/test1 $M0/a/test2
TEST file_exists $V0 a/test1 a/test2
-# remove from eliminate pattern [20]
+# remove from eliminate pattern
rm -f $M0/a/test1
EXPECT "Y" wildcard_not_exists $M0/.trashcan/a/test1*
-# truncate from eliminate path [21-23]
+# truncate from eliminate path
truncate -s 2 $M0/a/test2
TEST [ -e $M0/a/test2 ]
TEST [ `ls -l $M0/a/test2 | awk '{print $5}'` -eq 2 ]
EXPECT "Y" wildcard_not_exists $M0/.trashcan/a/test2*
-# set internal op on [24-25]
+# set internal op on
TEST $CLI volume set $V0 features.trash-internal-op on
EXPECT 'on' volinfo_field $V0 'features.trash-internal-op'
-# again create two files and check [26]
+# again create two files and check
create_files $M0/inop1 $M0/inop2
TEST file_exists $V0 inop1 inop2
-# perform unlink [27]
+# perform unlink
TEST unlink_op inop1
-# perform truncate [28]
+# perform truncate
TEST truncate_op inop2 4
-# remove one brick and restart the volume [28-31]
+# remove one brick and restart the volume
TEST $CLI volume remove-brick $V0 $H0:$B0/${V0}2 force
EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0
TEST $CLI volume stop $V0
TEST start_vol $V0 $M0 $M0/.trashcan
-# again create two files and check [33]
+# again create two files and check
create_files $M0/rebal1 $M0/rebal2
TEST file_exists $V0 rebal1 rebal2
-# add one brick [34-35]
+# add one brick
TEST $CLI volume add-brick $V0 $H0:$B0/${V0}3
TEST [ -d $B0/${V0}3 ]
-# perform rebalance [36]
+# perform rebalance
TEST $CLI volume rebalance $V0 start force
EXPECT_WITHIN $REBALANCE_TIMEOUT "0" rebalance_completed
#Find out which file was migrated to the new brick
file_name=$(ls $B0/${V0}3/rebal*| xargs basename)
-# check whether rebalance was succesful [37-40]
+# check whether rebalance was succesful
EXPECT "Y" wildcard_exists $B0/${V0}3/$file_name*
EXPECT "Y" wildcard_exists $B0/${V0}1/.trashcan/internal_op/$file_name*
@@ -201,52 +201,42 @@ EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0
# force required in case rebalance is not over
TEST $CLI volume stop $V0 force
-# create a replicated volume [41]
+# create a replicated volume
TEST $CLI volume create $V1 replica 2 $H0:$B0/${V1}{1,2}
-# checking volume status [42-45]
+# checking volume status
EXPECT "$V1" volinfo_field $V1 'Volume Name'
EXPECT 'Replicate' volinfo_field $V1 'Type'
EXPECT 'Created' volinfo_field $V1 'Status'
EXPECT '2' brick_count $V1
-# enable trash with options and start the replicate volume by disabling automatic self-heal [46-50]
+# enable trash with options and start the replicate volume by disabling automatic self-heal
TEST $CLI volume set $V1 features.trash on
TEST $CLI volume set $V1 features.trash-internal-op on
EXPECT 'on' volinfo_field $V1 'features.trash'
EXPECT 'on' volinfo_field $V1 'features.trash-internal-op'
TEST start_vol $V1 $M1 $M1/.trashcan
-# mount and check for trash directory [51]
+# mount and check for trash directory
TEST [ -d $M1/.trashcan/internal_op ]
-# create a file and check [52]
+# create a file and check
touch $M1/self
TEST [ -e $B0/${V1}1/self -a -e $B0/${V1}2/self ]
-# kill one brick and delete the file from mount point [53-54]
+# kill one brick and delete the file from mount point
kill_brick $V1 $H0 $B0/${V1}1
EXPECT_WITHIN ${PROCESS_UP_TIMEOUT} "1" online_brick_count
rm -f $M1/self
EXPECT "Y" wildcard_exists $B0/${V1}2/.trashcan/self*
-# force start the volume and trigger the self-heal manually [55-57]
-TEST $CLI volume start $V1 force
-EXPECT_WITHIN $PROCESS_UP_TIMEOUT "2" online_brick_count
-EXPECT_WITHIN $PROCESS_UP_TIMEOUT "Y" glustershd_up_status
-# Since we created the file under root of the volume, it will be
-# healed automatically
-
-# check for the removed file in trashcan [58]
-EXPECT_WITHIN $HEAL_TIMEOUT "Y" wildcard_exists $B0/${V1}1/.trashcan/internal_op/self*
-
-# check renaming of trash directory through cli [59-62]
+# check renaming of trash directory through cli
TEST $CLI volume set $V0 trash-dir abc
TEST start_vol $V0 $M0 $M0/abc
TEST [ -e $M0/abc -a ! -e $M0/.trashcan ]
EXPECT "Y" wildcard_exists $B0/${V0}1/abc/internal_op/rebal*
-# ensure that rename and delete operation on trash directory fails [63-65]
+# ensure that rename and delete operation on trash directory fails
rm -rf $M0/abc/internal_op
TEST [ -e $M0/abc/internal_op ]
rm -rf $M0/abc/
diff --git a/tests/geo-rep.rc b/tests/geo-rep.rc
index 84c2c2251c0..9ba4262730e 100644
--- a/tests/geo-rep.rc
+++ b/tests/geo-rep.rc
@@ -1,4 +1,9 @@
GEO_REP_TIMEOUT=120
+CHECK_MOUNT_TIMEOUT=50
+#check for mount point
+function check_mounted () {
+ df | grep $1 | wc -l
+}
function check_status_num_rows()
{
diff --git a/tests/line-coverage/cli-peer-and-volume-operations.t b/tests/line-coverage/cli-peer-and-volume-operations.t
index a1545048773..0cf8dbe81f9 100644
--- a/tests/line-coverage/cli-peer-and-volume-operations.t
+++ b/tests/line-coverage/cli-peer-and-volume-operations.t
@@ -77,6 +77,46 @@ TEST ! $CLI_1 volume status $V0 shd
#test explicitly provided options
TEST $CLI_1 --timeout=120 --log-level=INFO volume status
+#changing timezone to a different one, to check localtime logging feature
+TEST export TZ='Asia/Kolkata'
+TEST restart_glusterd 1
+
+#localtime logging enable
+TEST $CLI_1 volume set all cluster.localtime-logging enable
+EXPECT '1' logging_time_check $LOGDIR
+
+#localtime logging disable
+TEST $CLI_1 volume set all cluster.localtime-logging disable
+EXPECT '0' logging_time_check $LOGDIR
+
+#changing timezone back to original timezone
+TEST export TZ='UTC'
+
+#negative tests for volume options
+#'set' option to enable quota/inode-quota is now depreciated
+TEST ! $CLI_1 volume set $V0 quota enable
+TEST ! $CLI_1 volume set $V0 inode-quota enable
+
+#invalid transport type 'rcp'
+TEST ! $CLI_1 volume set $V0 config.transport rcp
+
+#'op-version' option is not valid for a single volume
+TEST ! $CLI_1 volume set $V0 cluster.op-version 72000
+
+#'op-version' option can't be used with any other option
+TEST ! $CLI_1 volume set all cluster.localtime-logging disable cluster.op-version 72000
+
+#invalid format of 'op-version'
+TEST ! $CLI_1 volume set all cluster.op-version 72-000
+
+#provided 'op-version' value is greater than max allowed op-version
+op_version=$($CLI_1 volume get all cluster.max-op-version | awk 'NR==3 {print$2}')
+op_version=$((op_version+1000)) #this can be any number greater than 0
+TEST ! $CLI_1 volume set all cluster.op-version $op_version
+
+#provided 'op-verison' value cannot be less than the current cluster op-version value
+TEST ! $CLI_1 volume set all cluster.op-version 00000
+
# system commnds
TEST $CLI_1 system help
TEST $CLI_1 system uuid get
diff --git a/tests/line-coverage/log-and-brick-ops-negative-case.t b/tests/line-coverage/log-and-brick-ops-negative-case.t
new file mode 100644
index 00000000000..d86cb452282
--- /dev/null
+++ b/tests/line-coverage/log-and-brick-ops-negative-case.t
@@ -0,0 +1,82 @@
+#!/bin/bash
+
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
+
+cleanup
+TEST glusterd
+TEST pidof glusterd
+
+#create volumes
+TEST $CLI volume create ${V0}_1 $H0:$B0/v{1..2}
+
+TEST $CLI volume create ${V0}_2 replica 3 arbiter 1 $H0:$B0/v{3..5}
+
+TEST $CLI volume create ${V0}_3 disperse 3 redundancy 1 $H0:$B0/v{6..8}
+TEST $CLI volume start ${V0}_3
+EXPECT 'Started' volinfo_field ${V0}_3 'Status'
+
+TEST $CLI volume create ${V0}_4 replica 3 $H0:$B0/v{9..14}
+TEST $CLI volume start ${V0}_4
+EXPECT 'Started' volinfo_field ${V0}_4 'Status'
+
+#log rotate option
+#provided volume does not exist
+TEST ! $CLI volume log ${V0}_5 rotate
+
+#volume must be started before using log rotate option
+TEST ! $CLI volume log ${V0}_1 rotate
+TEST $CLI volume start ${V0}_1
+EXPECT 'Started' volinfo_field ${V0}_1 'Status'
+
+#incorrect brick provided for the volume
+TEST ! $CLI volume log ${V0}_1 rotate $H0:$B0/v15
+
+#add-brick operations
+#volume must be in started to state to increase replica count
+TEST ! $CLI volume add-brick ${V0}_2 replica 4 $H0:$B0/v15
+TEST $CLI volume start ${V0}_2
+EXPECT 'Started' volinfo_field ${V0}_2 'Status'
+
+#incorrect number of bricks for a replica 4 volume
+TEST ! $CLI volume add-brick ${V0}_1 replica 4 $H0:$B0/v15
+
+#replica count provided is less than the current replica count
+TEST ! $CLI volume add-brick ${V0}_2 replica 2 $H0:$B0/v15
+
+#dispersed to replicated dispersed not possible
+TEST ! $CLI volume add-brick ${V0}_3 replica 2 $H0:$B0/v15
+
+#remove-brick operations
+#replica count option provided for dispersed vol
+TEST ! $CLI volume remove-brick ${V0}_3 replica 2 $H0:$B0/v8 start
+
+#given replica count is greater than the current replica count
+TEST ! $CLI volume remove-brick ${V0}_2 replica 4 $H0:$B0/v5 start
+
+#number of bricks to be removed, must be a multiple of replica count
+TEST ! $CLI volume remove-brick ${V0}_2 replica 3 $H0:$B0/v{3..4} start
+
+#less number of bricks given to reduce the replica count
+TEST ! $CLI volume remove-brick ${V0}_2 replica 1 $H0:$B0/v3 start
+
+#bricks should be from different subvol
+TEST ! $CLI volume remove-brick ${V0}_4 replica 2 $H0:$B0/v{13..14} start
+
+#arbiter must be removed to reduce replica count
+TEST ! $CLI volume remove-brick ${V0}_2 replica 1 $H0:$B0/v{3..4} start
+
+#removal of bricks is not allowed without reducing the replica count explicitly
+TEST ! $CLI volume remove-brick ${V0}_2 replica 3 $H0:$B0/v{3..5} start
+
+#incorrect brick for given vol
+TEST ! $CLI volume remove-brick ${V0}_1 $H0:$B0/v15 start
+
+#removing all the bricks are not allowed
+TEST ! $CLI volume remove-brick ${V0}_1 $H0:$B0/v{1..2} start
+
+#volume must not be stopped state while removing bricks
+TEST $CLI volume stop ${V0}_1
+TEST ! $CLI volume remove-brick ${V0}_1 $H0:$B0/v1 start
+
+cleanup \ No newline at end of file
diff --git a/tests/volume.rc b/tests/volume.rc
index bc768c9434f..b38848c0e52 100644
--- a/tests/volume.rc
+++ b/tests/volume.rc
@@ -992,3 +992,11 @@ function get-xml()
{
$CLI $1 --xml | xmllint --format - | grep $2 | sed 's/\(<"$2">\|<\/"$2">\)//g'
}
+
+function logging_time_check()
+{
+ local logdir=$1
+ local logfile=`echo ${0##*/}`_glusterd1.log
+
+ cat $logdir/1/$logfile | tail -n 2 | head -n 1 | grep $(date +%H:%M) | wc -l
+}
diff --git a/tools/gfind_missing_files/gfind_missing_files.sh b/tools/gfind_missing_files/gfind_missing_files.sh
index f42fe7b05af..e7aaa0b5dd4 100644
--- a/tools/gfind_missing_files/gfind_missing_files.sh
+++ b/tools/gfind_missing_files/gfind_missing_files.sh
@@ -61,7 +61,7 @@ mount_slave()
parse_cli()
{
- if [[ $# -ne 4 ]]; then
+ if [ "$#" -ne 4 ]; then
echo "Usage: gfind_missing_files <brick-path> <slave-host> <slave-vol> <OUTFILE>"
exit 1
else
diff --git a/tools/glusterfind/src/utils.py b/tools/glusterfind/src/utils.py
index cc099031ef8..906ebd8f252 100644
--- a/tools/glusterfind/src/utils.py
+++ b/tools/glusterfind/src/utils.py
@@ -230,7 +230,11 @@ def get_changelog_rollover_time(volumename):
try:
tree = etree.fromstring(out)
- return int(tree.find('volGetopts/Opt/Value').text)
+ val = tree.find('volGetopts/Opt/Value').text
+ if val is not None:
+ # Filter the value by split, as it may be 'X (DEFAULT)'
+ # and we only need 'X'
+ return int(val.split(' ', 1)[0])
except ParseError:
return DEFAULT_CHANGELOG_INTERVAL
diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c
index f308088ed5f..032ab5c8001 100644
--- a/xlators/cluster/afr/src/afr-common.c
+++ b/xlators/cluster/afr/src/afr-common.c
@@ -45,6 +45,41 @@ afr_quorum_errno(afr_private_t *priv)
return ENOTCONN;
}
+gf_boolean_t
+afr_is_private_directory(afr_private_t *priv, uuid_t pargfid, const char *name,
+ pid_t pid)
+{
+ if (!__is_root_gfid(pargfid)) {
+ return _gf_false;
+ }
+
+ if (strcmp(name, GF_REPLICATE_TRASH_DIR) == 0) {
+ /*For backward compatibility /.landfill is private*/
+ return _gf_true;
+ }
+
+ if (pid == GF_CLIENT_PID_GSYNCD) {
+ /*geo-rep needs to create/sync private directory on slave because
+ * it appears in changelog*/
+ return _gf_false;
+ }
+
+ if (pid == GF_CLIENT_PID_GLFS_HEAL || pid == GF_CLIENT_PID_SELF_HEALD) {
+ if (strcmp(name, priv->anon_inode_name) == 0) {
+ /* anonymous-inode dir is private*/
+ return _gf_true;
+ }
+ } else {
+ if (strncmp(name, AFR_ANON_DIR_PREFIX, strlen(AFR_ANON_DIR_PREFIX)) ==
+ 0) {
+ /* anonymous-inode dir prefix is private for geo-rep to work*/
+ return _gf_true;
+ }
+ }
+
+ return _gf_false;
+}
+
void
afr_fill_success_replies(afr_local_t *local, afr_private_t *priv,
unsigned char *replies)
@@ -1197,12 +1232,11 @@ afr_inode_get_readable(call_frame_t *frame, inode_t *inode, xlator_t *this,
return 0;
}
-int
+static int
afr_inode_split_brain_choice_get(inode_t *inode, xlator_t *this,
int *spb_choice)
{
int ret = -1;
-
GF_VALIDATE_OR_GOTO(this->name, inode, out);
LOCK(&inode->lock);
@@ -1214,6 +1248,40 @@ out:
return ret;
}
+/*
+ * frame is used to get the favourite policy. Since
+ * afr_inode_split_brain_choice_get was called with afr_open, it is possible to
+ * have a frame with out local->replies. So in that case, frame is passed as
+ * null, hence this function will handle the frame NULL case.
+ */
+int
+afr_split_brain_read_subvol_get(inode_t *inode, xlator_t *this,
+ call_frame_t *frame, int *spb_subvol)
+{
+ int ret = -1;
+ afr_local_t *local = NULL;
+ afr_private_t *priv = NULL;
+
+ GF_VALIDATE_OR_GOTO("afr", this, out);
+ GF_VALIDATE_OR_GOTO(this->name, this->private, out);
+ GF_VALIDATE_OR_GOTO(this->name, inode, out);
+ GF_VALIDATE_OR_GOTO(this->name, spb_subvol, out);
+
+ priv = this->private;
+
+ ret = afr_inode_split_brain_choice_get(inode, this, spb_subvol);
+ if (*spb_subvol < 0 && priv->fav_child_policy && frame && frame->local) {
+ local = frame->local;
+ *spb_subvol = afr_sh_get_fav_by_policy(this, local->replies, inode,
+ NULL);
+ if (*spb_subvol >= 0) {
+ ret = 0;
+ }
+ }
+
+out:
+ return ret;
+}
int
afr_inode_read_subvol_set(inode_t *inode, xlator_t *this, unsigned char *data,
unsigned char *metadata, int event)
@@ -2238,8 +2306,9 @@ afr_hash_child(afr_read_subvol_args_t *args, afr_private_t *priv,
* need is a low probability that multiple clients
* won't converge on the same subvolume.
*/
+ gf_uuid_copy(gfid_copy, args->gfid);
pid = getpid();
- memcpy(gfid_copy, &pid, sizeof(pid));
+ *(pid_t *)gfid_copy ^= pid;
}
child = SuperFastHash((char *)gfid_copy, sizeof(gfid_copy)) %
priv->child_count;
@@ -2823,7 +2892,7 @@ afr_attempt_readsubvol_set(call_frame_t *frame, xlator_t *this,
{
afr_private_t *priv = NULL;
afr_local_t *local = NULL;
- int spb_choice = -1;
+ int spb_subvol = -1;
int child_count = -1;
if (*read_subvol != -1)
@@ -2833,10 +2902,10 @@ afr_attempt_readsubvol_set(call_frame_t *frame, xlator_t *this,
local = frame->local;
child_count = priv->child_count;
- afr_inode_split_brain_choice_get(local->inode, this, &spb_choice);
- if ((spb_choice >= 0) &&
+ afr_split_brain_read_subvol_get(local->inode, this, frame, &spb_subvol);
+ if ((spb_subvol >= 0) &&
(AFR_COUNT(success_replies, child_count) == child_count)) {
- *read_subvol = spb_choice;
+ *read_subvol = spb_subvol;
} else if (!priv->quorum_count ||
frame->root->pid == GF_CLIENT_PID_GLFS_HEAL) {
*read_subvol = afr_first_up_child(frame, this);
@@ -3945,11 +4014,10 @@ afr_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xattr_req)
return 0;
}
- if (__is_root_gfid(loc->parent->gfid)) {
- if (!strcmp(loc->name, GF_REPLICATE_TRASH_DIR)) {
- op_errno = EPERM;
- goto out;
- }
+ if (afr_is_private_directory(this->private, loc->parent->gfid, loc->name,
+ frame->root->pid)) {
+ op_errno = EPERM;
+ goto out;
}
local = AFR_FRAME_INIT(frame, op_errno);
@@ -5627,6 +5695,7 @@ afr_priv_dump(xlator_t *this)
priv->background_self_heal_count);
gf_proc_dump_write("healers", "%d", priv->healers);
gf_proc_dump_write("read-hash-mode", "%d", priv->hash_mode);
+ gf_proc_dump_write("use-anonymous-inode", "%d", priv->use_anon_inode);
if (priv->quorum_count == AFR_QUORUM_AUTO) {
gf_proc_dump_write("quorum-type", "auto");
} else if (priv->quorum_count == 0) {
@@ -6603,6 +6672,8 @@ afr_priv_destroy(afr_private_t *priv)
if (!priv)
goto out;
+
+ GF_FREE(priv->sh_domain);
GF_FREE(priv->last_event);
child_count = priv->child_count;
@@ -6618,6 +6689,7 @@ afr_priv_destroy(afr_private_t *priv)
GF_FREE(priv->local);
GF_FREE(priv->pending_key);
GF_FREE(priv->children);
+ GF_FREE(priv->anon_inode);
GF_FREE(priv->child_up);
GF_FREE(priv->halo_child_up);
GF_FREE(priv->child_latency);
diff --git a/xlators/cluster/afr/src/afr-dir-read.c b/xlators/cluster/afr/src/afr-dir-read.c
index 74f71fdc76a..f8bf8340dab 100644
--- a/xlators/cluster/afr/src/afr-dir-read.c
+++ b/xlators/cluster/afr/src/afr-dir-read.c
@@ -67,7 +67,8 @@ afr_opendir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
}
int
-afr_opendir(call_frame_t *frame, xlator_t *this, loc_t *loc, fd_t *fd)
+afr_opendir(call_frame_t *frame, xlator_t *this, loc_t *loc, fd_t *fd,
+ dict_t *xdata)
{
afr_private_t *priv = NULL;
afr_local_t *local = NULL;
@@ -163,8 +164,8 @@ afr_validate_read_subvol(inode_t *inode, xlator_t *this, int par_read_subvol)
}
static void
-afr_readdir_transform_entries(gf_dirent_t *subvol_entries, int subvol,
- gf_dirent_t *entries, fd_t *fd)
+afr_readdir_transform_entries(call_frame_t *frame, gf_dirent_t *subvol_entries,
+ int subvol, gf_dirent_t *entries, fd_t *fd)
{
int ret = -1;
gf_dirent_t *entry = NULL;
@@ -182,8 +183,8 @@ afr_readdir_transform_entries(gf_dirent_t *subvol_entries, int subvol,
list_for_each_entry_safe(entry, tmp, &subvol_entries->list, list)
{
- if (__is_root_gfid(fd->inode->gfid) &&
- !strcmp(entry->d_name, GF_REPLICATE_TRASH_DIR)) {
+ if (afr_is_private_directory(priv, fd->inode->gfid, entry->d_name,
+ frame->root->pid)) {
continue;
}
@@ -227,8 +228,8 @@ afr_readdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
}
if (op_ret >= 0)
- afr_readdir_transform_entries(subvol_entries, (long)cookie, &entries,
- local->fd);
+ afr_readdir_transform_entries(frame, subvol_entries, (long)cookie,
+ &entries, local->fd);
AFR_STACK_UNWIND(readdir, frame, op_ret, op_errno, &entries, xdata);
diff --git a/xlators/cluster/afr/src/afr-open.c b/xlators/cluster/afr/src/afr-open.c
index a5b004f4258..64856042b65 100644
--- a/xlators/cluster/afr/src/afr-open.c
+++ b/xlators/cluster/afr/src/afr-open.c
@@ -137,7 +137,7 @@ afr_open(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
{
afr_private_t *priv = NULL;
afr_local_t *local = NULL;
- int spb_choice = 0;
+ int spb_subvol = 0;
int event_generation = 0;
int ret = 0;
int32_t op_errno = 0;
@@ -179,9 +179,9 @@ afr_open(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
ret = afr_inode_get_readable(frame, local->inode, this, NULL,
&event_generation, AFR_DATA_TRANSACTION);
if ((ret < 0) &&
- (afr_inode_split_brain_choice_get(local->inode, this, &spb_choice) ==
- 0) &&
- spb_choice < 0) {
+ (afr_split_brain_read_subvol_get(local->inode, this, NULL,
+ &spb_subvol) == 0) &&
+ spb_subvol < 0) {
afr_inode_refresh(frame, this, local->inode, local->inode->gfid,
afr_open_continue);
} else {
diff --git a/xlators/cluster/afr/src/afr-read-txn.c b/xlators/cluster/afr/src/afr-read-txn.c
index 88ea9b4cfc6..6fc2c75145c 100644
--- a/xlators/cluster/afr/src/afr-read-txn.c
+++ b/xlators/cluster/afr/src/afr-read-txn.c
@@ -272,7 +272,7 @@ afr_read_txn_refresh_done(call_frame_t *frame, xlator_t *this, int err)
int read_subvol = -1;
inode_t *inode = NULL;
int ret = -1;
- int spb_choice = -1;
+ int spb_subvol = -1;
local = frame->local;
inode = local->inode;
@@ -303,9 +303,9 @@ afr_read_txn_refresh_done(call_frame_t *frame, xlator_t *this, int err)
local->read_attempted[read_subvol] = 1;
readfn:
if (read_subvol == -1) {
- ret = afr_inode_split_brain_choice_get(inode, this, &spb_choice);
- if ((ret == 0) && spb_choice >= 0)
- read_subvol = spb_choice;
+ ret = afr_split_brain_read_subvol_get(inode, this, frame, &spb_subvol);
+ if ((ret == 0) && spb_subvol >= 0)
+ read_subvol = spb_subvol;
}
if (read_subvol == -1) {
diff --git a/xlators/cluster/afr/src/afr-self-heal-common.c b/xlators/cluster/afr/src/afr-self-heal-common.c
index fdec66340ba..a580a1584cc 100644
--- a/xlators/cluster/afr/src/afr-self-heal-common.c
+++ b/xlators/cluster/afr/src/afr-self-heal-common.c
@@ -140,7 +140,7 @@ heal:
}
}
out:
- if (gfid_idx && (*gfid_idx == -1) && (ret == 0)) {
+ if (gfid_idx && (*gfid_idx == -1) && (ret == 0) && local) {
ret = -afr_final_errno(local, priv);
}
loc_wipe(&loc);
@@ -1909,7 +1909,8 @@ afr_selfheal_unlocked_discover(call_frame_t *frame, inode_t *inode, uuid_t gfid,
dict_t *dict = NULL;
local = frame->local;
- if (local && local->xattr_req)
+
+ if (local->xattr_req)
dict = local->xattr_req;
return afr_selfheal_unlocked_discover_on(frame, inode, gfid, replies,
@@ -2749,3 +2750,185 @@ afr_choose_source_by_policy(afr_private_t *priv, unsigned char *sources,
out:
return source;
}
+
+static int
+afr_anon_inode_mkdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+ int32_t op_ret, int32_t op_errno, inode_t *inode,
+ struct iatt *buf, struct iatt *preparent,
+ struct iatt *postparent, dict_t *xdata)
+{
+ afr_local_t *local = frame->local;
+ int i = (long)cookie;
+
+ local->replies[i].valid = 1;
+ local->replies[i].op_ret = op_ret;
+ local->replies[i].op_errno = op_errno;
+ if (op_ret == 0) {
+ local->op_ret = 0;
+ local->replies[i].poststat = *buf;
+ local->replies[i].preparent = *preparent;
+ local->replies[i].postparent = *postparent;
+ }
+ if (xdata) {
+ local->replies[i].xdata = dict_ref(xdata);
+ }
+
+ syncbarrier_wake(&local->barrier);
+ return 0;
+}
+
+int
+afr_anon_inode_create(xlator_t *this, int child, inode_t **linked_inode)
+{
+ call_frame_t *frame = NULL;
+ afr_local_t *local = NULL;
+ afr_private_t *priv = this->private;
+ unsigned char *mkdir_on = alloca0(priv->child_count);
+ unsigned char *lookup_on = alloca0(priv->child_count);
+ loc_t loc = {0};
+ int32_t op_errno = 0;
+ int32_t child_op_errno = 0;
+ struct iatt iatt = {0};
+ dict_t *xdata = NULL;
+ uuid_t anon_inode_gfid = {0};
+ int mkdir_count = 0;
+ int i = 0;
+
+ /*Try to mkdir everywhere and return success if the dir exists on 'child'
+ */
+
+ if (!priv->use_anon_inode) {
+ op_errno = EINVAL;
+ goto out;
+ }
+
+ frame = afr_frame_create(this, &op_errno);
+ if (op_errno) {
+ goto out;
+ }
+ local = frame->local;
+ if (!local->child_up[child]) {
+ /*Other bricks may need mkdir so don't error out yet*/
+ child_op_errno = ENOTCONN;
+ }
+ gf_uuid_parse(priv->anon_gfid_str, anon_inode_gfid);
+ for (i = 0; i < priv->child_count; i++) {
+ if (!local->child_up[i])
+ continue;
+
+ if (priv->anon_inode[i]) {
+ mkdir_on[i] = 0;
+ } else {
+ mkdir_on[i] = 1;
+ mkdir_count++;
+ }
+ }
+
+ if (mkdir_count == 0) {
+ *linked_inode = inode_find(this->itable, anon_inode_gfid);
+ if (*linked_inode) {
+ op_errno = 0;
+ goto out;
+ }
+ }
+
+ loc.parent = inode_ref(this->itable->root);
+ loc.name = priv->anon_inode_name;
+ loc.inode = inode_new(this->itable);
+ if (!loc.inode) {
+ op_errno = ENOMEM;
+ goto out;
+ }
+
+ xdata = dict_new();
+ if (!xdata) {
+ op_errno = ENOMEM;
+ goto out;
+ }
+
+ op_errno = -dict_set_gfuuid(xdata, "gfid-req", anon_inode_gfid, _gf_true);
+ if (op_errno) {
+ goto out;
+ }
+
+ if (mkdir_count == 0) {
+ memcpy(lookup_on, local->child_up, priv->child_count);
+ goto lookup;
+ }
+
+ AFR_ONLIST(mkdir_on, frame, afr_anon_inode_mkdir_cbk, mkdir, &loc, 0755, 0,
+ xdata);
+
+ for (i = 0; i < priv->child_count; i++) {
+ if (!mkdir_on[i]) {
+ continue;
+ }
+
+ if (local->replies[i].op_ret == 0) {
+ priv->anon_inode[i] = 1;
+ iatt = local->replies[i].poststat;
+ } else if (local->replies[i].op_ret < 0 &&
+ local->replies[i].op_errno == EEXIST) {
+ lookup_on[i] = 1;
+ } else if (i == child) {
+ child_op_errno = local->replies[i].op_errno;
+ }
+ }
+
+ if (AFR_COUNT(lookup_on, priv->child_count) == 0) {
+ goto link;
+ }
+
+lookup:
+ AFR_ONLIST(lookup_on, frame, afr_selfheal_discover_cbk, lookup, &loc,
+ xdata);
+ for (i = 0; i < priv->child_count; i++) {
+ if (!lookup_on[i]) {
+ continue;
+ }
+
+ if (local->replies[i].op_ret == 0) {
+ if (gf_uuid_compare(anon_inode_gfid,
+ local->replies[i].poststat.ia_gfid) == 0) {
+ priv->anon_inode[i] = 1;
+ iatt = local->replies[i].poststat;
+ } else {
+ if (i == child)
+ child_op_errno = EINVAL;
+ gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_INVALID_DATA,
+ "%s has gfid: %s", priv->anon_inode_name,
+ uuid_utoa(local->replies[i].poststat.ia_gfid));
+ }
+ } else if (i == child) {
+ child_op_errno = local->replies[i].op_errno;
+ }
+ }
+link:
+ if (!gf_uuid_is_null(iatt.ia_gfid)) {
+ *linked_inode = inode_link(loc.inode, loc.parent, loc.name, &iatt);
+ if (*linked_inode) {
+ op_errno = 0;
+ inode_lookup(*linked_inode);
+ } else {
+ op_errno = ENOMEM;
+ }
+ goto out;
+ }
+
+out:
+ if (xdata)
+ dict_unref(xdata);
+ loc_wipe(&loc);
+ /*child_op_errno takes precedence*/
+ if (child_op_errno == 0) {
+ child_op_errno = op_errno;
+ }
+
+ if (child_op_errno && *linked_inode) {
+ inode_unref(*linked_inode);
+ *linked_inode = NULL;
+ }
+ if (frame)
+ AFR_STACK_DESTROY(frame);
+ return -child_op_errno;
+}
diff --git a/xlators/cluster/afr/src/afr-self-heal-entry.c b/xlators/cluster/afr/src/afr-self-heal-entry.c
index ac31751997f..64893f441e3 100644
--- a/xlators/cluster/afr/src/afr-self-heal-entry.c
+++ b/xlators/cluster/afr/src/afr-self-heal-entry.c
@@ -16,54 +16,170 @@
#include <glusterfs/syncop-utils.h>
#include <glusterfs/events.h>
-static int
-afr_selfheal_entry_delete(xlator_t *this, inode_t *dir, const char *name,
- inode_t *inode, int child, struct afr_reply *replies)
+int
+afr_selfheal_entry_anon_inode(xlator_t *this, inode_t *dir, const char *name,
+ inode_t *inode, int child,
+ struct afr_reply *replies,
+ gf_boolean_t *anon_inode)
{
afr_private_t *priv = NULL;
+ afr_local_t *local = NULL;
xlator_t *subvol = NULL;
int ret = 0;
+ int i = 0;
+ char g[64] = {0};
+ unsigned char *lookup_success = NULL;
+ call_frame_t *frame = NULL;
+ loc_t loc2 = {
+ 0,
+ };
loc_t loc = {
0,
};
- char g[64];
priv = this->private;
-
subvol = priv->children[child];
+ lookup_success = alloca0(priv->child_count);
+ uuid_utoa_r(replies[child].poststat.ia_gfid, g);
+ loc.inode = inode_new(inode->table);
+ if (!loc.inode) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ if (replies[child].poststat.ia_type == IA_IFDIR) {
+ /* This directory may have sub-directory hierarchy which may need to
+ * be preserved for subsequent heals. So unconditionally move the
+ * directory to anonymous-inode directory*/
+ *anon_inode = _gf_true;
+ goto anon_inode;
+ }
+
+ frame = afr_frame_create(this, &ret);
+ if (!frame) {
+ ret = -ret;
+ goto out;
+ }
+ local = frame->local;
+ gf_uuid_copy(loc.gfid, replies[child].poststat.ia_gfid);
+ AFR_ONLIST(local->child_up, frame, afr_selfheal_discover_cbk, lookup, &loc,
+ NULL);
+ for (i = 0; i < priv->child_count; i++) {
+ if (local->replies[i].op_ret == 0) {
+ lookup_success[i] = 1;
+ } else if (local->replies[i].op_errno != ENOENT &&
+ local->replies[i].op_errno != ESTALE) {
+ ret = -local->replies[i].op_errno;
+ }
+ }
+
+ if (priv->quorum_count) {
+ if (afr_has_quorum(lookup_success, this, NULL)) {
+ *anon_inode = _gf_true;
+ }
+ } else if (AFR_COUNT(lookup_success, priv->child_count) > 1) {
+ *anon_inode = _gf_true;
+ } else if (ret) {
+ goto out;
+ }
+
+anon_inode:
+ if (!*anon_inode) {
+ ret = 0;
+ goto out;
+ }
loc.parent = inode_ref(dir);
gf_uuid_copy(loc.pargfid, dir->gfid);
loc.name = name;
- loc.inode = inode_ref(inode);
- if (replies[child].valid && replies[child].op_ret == 0) {
- switch (replies[child].poststat.ia_type) {
- case IA_IFDIR:
- gf_msg(this->name, GF_LOG_WARNING, 0,
- AFR_MSG_EXPUNGING_FILE_OR_DIR,
- "expunging dir %s/%s (%s) on %s", uuid_utoa(dir->gfid),
- name, uuid_utoa_r(replies[child].poststat.ia_gfid, g),
- subvol->name);
- ret = syncop_rmdir(subvol, &loc, 1, NULL, NULL);
- break;
- default:
- gf_msg(this->name, GF_LOG_WARNING, 0,
- AFR_MSG_EXPUNGING_FILE_OR_DIR,
- "expunging file %s/%s (%s) on %s", uuid_utoa(dir->gfid),
- name, uuid_utoa_r(replies[child].poststat.ia_gfid, g),
- subvol->name);
- ret = syncop_unlink(subvol, &loc, NULL, NULL);
- break;
- }
+ ret = afr_anon_inode_create(this, child, &loc2.parent);
+ if (ret < 0)
+ goto out;
+
+ loc2.name = g;
+ ret = syncop_rename(subvol, &loc, &loc2, NULL, NULL);
+ if (ret < 0) {
+ gf_msg(this->name, GF_LOG_WARNING, -ret, AFR_MSG_EXPUNGING_FILE_OR_DIR,
+ "Rename to %s dir %s/%s (%s) on %s failed",
+ priv->anon_inode_name, uuid_utoa(dir->gfid), name, g,
+ subvol->name);
+ } else {
+ gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_EXPUNGING_FILE_OR_DIR,
+ "Rename to %s dir %s/%s (%s) on %s successful",
+ priv->anon_inode_name, uuid_utoa(dir->gfid), name, g,
+ subvol->name);
}
+out:
loc_wipe(&loc);
+ loc_wipe(&loc2);
+ if (frame) {
+ AFR_STACK_DESTROY(frame);
+ }
return ret;
}
int
+afr_selfheal_entry_delete(xlator_t *this, inode_t *dir, const char *name,
+ inode_t *inode, int child, struct afr_reply *replies)
+{
+ char g[64] = {0};
+ afr_private_t *priv = NULL;
+ xlator_t *subvol = NULL;
+ int ret = 0;
+ loc_t loc = {
+ 0,
+ };
+ gf_boolean_t anon_inode = _gf_false;
+
+ priv = this->private;
+ subvol = priv->children[child];
+
+ if ((!replies[child].valid) || (replies[child].op_ret < 0)) {
+ /*Nothing to do*/
+ ret = 0;
+ goto out;
+ }
+
+ if (priv->use_anon_inode) {
+ ret = afr_selfheal_entry_anon_inode(this, dir, name, inode, child,
+ replies, &anon_inode);
+ if (ret < 0 || anon_inode)
+ goto out;
+ }
+
+ loc.parent = inode_ref(dir);
+ loc.inode = inode_new(inode->table);
+ if (!loc.inode) {
+ ret = -ENOMEM;
+ goto out;
+ }
+ loc.name = name;
+ switch (replies[child].poststat.ia_type) {
+ case IA_IFDIR:
+ gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_EXPUNGING_FILE_OR_DIR,
+ "expunging dir %s/%s (%s) on %s", uuid_utoa(dir->gfid), name,
+ uuid_utoa_r(replies[child].poststat.ia_gfid, g),
+ subvol->name);
+ ret = syncop_rmdir(subvol, &loc, 1, NULL, NULL);
+ break;
+ default:
+ gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_EXPUNGING_FILE_OR_DIR,
+ "expunging file %s/%s (%s) on %s", uuid_utoa(dir->gfid),
+ name, uuid_utoa_r(replies[child].poststat.ia_gfid, g),
+ subvol->name);
+ ret = syncop_unlink(subvol, &loc, NULL, NULL);
+ break;
+ }
+
+out:
+ loc_wipe(&loc);
+ return ret;
+}
+
+int
afr_selfheal_recreate_entry(call_frame_t *frame, int dst, int source,
unsigned char *sources, inode_t *dir,
const char *name, inode_t *inode,
@@ -76,6 +192,9 @@ afr_selfheal_recreate_entry(call_frame_t *frame, int dst, int source,
loc_t srcloc = {
0,
};
+ loc_t anonloc = {
+ 0,
+ };
xlator_t *this = frame->this;
afr_private_t *priv = NULL;
dict_t *xdata = NULL;
@@ -86,15 +205,17 @@ afr_selfheal_recreate_entry(call_frame_t *frame, int dst, int source,
0,
};
unsigned char *newentry = NULL;
- char dir_uuid_str[64] = {0}, iatt_uuid_str[64] = {0};
+ char iatt_uuid_str[64] = {0};
+ char dir_uuid_str[64] = {0};
priv = this->private;
iatt = &replies[source].poststat;
+ uuid_utoa_r(iatt->ia_gfid, iatt_uuid_str);
if (iatt->ia_type == IA_INVAL || gf_uuid_is_null(iatt->ia_gfid)) {
gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SELF_HEAL_FAILED,
"Invalid ia_type (%d) or gfid(%s). source brick=%d, "
"pargfid=%s, name=%s",
- iatt->ia_type, uuid_utoa_r(iatt->ia_gfid, iatt_uuid_str), source,
+ iatt->ia_type, iatt_uuid_str, source,
uuid_utoa_r(dir->gfid, dir_uuid_str), name);
ret = -EINVAL;
goto out;
@@ -120,14 +241,24 @@ afr_selfheal_recreate_entry(call_frame_t *frame, int dst, int source,
srcloc.inode = inode_ref(inode);
gf_uuid_copy(srcloc.gfid, iatt->ia_gfid);
- if (iatt->ia_type != IA_IFDIR)
- ret = syncop_lookup(priv->children[dst], &srcloc, 0, 0, 0, 0);
- if (iatt->ia_type == IA_IFDIR || ret == -ENOENT || ret == -ESTALE) {
+ ret = syncop_lookup(priv->children[dst], &srcloc, 0, 0, 0, 0);
+ if (ret == -ENOENT || ret == -ESTALE) {
newentry[dst] = 1;
ret = afr_selfheal_newentry_mark(frame, this, inode, source, replies,
sources, newentry);
if (ret)
goto out;
+ } else if (ret == 0 && iatt->ia_type == IA_IFDIR && priv->use_anon_inode) {
+ // Try rename from hidden directory
+ ret = afr_anon_inode_create(this, dst, &anonloc.parent);
+ if (ret < 0)
+ goto out;
+ anonloc.inode = inode_ref(inode);
+ anonloc.name = iatt_uuid_str;
+ ret = syncop_rename(priv->children[dst], &anonloc, &loc, NULL, NULL);
+ if (ret == -ENOENT || ret == -ESTALE)
+ ret = -1; /*This sets 'mismatch' to true*/
+ goto out;
}
mode = st_mode_from_ia(iatt->ia_prot, iatt->ia_type);
@@ -166,6 +297,7 @@ out:
GF_FREE(linkname);
loc_wipe(&loc);
loc_wipe(&srcloc);
+ loc_wipe(&anonloc);
return ret;
}
@@ -578,6 +710,11 @@ afr_selfheal_entry_dirent(call_frame_t *frame, xlator_t *this, fd_t *fd,
priv = this->private;
+ if (afr_is_private_directory(priv, fd->inode->gfid, name,
+ GF_CLIENT_PID_SELF_HEALD)) {
+ return 0;
+ }
+
xattr = dict_new();
if (!xattr)
return -ENOMEM;
@@ -626,7 +763,7 @@ afr_selfheal_entry_dirent(call_frame_t *frame, xlator_t *this, fd_t *fd,
replies);
if ((ret == 0) && (priv->esh_granular) && parent_idx_inode) {
- ret = afr_shd_index_purge(subvol, parent_idx_inode, name,
+ ret = afr_shd_entry_purge(subvol, parent_idx_inode, name,
inode->ia_type);
/* Why is ret force-set to 0? We do not care about
* index purge failing for full heal as it is quite
@@ -756,10 +893,6 @@ afr_selfheal_entry_do_subvol(call_frame_t *frame, xlator_t *this, fd_t *fd,
if (!strcmp(entry->d_name, ".") || !strcmp(entry->d_name, ".."))
continue;
- if (__is_root_gfid(fd->inode->gfid) &&
- !strcmp(entry->d_name, GF_REPLICATE_TRASH_DIR))
- continue;
-
ret = afr_selfheal_entry_dirent(iter_frame, this, fd, entry->d_name,
loc.inode, subvol,
local->need_full_crawl);
@@ -822,7 +955,7 @@ afr_selfheal_entry_granular_dirent(xlator_t *subvol, gf_dirent_t *entry,
/* The name indices under the pgfid index dir are guaranteed
* to be regular files. Hence the hardcoding.
*/
- afr_shd_index_purge(subvol, parent->inode, entry->d_name, IA_IFREG);
+ afr_shd_entry_purge(subvol, parent->inode, entry->d_name, IA_IFREG);
ret = 0;
goto out;
}
diff --git a/xlators/cluster/afr/src/afr-self-heal-name.c b/xlators/cluster/afr/src/afr-self-heal-name.c
index dace07131cb..834aac86d48 100644
--- a/xlators/cluster/afr/src/afr-self-heal-name.c
+++ b/xlators/cluster/afr/src/afr-self-heal-name.c
@@ -98,21 +98,12 @@ __afr_selfheal_name_expunge(xlator_t *this, inode_t *parent, uuid_t pargfid,
const char *bname, inode_t *inode,
struct afr_reply *replies)
{
- loc_t loc = {
- 0,
- };
int i = 0;
afr_private_t *priv = NULL;
- char g[64];
int ret = 0;
priv = this->private;
- loc.parent = inode_ref(parent);
- gf_uuid_copy(loc.pargfid, pargfid);
- loc.name = bname;
- loc.inode = inode_ref(inode);
-
for (i = 0; i < priv->child_count; i++) {
if (!replies[i].valid)
continue;
@@ -120,30 +111,10 @@ __afr_selfheal_name_expunge(xlator_t *this, inode_t *parent, uuid_t pargfid,
if (replies[i].op_ret)
continue;
- switch (replies[i].poststat.ia_type) {
- case IA_IFDIR:
- gf_msg(this->name, GF_LOG_WARNING, 0,
- AFR_MSG_EXPUNGING_FILE_OR_DIR,
- "expunging dir %s/%s (%s) on %s", uuid_utoa(pargfid),
- bname, uuid_utoa_r(replies[i].poststat.ia_gfid, g),
- priv->children[i]->name);
-
- ret |= syncop_rmdir(priv->children[i], &loc, 1, NULL, NULL);
- break;
- default:
- gf_msg(this->name, GF_LOG_WARNING, 0,
- AFR_MSG_EXPUNGING_FILE_OR_DIR,
- "expunging file %s/%s (%s) on %s", uuid_utoa(pargfid),
- bname, uuid_utoa_r(replies[i].poststat.ia_gfid, g),
- priv->children[i]->name);
-
- ret |= syncop_unlink(priv->children[i], &loc, NULL, NULL);
- break;
- }
+ ret |= afr_selfheal_entry_delete(this, parent, bname, inode, i,
+ replies);
}
- loc_wipe(&loc);
-
return ret;
}
@@ -381,7 +352,7 @@ __afr_selfheal_name_do(call_frame_t *frame, xlator_t *this, inode_t *parent,
ret = __afr_selfheal_assign_gfid(this, parent, pargfid, bname, inode,
replies, gfid, locked_on, source, sources,
is_gfid_absent, &gfid_idx);
- if (ret)
+ if (ret || (gfid_idx < 0))
return ret;
ret = __afr_selfheal_name_impunge(frame, this, parent, pargfid, bname,
diff --git a/xlators/cluster/afr/src/afr-self-heal.h b/xlators/cluster/afr/src/afr-self-heal.h
index 7a038fa7fe3..48e6dbcfb18 100644
--- a/xlators/cluster/afr/src/afr-self-heal.h
+++ b/xlators/cluster/afr/src/afr-self-heal.h
@@ -369,4 +369,9 @@ gf_boolean_t
afr_is_file_empty_on_all_children(afr_private_t *priv,
struct afr_reply *replies);
+int
+afr_selfheal_entry_delete(xlator_t *this, inode_t *dir, const char *name,
+ inode_t *inode, int child, struct afr_reply *replies);
+int
+afr_anon_inode_create(xlator_t *this, int child, inode_t **linked_inode);
#endif /* !_AFR_SELFHEAL_H */
diff --git a/xlators/cluster/afr/src/afr-self-heald.c b/xlators/cluster/afr/src/afr-self-heald.c
index 2219a53b277..109fd4b7421 100644
--- a/xlators/cluster/afr/src/afr-self-heald.c
+++ b/xlators/cluster/afr/src/afr-self-heald.c
@@ -94,7 +94,7 @@ __afr_shd_healer_wait(struct subvol_healer *healer)
priv = healer->this->private;
disabled_loop:
- wait_till.tv_sec = time(NULL) + priv->shd.timeout;
+ wait_till.tv_sec = gf_time() + priv->shd.timeout;
while (!healer->rerun) {
ret = pthread_cond_timedwait(&healer->cond, &healer->mutex, &wait_till);
@@ -222,7 +222,7 @@ out:
}
int
-afr_shd_index_purge(xlator_t *subvol, inode_t *inode, char *name,
+afr_shd_entry_purge(xlator_t *subvol, inode_t *inode, char *name,
ia_type_t type)
{
int ret = 0;
@@ -371,7 +371,7 @@ afr_shd_sweep_prepare(struct subvol_healer *healer)
event->split_brain_count = 0;
event->heal_failed_count = 0;
- time(&event->start_time);
+ event->start_time = gf_time();
event->end_time = 0;
_mask_cancellation();
}
@@ -386,7 +386,7 @@ afr_shd_sweep_done(struct subvol_healer *healer)
event = &healer->crawl_event;
shd = &(((afr_private_t *)healer->this->private)->shd);
- time(&event->end_time);
+ event->end_time = gf_time();
history = gf_memdup(event, sizeof(*event));
event->start_time = 0;
@@ -424,7 +424,7 @@ afr_shd_index_heal(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
ret = afr_shd_selfheal(healer, healer->subvol, gfid);
if (ret == -ENOENT || ret == -ESTALE)
- afr_shd_index_purge(subvol, parent->inode, entry->d_name, val);
+ afr_shd_entry_purge(subvol, parent->inode, entry->d_name, val);
if (ret == 2)
/* If bricks crashed in pre-op after creating indices/xattrop
@@ -843,6 +843,176 @@ out:
return need_heal;
}
+static int
+afr_shd_anon_inode_cleaner(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
+ void *data)
+{
+ struct subvol_healer *healer = data;
+ afr_private_t *priv = healer->this->private;
+ call_frame_t *frame = NULL;
+ afr_local_t *local = NULL;
+ int ret = 0;
+ loc_t loc = {0};
+ int count = 0;
+ int i = 0;
+ int op_errno = 0;
+ struct iatt *iatt = NULL;
+ gf_boolean_t multiple_links = _gf_false;
+ unsigned char *gfid_present = alloca0(priv->child_count);
+ unsigned char *entry_present = alloca0(priv->child_count);
+ char *type = "file";
+
+ frame = afr_frame_create(healer->this, &ret);
+ if (!frame) {
+ ret = -ret;
+ goto out;
+ }
+ local = frame->local;
+ if (AFR_COUNT(local->child_up, priv->child_count) != priv->child_count) {
+ gf_msg_debug(healer->this->name, 0,
+ "Not all bricks are up. Skipping "
+ "cleanup of %s on %s",
+ entry->d_name, subvol->name);
+ ret = 0;
+ goto out;
+ }
+
+ loc.inode = inode_new(parent->inode->table);
+ if (!loc.inode) {
+ ret = -ENOMEM;
+ goto out;
+ }
+ ret = gf_uuid_parse(entry->d_name, loc.gfid);
+ if (ret) {
+ ret = 0;
+ goto out;
+ }
+ AFR_ONLIST(local->child_up, frame, afr_selfheal_discover_cbk, lookup, &loc,
+ NULL);
+ for (i = 0; i < priv->child_count; i++) {
+ if (local->replies[i].op_ret == 0) {
+ count++;
+ gfid_present[i] = 1;
+ iatt = &local->replies[i].poststat;
+ if (iatt->ia_type == IA_IFDIR) {
+ type = "dir";
+ }
+
+ if (i == healer->subvol) {
+ if (local->replies[i].poststat.ia_nlink > 1) {
+ multiple_links = _gf_true;
+ }
+ }
+ } else if (local->replies[i].op_errno != ENOENT &&
+ local->replies[i].op_errno != ESTALE) {
+ /*We don't have complete view. Skip the entry*/
+ gf_msg_debug(healer->this->name, local->replies[i].op_errno,
+ "Skipping cleanup of %s on %s", entry->d_name,
+ subvol->name);
+ ret = 0;
+ goto out;
+ }
+ }
+
+ /*Inode is deleted from subvol*/
+ if (count == 1 || (iatt->ia_type != IA_IFDIR && multiple_links)) {
+ gf_msg(healer->this->name, GF_LOG_WARNING, 0,
+ AFR_MSG_EXPUNGING_FILE_OR_DIR, "expunging %s %s/%s on %s", type,
+ priv->anon_inode_name, entry->d_name, subvol->name);
+ ret = afr_shd_entry_purge(subvol, parent->inode, entry->d_name,
+ iatt->ia_type);
+ if (ret == -ENOENT || ret == -ESTALE)
+ ret = 0;
+ } else if (count > 1) {
+ loc_wipe(&loc);
+ loc.parent = inode_ref(parent->inode);
+ loc.name = entry->d_name;
+ loc.inode = inode_new(parent->inode->table);
+ if (!loc.inode) {
+ ret = -ENOMEM;
+ goto out;
+ }
+ AFR_ONLIST(local->child_up, frame, afr_selfheal_discover_cbk, lookup,
+ &loc, NULL);
+ count = 0;
+ for (i = 0; i < priv->child_count; i++) {
+ if (local->replies[i].op_ret == 0) {
+ count++;
+ entry_present[i] = 1;
+ iatt = &local->replies[i].poststat;
+ } else if (local->replies[i].op_errno != ENOENT &&
+ local->replies[i].op_errno != ESTALE) {
+ /*We don't have complete view. Skip the entry*/
+ gf_msg_debug(healer->this->name, local->replies[i].op_errno,
+ "Skipping cleanup of %s on %s", entry->d_name,
+ subvol->name);
+ ret = 0;
+ goto out;
+ }
+ }
+ for (i = 0; i < priv->child_count; i++) {
+ if (gfid_present[i] && !entry_present[i]) {
+ /*Entry is not anonymous on at least one subvol*/
+ gf_msg_debug(healer->this->name, 0,
+ "Valid entry present on %s "
+ "Skipping cleanup of %s on %s",
+ priv->children[i]->name, entry->d_name,
+ subvol->name);
+ ret = 0;
+ goto out;
+ }
+ }
+
+ gf_msg(healer->this->name, GF_LOG_WARNING, 0,
+ AFR_MSG_EXPUNGING_FILE_OR_DIR,
+ "expunging %s %s/%s on all subvols", type, priv->anon_inode_name,
+ entry->d_name);
+ ret = 0;
+ for (i = 0; i < priv->child_count; i++) {
+ op_errno = -afr_shd_entry_purge(priv->children[i], loc.parent,
+ entry->d_name, iatt->ia_type);
+ if (op_errno != ENOENT && op_errno != ESTALE) {
+ ret |= -op_errno;
+ }
+ }
+ }
+
+out:
+ if (frame)
+ AFR_STACK_DESTROY(frame);
+ loc_wipe(&loc);
+ return ret;
+}
+
+static void
+afr_cleanup_anon_inode_dir(struct subvol_healer *healer)
+{
+ int ret = 0;
+ call_frame_t *frame = NULL;
+ afr_private_t *priv = healer->this->private;
+ loc_t loc = {0};
+
+ ret = afr_anon_inode_create(healer->this, healer->subvol, &loc.inode);
+ if (ret)
+ goto out;
+
+ frame = afr_frame_create(healer->this, &ret);
+ if (!frame) {
+ ret = -ret;
+ goto out;
+ }
+
+ ret = syncop_mt_dir_scan(frame, priv->children[healer->subvol], &loc,
+ GF_CLIENT_PID_SELF_HEALD, healer,
+ afr_shd_anon_inode_cleaner, NULL,
+ priv->shd.max_threads, priv->shd.wait_qlength);
+out:
+ if (frame)
+ AFR_STACK_DESTROY(frame);
+ loc_wipe(&loc);
+ return;
+}
+
void *
afr_shd_index_healer(void *data)
{
@@ -900,6 +1070,10 @@ afr_shd_index_healer(void *data)
sleep(1);
} while (ret > 0);
+ if (ret == 0) {
+ afr_cleanup_anon_inode_dir(healer);
+ }
+
if (ret == 0 && pre_crawl_xdata &&
!healer->crawl_event.heal_failed_count) {
afr_shd_ta_check_and_unset_xattrs(this, &loc, healer,
diff --git a/xlators/cluster/afr/src/afr-self-heald.h b/xlators/cluster/afr/src/afr-self-heald.h
index 687c28e6472..18db728ea7b 100644
--- a/xlators/cluster/afr/src/afr-self-heald.h
+++ b/xlators/cluster/afr/src/afr-self-heald.h
@@ -70,6 +70,6 @@ afr_shd_gfid_to_path(xlator_t *this, xlator_t *subvol, uuid_t gfid,
char **path_p);
int
-afr_shd_index_purge(xlator_t *subvol, inode_t *inode, char *name,
+afr_shd_entry_purge(xlator_t *subvol, inode_t *inode, char *name,
ia_type_t type);
#endif /* !_AFR_SELF_HEALD_H */
diff --git a/xlators/cluster/afr/src/afr.c b/xlators/cluster/afr/src/afr.c
index a38489d9932..df7366f0a65 100644
--- a/xlators/cluster/afr/src/afr.c
+++ b/xlators/cluster/afr/src/afr.c
@@ -135,6 +135,27 @@ set_data_self_heal_algorithm(afr_private_t *priv, char *algo)
}
}
+void
+afr_handle_anon_inode_options(afr_private_t *priv, dict_t *options)
+{
+ char *volfile_id_str = NULL;
+ uuid_t anon_inode_gfid = {0};
+
+ /*If volume id is not present don't enable anything*/
+ if (dict_get_str(options, "volume-id", &volfile_id_str))
+ return;
+ GF_ASSERT(strlen(AFR_ANON_DIR_PREFIX) + strlen(volfile_id_str) <= NAME_MAX);
+ /*anon_inode_name is not supposed to change once assigned*/
+ if (!priv->anon_inode_name[0]) {
+ snprintf(priv->anon_inode_name, sizeof(priv->anon_inode_name), "%s-%s",
+ AFR_ANON_DIR_PREFIX, volfile_id_str);
+ gf_uuid_parse(volfile_id_str, anon_inode_gfid);
+ /*Flip a bit to make sure volfile-id and anon-gfid are not same*/
+ anon_inode_gfid[0] ^= 1;
+ uuid_utoa_r(anon_inode_gfid, priv->anon_gfid_str);
+ }
+}
+
int
reconfigure(xlator_t *this, dict_t *options)
{
@@ -168,7 +189,8 @@ reconfigure(xlator_t *this, dict_t *options)
bool, out);
GF_OPTION_RECONF("data-self-heal", data_self_heal, options, str, out);
- gf_string2boolean(data_self_heal, &priv->data_self_heal);
+ if (gf_string2boolean(data_self_heal, &priv->data_self_heal) == -1)
+ goto out;
GF_OPTION_RECONF("entry-self-heal", priv->entry_self_heal, options, bool,
out);
@@ -289,6 +311,10 @@ reconfigure(xlator_t *this, dict_t *options)
consistent_io = _gf_false;
priv->consistent_io = consistent_io;
+ afr_handle_anon_inode_options(priv, options);
+
+ GF_OPTION_RECONF("use-anonymous-inode", priv->use_anon_inode, options, bool,
+ out);
if (priv->shd.enabled) {
if ((priv->shd.enabled != enabled_old) ||
(timeout_old != priv->shd.timeout))
@@ -485,7 +511,8 @@ init(xlator_t *this)
GF_OPTION_INIT("heal-wait-queue-length", priv->heal_wait_qlen, uint32, out);
GF_OPTION_INIT("data-self-heal", data_self_heal, str, out);
- gf_string2boolean(data_self_heal, &priv->data_self_heal);
+ if (gf_string2boolean(data_self_heal, &priv->data_self_heal) == -1)
+ goto out;
GF_OPTION_INIT("data-self-heal-algorithm", data_self_heal_algorithm, str,
out);
@@ -539,7 +566,9 @@ init(xlator_t *this)
GF_OPTION_INIT("consistent-metadata", priv->consistent_metadata, bool, out);
GF_OPTION_INIT("consistent-io", priv->consistent_io, bool, out);
+ afr_handle_anon_inode_options(priv, this->options);
+ GF_OPTION_INIT("use-anonymous-inode", priv->use_anon_inode, bool, out);
if (priv->quorum_count != 0)
priv->consistent_io = _gf_false;
@@ -551,6 +580,9 @@ init(xlator_t *this)
goto out;
}
+ priv->anon_inode = GF_CALLOC(sizeof(unsigned char), child_count,
+ gf_afr_mt_char);
+
priv->child_up = GF_CALLOC(sizeof(unsigned char), child_count,
gf_afr_mt_char);
@@ -559,7 +591,8 @@ init(xlator_t *this)
priv->halo_child_up = GF_CALLOC(sizeof(unsigned char), child_count,
gf_afr_mt_char);
- if (!priv->child_up || !priv->child_latency || !priv->halo_child_up) {
+ if (!priv->child_up || !priv->child_latency || !priv->halo_child_up ||
+ !priv->anon_inode) {
ret = -ENOMEM;
goto out;
}
@@ -1284,6 +1317,14 @@ struct volume_options options[] = {
.tags = {"replicate"},
.description = "This option exists only for backward compatibility "
"and configuring it doesn't have any effect"},
+ {.key = {"use-anonymous-inode"},
+ .type = GF_OPTION_TYPE_BOOL,
+ .default_value = "no",
+ .op_version = {GD_OP_VERSION_8_0},
+ .flags = OPT_FLAG_CLIENT_OPT | OPT_FLAG_SETTABLE,
+ .tags = {"replicate"},
+ .description = "Setting this option heals directory renames efficiently"},
+
{.key = {NULL}},
};
diff --git a/xlators/cluster/afr/src/afr.h b/xlators/cluster/afr/src/afr.h
index 23a0469f3d4..d62f9a9caf2 100644
--- a/xlators/cluster/afr/src/afr.h
+++ b/xlators/cluster/afr/src/afr.h
@@ -42,6 +42,7 @@
#define AFR_LK_HEAL_DOM "afr.lock-heal.domain"
#define AFR_HALO_MAX_LATENCY 99999
+#define AFR_ANON_DIR_PREFIX ".glusterfs-anonymous-inode"
#define PFLAG_PENDING (1 << 0)
#define PFLAG_SBRAIN (1 << 1)
@@ -190,6 +191,7 @@ typedef struct _afr_private {
struct list_head ta_waitq;
struct list_head ta_onwireq;
+ unsigned char *anon_inode;
unsigned char *child_up;
unsigned char *halo_child_up;
int64_t *child_latency;
@@ -275,10 +277,15 @@ typedef struct _afr_private {
gf_boolean_t esh_granular;
gf_boolean_t consistent_io;
gf_boolean_t data_self_heal; /* on/off */
+ gf_boolean_t use_anon_inode;
/*For lock healing.*/
struct list_head saved_locks;
struct list_head lk_healq;
+
+ /*For anon-inode handling */
+ char anon_inode_name[NAME_MAX + 1];
+ char anon_gfid_str[UUID_SIZE + 1];
} afr_private_t;
typedef enum {
@@ -1271,8 +1278,8 @@ int
afr_inode_split_brain_choice_set(inode_t *inode, xlator_t *this,
int spb_choice);
int
-afr_inode_split_brain_choice_get(inode_t *inode, xlator_t *this,
- int *spb_choice);
+afr_split_brain_read_subvol_get(inode_t *inode, xlator_t *this,
+ call_frame_t *frame, int *spb_subvol);
int
afr_get_child_index_from_name(xlator_t *this, char *name);
@@ -1409,4 +1416,8 @@ afr_dom_lock_release(call_frame_t *frame);
void
afr_fill_success_replies(afr_local_t *local, afr_private_t *priv,
unsigned char *replies);
+
+gf_boolean_t
+afr_is_private_directory(afr_private_t *priv, uuid_t pargfid, const char *name,
+ pid_t pid);
#endif /* __AFR_H__ */
diff --git a/xlators/cluster/dht/src/dht-common.c b/xlators/cluster/dht/src/dht-common.c
index e3aed576d75..8ba0cc4c732 100644
--- a/xlators/cluster/dht/src/dht-common.c
+++ b/xlators/cluster/dht/src/dht-common.c
@@ -17,6 +17,7 @@
#include <glusterfs/quota-common-utils.h>
#include <glusterfs/upcall-utils.h>
#include "glusterfs/compat-errno.h" // for ENODATA on BSD
+#include <glusterfs/common-utils.h>
#include <sys/time.h>
#include <libgen.h>
@@ -43,15 +44,6 @@ dht_common_mark_mdsxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
static int
dht_rmdir_unlock(call_frame_t *frame, xlator_t *this);
-char *xattrs_to_heal[] = {"user.",
- POSIX_ACL_ACCESS_XATTR,
- POSIX_ACL_DEFAULT_XATTR,
- QUOTA_LIMIT_KEY,
- QUOTA_LIMIT_OBJECTS_KEY,
- GF_SELINUX_XATTR_KEY,
- GF_XATTR_MDATA_KEY,
- NULL};
-
static const char *dht_dbg_vxattrs[] = {DHT_DBG_HASHED_SUBVOL_PATTERN, NULL};
/* Check the xdata to make sure EBADF has been set by client xlator */
@@ -84,6 +76,8 @@ dht_set_fixed_dir_stat(struct iatt *stat)
static gf_boolean_t
dht_match_xattr(const char *key)
{
+ char **xattrs_to_heal = get_xattrs_to_heal();
+
return gf_get_index_by_elem(xattrs_to_heal, (char *)key) >= 0;
}
@@ -388,7 +382,7 @@ out:
/* Code to save hashed subvol on inode ctx as a mds subvol
*/
-static int
+int
dht_inode_ctx_mdsvol_set(inode_t *inode, xlator_t *this, xlator_t *mds_subvol)
{
dht_inode_ctx_t *ctx = NULL;
@@ -619,13 +613,14 @@ dht_discover_complete(xlator_t *this, call_frame_t *discover_frame)
if (local->need_xattr_heal && !heal_path) {
local->need_xattr_heal = 0;
- ret = dht_dir_xattr_heal(this, local);
- if (ret)
- gf_msg(this->name, GF_LOG_ERROR, ret,
+ ret = dht_dir_xattr_heal(this, local, &op_errno);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, op_errno,
DHT_MSG_DIR_XATTR_HEAL_FAILED,
"xattr heal failed for "
"directory gfid is %s ",
gfid_local);
+ }
}
}
@@ -695,6 +690,7 @@ dht_common_mark_mdsxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
int ret = -1;
dht_conf_t *conf = 0;
dht_layout_t *layout = NULL;
+ int32_t mds_heal_fresh_lookup = 0;
GF_VALIDATE_OR_GOTO(this->name, frame, out);
GF_VALIDATE_OR_GOTO(this->name, frame->local, out);
@@ -702,6 +698,7 @@ dht_common_mark_mdsxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
local = frame->local;
conf = this->private;
layout = local->selfheal.layout;
+ mds_heal_fresh_lookup = local->mds_heal_fresh_lookup;
if (op_ret) {
gf_msg_debug(this->name, op_ret,
@@ -722,7 +719,7 @@ dht_common_mark_mdsxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
layout);
}
out:
- if (local && local->mds_heal_fresh_lookup)
+ if (mds_heal_fresh_lookup)
DHT_STACK_DESTROY(frame);
return 0;
}
@@ -1256,7 +1253,7 @@ err:
to non hashed subvol
*/
int
-dht_dir_xattr_heal(xlator_t *this, dht_local_t *local)
+dht_dir_xattr_heal(xlator_t *this, dht_local_t *local, int *op_errno)
{
dht_local_t *copy_local = NULL;
call_frame_t *copy = NULL;
@@ -1268,6 +1265,7 @@ dht_dir_xattr_heal(xlator_t *this, dht_local_t *local)
"No gfid exists for path %s "
"so healing xattr is not possible",
local->loc.path);
+ *op_errno = EIO;
goto out;
}
@@ -1281,6 +1279,7 @@ dht_dir_xattr_heal(xlator_t *this, dht_local_t *local)
"Memory allocation failed "
"for path %s gfid %s ",
local->loc.path, gfid_local);
+ *op_errno = ENOMEM;
DHT_STACK_DESTROY(copy);
} else {
copy_local->stbuf = local->stbuf;
@@ -1295,6 +1294,7 @@ dht_dir_xattr_heal(xlator_t *this, dht_local_t *local)
"Synctask creation failed to heal xattr "
"for path %s gfid %s ",
local->loc.path, gfid_local);
+ *op_errno = ENOMEM;
DHT_STACK_DESTROY(copy);
}
}
@@ -1651,7 +1651,7 @@ dht_revalidate_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
uint32_t vol_commit_hash = 0;
xlator_t *subvol = NULL;
int32_t check_mds = 0;
- int errst = 0;
+ int errst = 0, i = 0;
int32_t mds_xattr_val[1] = {0};
GF_VALIDATE_OR_GOTO("dht", frame, err);
@@ -1718,6 +1718,14 @@ dht_revalidate_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
local->need_lookup_everywhere = 1;
} else if (IA_ISDIR(local->loc.inode->ia_type)) {
+ layout = local->layout;
+ for (i = 0; i < layout->cnt; i++) {
+ if (layout->list[i].xlator == prev) {
+ layout->list[i].err = op_errno;
+ break;
+ }
+ }
+
local->need_selfheal = 1;
}
}
@@ -2153,31 +2161,18 @@ static int
dht_fill_dict_to_avoid_unlink_of_migrating_file(dict_t *dict)
{
int ret = 0;
- xlator_t *this = NULL;
- char *linktoskip_key = NULL;
- this = THIS;
- GF_VALIDATE_OR_GOTO("dht", this, err);
-
- if (dht_is_tier_xlator(this))
- linktoskip_key = TIER_SKIP_NON_LINKTO_UNLINK;
- else
- linktoskip_key = DHT_SKIP_NON_LINKTO_UNLINK;
-
- ret = dict_set_int32(dict, linktoskip_key, 1);
+ ret = dict_set_int32_sizen(dict, DHT_SKIP_NON_LINKTO_UNLINK, 1);
if (ret)
- goto err;
+ return -1;
- ret = dict_set_int32(dict, DHT_SKIP_OPEN_FD_UNLINK, 1);
+ ret = dict_set_int32_sizen(dict, DHT_SKIP_OPEN_FD_UNLINK, 1);
if (ret)
- goto err;
+ return -1;
return 0;
-
-err:
- return -1;
}
static int32_t
@@ -4306,6 +4301,8 @@ dht_find_local_subvol_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
index = conf->local_subvols_cnt;
uuid_list_copy = gf_strdup(uuid_list);
+ if (!uuid_list_copy)
+ goto unlock;
for (uuid_str = strtok_r(uuid_list, " ", &saveptr); uuid_str;
uuid_str = next_uuid_str) {
@@ -4596,18 +4593,8 @@ dht_getxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
dict_del(xattr, conf->xattr_name);
dict_del(xattr, conf->mds_xattr_key);
- /* filter out following two xattrs that need not
- * be visible on the mount point for geo-rep -
- * trusted.tier.fix.layout.complete and
- * trusted.tier.tier-dht.commithash
- */
-
dict_del(xattr, conf->commithash_xattr_name);
- if (frame->root->pid >= 0 && dht_is_tier_xlator(this)) {
- dict_del(xattr, GF_XATTR_TIER_LAYOUT_FIXED_KEY);
- }
-
if (frame->root->pid >= 0) {
GF_REMOVE_INTERNAL_XATTR("trusted.glusterfs.quota*", xattr);
GF_REMOVE_INTERNAL_XATTR("trusted.pgfid*", xattr);
@@ -5446,11 +5433,13 @@ dht_dir_common_set_remove_xattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
int call_cnt = 0;
dht_local_t *local = NULL;
char gfid_local[GF_UUID_BUF_SIZE] = {0};
+ char **xattrs_to_heal;
conf = this->private;
local = frame->local;
call_cnt = conf->subvolume_cnt;
local->flags = flags;
+ xattrs_to_heal = get_xattrs_to_heal();
if (!gf_uuid_is_null(local->gfid)) {
gf_uuid_unparse(local->gfid, gfid_local);
@@ -5883,22 +5872,7 @@ dht_setxattr(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xattr,
if (local->rebalance.target_node) {
local->flags = forced_rebalance;
- /* Flag to suggest its a tiering migration
- * The reason for this dic key-value is that
- * promotions and demotions are multithreaded
- * so the original frame from gf_defrag_start()
- * is not carried. A new frame will be created when
- * we do syncop_setxattr(). This does not have the
- * frame->root->pid of the original frame. So we pass
- * this dic key-value when we do syncop_setxattr() to do
- * data migration and set the frame->root->pid to
- * GF_CLIENT_PID_TIER_DEFRAG in dht_setxattr() just before
- * calling dht_start_rebalance_task() */
- tmp = dict_get(xattr, TIERING_MIGRATION_KEY);
- if (tmp)
- frame->root->pid = GF_CLIENT_PID_TIER_DEFRAG;
- else
- frame->root->pid = GF_CLIENT_PID_DEFRAG;
+ frame->root->pid = GF_CLIENT_PID_DEFRAG;
ret = dht_start_rebalance_task(this, frame);
if (!ret)
@@ -6710,10 +6684,9 @@ dht_readdirp_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
layout = local->layout;
- /* We have seen crashes in while running "rm -rf" on tier volumes
- when the layout was NULL on the hot tier. This will skip the
- entries on the subvol without a layout, hence preventing the crash
- but rmdir might fail with "directory not empty" errors*/
+ /* This will skip the entries on the subvol without a layout,
+ * hence preventing the crash but rmdir might fail with
+ * "directory not empty" errors*/
if (layout == NULL)
goto done;
@@ -10840,23 +10813,17 @@ dht_notify(xlator_t *this, int event, void *data, ...)
int had_heard_from_all = 0;
int have_heard_from_all = 0;
- struct timeval time = {
- 0,
- };
gf_defrag_info_t *defrag = NULL;
dict_t *dict = NULL;
gf_defrag_type cmd = 0;
dict_t *output = NULL;
va_list ap;
- dht_methods_t *methods = NULL;
struct gf_upcall *up_data = NULL;
struct gf_upcall_cache_invalidation *up_ci = NULL;
conf = this->private;
GF_VALIDATE_OR_GOTO(this->name, conf, out);
- methods = &(conf->methods);
-
/* had all subvolumes reported status once till now? */
had_heard_from_all = 1;
for (i = 0; i < conf->subvolume_cnt; i++) {
@@ -10886,12 +10853,11 @@ dht_notify(xlator_t *this, int event, void *data, ...)
break;
}
- gettimeofday(&time, NULL);
LOCK(&conf->subvolume_lock);
{
conf->subvolume_status[cnt] = 1;
conf->last_event[cnt] = event;
- conf->subvol_up_time[cnt] = time.tv_sec;
+ conf->subvol_up_time[cnt] = gf_time();
}
UNLOCK(&conf->subvolume_lock);
@@ -11080,15 +11046,13 @@ dht_notify(xlator_t *this, int event, void *data, ...)
* thread has already started.
*/
if (conf->defrag && !run_defrag) {
- if (methods->migration_needed(this)) {
- run_defrag = 1;
- ret = gf_thread_create(&conf->defrag->th, NULL, gf_defrag_start,
- this, "dhtdg");
- if (ret) {
- GF_FREE(conf->defrag);
- conf->defrag = NULL;
- kill(getpid(), SIGTERM);
- }
+ run_defrag = 1;
+ ret = gf_thread_create(&conf->defrag->th, NULL, gf_defrag_start,
+ this, "dhtdg");
+ if (ret) {
+ GF_FREE(conf->defrag);
+ conf->defrag = NULL;
+ kill(getpid(), SIGTERM);
}
}
}
@@ -11233,28 +11197,6 @@ out:
return ret;
}
-int32_t
-dht_migration_needed(xlator_t *this)
-{
- gf_defrag_info_t *defrag = NULL;
- dht_conf_t *conf = NULL;
- int ret = 0;
-
- conf = this->private;
-
- GF_VALIDATE_OR_GOTO("dht", conf, out);
- GF_VALIDATE_OR_GOTO("dht", conf->defrag, out);
-
- defrag = conf->defrag;
-
- if ((defrag->cmd != GF_DEFRAG_CMD_START_TIER) &&
- (defrag->cmd != GF_DEFRAG_CMD_START_DETACH_TIER))
- ret = 1;
-
-out:
- return ret;
-}
-
/*
This function should not be called more then once during a FOP
handling path. It is valid only for for ops on files
@@ -11289,14 +11231,6 @@ dht_set_local_rebalance(xlator_t *this, dht_local_t *local, struct iatt *stbuf,
return 0;
}
-gf_boolean_t
-dht_is_tier_xlator(xlator_t *this)
-{
- if (strcmp(this->type, "cluster/tier") == 0)
- return _gf_true;
- return _gf_false;
-}
-
int32_t
dht_release(xlator_t *this, fd_t *fd)
{
@@ -11436,3 +11370,22 @@ dht_pt_fgetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd, const char *key,
FIRST_CHILD(this)->fops->fgetxattr, fd, key, xdata);
return 0;
}
+
+/* The job of this function is to check if all the xlators have updated
+ * error in the layout. */
+int
+dht_dir_layout_error_check(xlator_t *this, inode_t *inode)
+{
+ dht_layout_t *layout = NULL;
+ int i = 0;
+
+ layout = dht_layout_get(this, inode);
+ for (i = 0; i < layout->cnt; i++) {
+ if (layout->list[i].err == 0) {
+ return 0;
+ }
+ }
+
+ /* Returning the first xlator error as all xlators have errors */
+ return layout->list[0].err;
+}
diff --git a/xlators/cluster/dht/src/dht-common.h b/xlators/cluster/dht/src/dht-common.h
index 8a49eefe9b4..fe0dc3db34a 100644
--- a/xlators/cluster/dht/src/dht-common.h
+++ b/xlators/cluster/dht/src/dht-common.h
@@ -24,7 +24,6 @@
#define _DHT_H
#define GF_XATTR_FIX_LAYOUT_KEY "distribute.fix.layout"
-#define GF_XATTR_TIER_LAYOUT_FIXED_KEY "trusted.tier.fix.layout.complete"
#define GF_XATTR_FILE_MIGRATE_KEY "trusted.distribute.migrate-data"
#define DHT_MDS_STR "mds"
#define GF_DHT_LOOKUP_UNHASHED_OFF 0
@@ -36,7 +35,6 @@
#define DHT_LAYOUT_HEAL_DOMAIN "dht.layout.heal"
/* Namespace synchronization */
#define DHT_ENTRY_SYNC_DOMAIN "dht.entry.sync"
-#define TIERING_MIGRATION_KEY "tiering.migration"
#define DHT_LAYOUT_HASH_INVALID 1
#define MAX_REBAL_THREADS sysconf(_SC_NPROCESSORS_ONLN)
@@ -52,10 +50,6 @@
#define DHT_DBG_HASHED_SUBVOL_PATTERN "dht.file.hashed-subvol.*"
#define DHT_DBG_HASHED_SUBVOL_KEY "dht.file.hashed-subvol."
-/* Array to hold custom xattr keys
- */
-extern char *xattrs_to_heal[];
-
/* Rebalance nodeuuid flags */
#define REBAL_NODEUUID_MINE 0x01
@@ -246,19 +240,6 @@ typedef gf_boolean_t (*dht_need_heal_t)(call_frame_t *frame,
dht_layout_t **inmem,
dht_layout_t **ondisk);
-typedef struct {
- uint64_t blocks_used;
- uint64_t pblocks_used;
- uint64_t files_used;
- uint64_t pfiles_used;
- uint64_t unhashed_blocks_used;
- uint64_t unhashed_pblocks_used;
- uint64_t unhashed_files_used;
- uint64_t unhashed_pfiles_used;
- uint64_t unhashed_fsid;
- uint64_t hashed_fsid;
-} tier_statvfs_t;
-
struct dht_local {
loc_t loc;
loc_t loc2;
@@ -276,7 +257,6 @@ struct dht_local {
struct iatt preparent;
struct iatt postparent;
struct statvfs statvfs;
- tier_statvfs_t tier_statvfs;
fd_t *fd;
inode_t *inode;
dict_t *params;
@@ -409,14 +389,7 @@ enum gf_defrag_type {
GF_DEFRAG_CMD_STATUS = 1 + 2,
GF_DEFRAG_CMD_START_LAYOUT_FIX = 1 + 3,
GF_DEFRAG_CMD_START_FORCE = 1 + 4,
- GF_DEFRAG_CMD_START_TIER = 1 + 5,
- GF_DEFRAG_CMD_STATUS_TIER = 1 + 6,
- GF_DEFRAG_CMD_START_DETACH_TIER = 1 + 7,
- GF_DEFRAG_CMD_STOP_DETACH_TIER = 1 + 8,
- GF_DEFRAG_CMD_PAUSE_TIER = 1 + 9,
- GF_DEFRAG_CMD_RESUME_TIER = 1 + 10,
GF_DEFRAG_CMD_DETACH_STATUS = 1 + 11,
- GF_DEFRAG_CMD_STOP_TIER = 1 + 12,
GF_DEFRAG_CMD_DETACH_START = 1 + 13,
GF_DEFRAG_CMD_DETACH_COMMIT = 1 + 14,
GF_DEFRAG_CMD_DETACH_COMMIT_FORCE = 1 + 15,
@@ -467,75 +440,6 @@ struct dht_container {
int local_subvol_index;
};
-typedef enum tier_mode_ {
- TIER_MODE_NONE = 0,
- TIER_MODE_TEST,
- TIER_MODE_WM
-} tier_mode_t;
-
-typedef enum tier_pause_state_ {
- TIER_RUNNING = 0,
- TIER_REQUEST_PAUSE,
- TIER_PAUSED
-} tier_pause_state_t;
-
-/* This Structure is only used in tiering fixlayout */
-typedef struct gf_tier_fix_layout_arg {
- xlator_t *this;
- dict_t *fix_layout;
- pthread_t thread_id;
-} gf_tier_fix_layout_arg_t;
-
-typedef struct gf_tier_conf {
- int is_tier;
- int watermark_hi;
- int watermark_low;
- int watermark_last;
- unsigned long block_size;
- fsblkcnt_t blocks_total;
- fsblkcnt_t blocks_used;
- uint64_t max_migrate_bytes;
- int max_migrate_files;
- int query_limit;
- tier_mode_t mode;
- int percent_full;
- /* These flags are only used for tier-compact */
- gf_boolean_t compact_active;
- /* These 3 flags are set to true when the client changes the */
- /* compaction mode on the command line. */
- /* When they are set, the daemon will trigger compaction as */
- /* soon as possible to activate or deactivate compaction. */
- /* If in the middle of a compaction, then the switches take */
- /* effect on the next compaction, not the current one. */
- /* If the user switches it off, we want to avoid needless */
- /* compactions. */
- /* If the user switches it on, they want to compact as soon */
- /* as possible. */
- gf_boolean_t compact_mode_switched;
- gf_boolean_t compact_mode_switched_hot;
- gf_boolean_t compact_mode_switched_cold;
- int tier_max_promote_size;
- int tier_promote_frequency;
- int tier_demote_frequency;
- int tier_compact_hot_frequency;
- int tier_compact_cold_frequency;
- uint64_t st_last_promoted_size;
- uint64_t st_last_demoted_size;
- struct synctask *pause_synctask;
- gf_timer_t *pause_timer;
- pthread_mutex_t pause_mutex;
- int promote_in_progress;
- int demote_in_progress;
- /* This Structure is only used in tiering fixlayout */
- gf_tier_fix_layout_arg_t tier_fix_layout_arg;
- /* Indicates the index of the first queryfile picked
- * in the last cycle of promote or demote */
- int32_t last_promote_qfile_index;
- int32_t last_demote_qfile_index;
- tier_pause_state_t pause_state;
- char volname[GD_VOLUME_NAME_MAX + 1];
-} gf_tier_conf_t;
-
typedef struct nodeuuid_info {
char info; /* Set to 1 is this is my node's uuid*/
uuid_t uuid; /* Store the nodeuuid as well for debugging*/
@@ -563,17 +467,10 @@ struct gf_defrag_info_ {
int cmd;
inode_t *root_inode;
uuid_t node_uuid;
- struct timeval start_time;
+ time_t start_time;
uint32_t new_commit_hash;
gf_defrag_status_t defrag_status;
gf_defrag_pattern_list_t *defrag_pattern;
- gf_tier_conf_t tier_conf;
-
- /*Data Tiering params for scanner*/
- uint64_t total_files_promoted;
- uint64_t total_files_demoted;
- int write_freq_threshold;
- int read_freq_threshold;
pthread_cond_t parallel_migration_cond;
pthread_mutex_t dfq_mutex;
@@ -609,7 +506,6 @@ typedef struct gf_defrag_info_ gf_defrag_info_t;
struct dht_methods_s {
int32_t (*migration_get_dst_subvol)(xlator_t *this, dht_local_t *local);
int32_t (*migration_other)(xlator_t *this, gf_defrag_info_t *defrag);
- int32_t (*migration_needed)(xlator_t *this);
xlator_t *(*layout_search)(xlator_t *this, dht_layout_t *layout,
const char *name);
};
@@ -630,7 +526,7 @@ struct dht_conf {
int subvolume_cnt;
int32_t refresh_interval;
gf_lock_t subvolume_lock;
- struct timeval last_stat_fetch;
+ time_t last_stat_fetch;
gf_lock_t layout_lock;
dict_t *leaf_to_subvol;
void *private; /* Can be used by wrapper xlators over
@@ -752,6 +648,8 @@ struct dir_dfmeta {
struct list_head **head;
struct list_head **iterator;
int *fetch_entries;
+ /* fds corresponding to local subvols only */
+ fd_t **lfd;
};
typedef struct dht_migrate_info {
@@ -1318,9 +1216,6 @@ dht_layout_missing_dirs(dht_layout_t *layout);
int
dht_refresh_layout(call_frame_t *frame);
-gf_boolean_t
-dht_is_tier_xlator(xlator_t *this);
-
int
dht_build_parent_loc(xlator_t *this, loc_t *parent, loc_t *child,
int32_t *op_errno);
@@ -1433,7 +1328,7 @@ dht_dir_set_heal_xattr(xlator_t *this, dht_local_t *local, dict_t *dst,
dict_t *src, int *uret, int *uflag);
int
-dht_dir_xattr_heal(xlator_t *this, dht_local_t *local);
+dht_dir_xattr_heal(xlator_t *this, dht_local_t *local, int *op_errno);
int
dht_common_mark_mdsxattr(call_frame_t *frame, int *errst, int flag);
@@ -1481,4 +1376,9 @@ dht_create_lock(call_frame_t *frame, xlator_t *subvol);
int
dht_set_parent_layout_in_dict(loc_t *loc, xlator_t *this, dht_local_t *local);
+int
+dht_dir_layout_error_check(xlator_t *this, inode_t *inode);
+
+int
+dht_inode_ctx_mdsvol_set(inode_t *inode, xlator_t *this, xlator_t *mds_subvol);
#endif /* _DHT_H */
diff --git a/xlators/cluster/dht/src/dht-diskusage.c b/xlators/cluster/dht/src/dht-diskusage.c
index 27097ca2475..c0588828fdb 100644
--- a/xlators/cluster/dht/src/dht-diskusage.c
+++ b/xlators/cluster/dht/src/dht-diskusage.c
@@ -151,22 +151,18 @@ dht_get_du_info(call_frame_t *frame, xlator_t *this, loc_t *loc)
dht_conf_t *conf = NULL;
call_frame_t *statfs_frame = NULL;
dht_local_t *statfs_local = NULL;
- struct timeval tv = {
- 0,
- };
loc_t tmp_loc = {
0,
};
+ time_t now;
conf = this->private;
-
- gettimeofday(&tv, NULL);
-
+ now = gf_time();
/* make it root gfid, should be enough to get the proper
info back */
tmp_loc.gfid[15] = 1;
- if (tv.tv_sec > (conf->refresh_interval + conf->last_stat_fetch.tv_sec)) {
+ if (now > (conf->refresh_interval + conf->last_stat_fetch)) {
statfs_frame = copy_frame(frame);
if (!statfs_frame) {
goto err;
@@ -198,7 +194,7 @@ dht_get_du_info(call_frame_t *frame, xlator_t *this, loc_t *loc)
statfs_local->params);
}
- conf->last_stat_fetch.tv_sec = tv.tv_sec;
+ conf->last_stat_fetch = now;
}
return 0;
err:
diff --git a/xlators/cluster/dht/src/dht-helper.c b/xlators/cluster/dht/src/dht-helper.c
index 73a89399efd..3f2fe43d5f3 100644
--- a/xlators/cluster/dht/src/dht-helper.c
+++ b/xlators/cluster/dht/src/dht-helper.c
@@ -2083,6 +2083,7 @@ dht_heal_full_path_done(int op_ret, call_frame_t *heal_frame, void *data)
dht_local_t *local = NULL;
xlator_t *this = NULL;
int ret = -1;
+ int op_errno = 0;
local = heal_frame->local;
main_frame = local->main_frame;
@@ -2092,11 +2093,12 @@ dht_heal_full_path_done(int op_ret, call_frame_t *heal_frame, void *data)
dht_set_fixed_dir_stat(&local->postparent);
if (local->need_xattr_heal) {
local->need_xattr_heal = 0;
- ret = dht_dir_xattr_heal(this, local);
- if (ret)
- gf_smsg(this->name, GF_LOG_ERROR, ret,
+ ret = dht_dir_xattr_heal(this, local, &op_errno);
+ if (ret) {
+ gf_smsg(this->name, GF_LOG_ERROR, op_errno,
DHT_MSG_DIR_XATTR_HEAL_FAILED, "path=%s", local->loc.path,
NULL);
+ }
}
DHT_STACK_UNWIND(lookup, main_frame, 0, 0, local->inode, &local->stbuf,
@@ -2265,6 +2267,7 @@ dht_dir_set_heal_xattr(xlator_t *this, dht_local_t *local, dict_t *dst,
int luret = -1;
int luflag = -1;
int i = 0;
+ char **xattrs_to_heal;
if (!src || !dst) {
gf_smsg(this->name, GF_LOG_WARNING, EINVAL, DHT_MSG_DST_NULL_SET_FAILED,
@@ -2279,6 +2282,9 @@ dht_dir_set_heal_xattr(xlator_t *this, dht_local_t *local, dict_t *dst,
and set it to dst dict, here index start from 1 because
user xattr already checked in previous statement
*/
+
+ xattrs_to_heal = get_xattrs_to_heal();
+
for (i = 1; xattrs_to_heal[i]; i++) {
keyval = dict_get(src, xattrs_to_heal[i]);
if (keyval) {
diff --git a/xlators/cluster/dht/src/dht-inode-write.c b/xlators/cluster/dht/src/dht-inode-write.c
index eda2491e0ff..2f23ce90fbd 100644
--- a/xlators/cluster/dht/src/dht-inode-write.c
+++ b/xlators/cluster/dht/src/dht-inode-write.c
@@ -93,30 +93,28 @@ dht_writev_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
/* Check if the rebalance phase1 is true */
if (IS_DHT_MIGRATION_PHASE1(postbuf)) {
- if (!dht_is_tier_xlator(this)) {
+ if (!local->xattr_req) {
+ local->xattr_req = dict_new();
if (!local->xattr_req) {
- local->xattr_req = dict_new();
- if (!local->xattr_req) {
- gf_msg(this->name, GF_LOG_ERROR, DHT_MSG_NO_MEMORY, ENOMEM,
- "insufficient memory");
- local->op_errno = ENOMEM;
- local->op_ret = -1;
- goto out;
- }
- }
-
- ret = dict_set_uint32(local->xattr_req,
- GF_PROTECT_FROM_EXTERNAL_WRITES, 1);
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, DHT_MSG_DICT_SET_FAILED, 0,
- "Failed to set key %s in dictionary",
- GF_PROTECT_FROM_EXTERNAL_WRITES);
+ gf_msg(this->name, GF_LOG_ERROR, DHT_MSG_NO_MEMORY, ENOMEM,
+ "insufficient memory");
local->op_errno = ENOMEM;
local->op_ret = -1;
goto out;
}
}
+ ret = dict_set_uint32(local->xattr_req, GF_PROTECT_FROM_EXTERNAL_WRITES,
+ 1);
+ if (ret) {
+ gf_msg(this->name, GF_LOG_ERROR, DHT_MSG_DICT_SET_FAILED, 0,
+ "Failed to set key %s in dictionary",
+ GF_PROTECT_FROM_EXTERNAL_WRITES);
+ local->op_errno = ENOMEM;
+ local->op_ret = -1;
+ goto out;
+ }
+
dht_iatt_merge(this, &local->stbuf, postbuf);
dht_iatt_merge(this, &local->prebuf, prebuf);
diff --git a/xlators/cluster/dht/src/dht-mem-types.h b/xlators/cluster/dht/src/dht-mem-types.h
index 33f9832395b..e3c4471334a 100644
--- a/xlators/cluster/dht/src/dht-mem-types.h
+++ b/xlators/cluster/dht/src/dht-mem-types.h
@@ -30,10 +30,7 @@ enum gf_dht_mem_types_ {
gf_dht_mt_container_t,
gf_dht_mt_octx_t,
gf_dht_mt_miginfo_t,
- gf_tier_mt_bricklist_t,
- gf_tier_mt_ipc_ctr_params_t,
gf_dht_mt_fd_ctx_t,
- gf_tier_mt_qfile_array_t,
gf_dht_ret_cache_t,
gf_dht_nodeuuids_t,
gf_dht_mt_end
diff --git a/xlators/cluster/dht/src/dht-messages.h b/xlators/cluster/dht/src/dht-messages.h
index 026879e14af..601f8dad78b 100644
--- a/xlators/cluster/dht/src/dht-messages.h
+++ b/xlators/cluster/dht/src/dht-messages.h
@@ -38,12 +38,11 @@ GLFS_MSGID(
DHT_MSG_REBALANCE_STATUS, DHT_MSG_REBALANCE_STOPPED, DHT_MSG_RENAME_FAILED,
DHT_MSG_SETATTR_FAILED, DHT_MSG_SUBVOL_INSUFF_INODES,
DHT_MSG_SUBVOL_INSUFF_SPACE, DHT_MSG_UNLINK_FAILED,
- DHT_MSG_LAYOUT_SET_FAILED, DHT_MSG_LOG_FIXED_LAYOUT, DHT_MSG_LOG_TIER_ERROR,
- DHT_MSG_LOG_TIER_STATUS, DHT_MSG_GET_XATTR_FAILED,
- DHT_MSG_FILE_LOOKUP_FAILED, DHT_MSG_OPEN_FD_FAILED,
- DHT_MSG_SET_INODE_CTX_FAILED, DHT_MSG_UNLOCKING_FAILED,
- DHT_MSG_DISK_LAYOUT_NULL, DHT_MSG_SUBVOL_INFO, DHT_MSG_CHUNK_SIZE_INFO,
- DHT_MSG_LAYOUT_FORM_FAILED, DHT_MSG_SUBVOL_ERROR,
+ DHT_MSG_LAYOUT_SET_FAILED, DHT_MSG_LOG_FIXED_LAYOUT,
+ DHT_MSG_GET_XATTR_FAILED, DHT_MSG_FILE_LOOKUP_FAILED,
+ DHT_MSG_OPEN_FD_FAILED, DHT_MSG_SET_INODE_CTX_FAILED,
+ DHT_MSG_UNLOCKING_FAILED, DHT_MSG_DISK_LAYOUT_NULL, DHT_MSG_SUBVOL_INFO,
+ DHT_MSG_CHUNK_SIZE_INFO, DHT_MSG_LAYOUT_FORM_FAILED, DHT_MSG_SUBVOL_ERROR,
DHT_MSG_LAYOUT_SORT_FAILED, DHT_MSG_REGEX_INFO, DHT_MSG_FOPEN_FAILED,
DHT_MSG_SET_HOSTNAME_FAILED, DHT_MSG_BRICK_ERROR, DHT_MSG_SYNCOP_FAILED,
DHT_MSG_MIGRATE_INFO, DHT_MSG_SOCKET_ERROR, DHT_MSG_CREATE_FD_FAILED,
@@ -69,8 +68,7 @@ GLFS_MSGID(
DHT_MSG_INIT_LOCAL_SUBVOL_FAILED, DHT_MSG_SYS_CALL_GET_TIME_FAILED,
DHT_MSG_NO_DISK_USAGE_STATUS, DHT_MSG_SUBVOL_DOWN_ERROR,
DHT_MSG_REBAL_THROTTLE_INFO, DHT_MSG_COMMIT_HASH_INFO,
- DHT_MSG_REBAL_STRUCT_SET, DHT_MSG_HAS_MIGINFO, DHT_MSG_LOG_IPC_TIER_ERROR,
- DHT_MSG_TIER_PAUSED, DHT_MSG_TIER_RESUME, DHT_MSG_SETTLE_HASH_FAILED,
+ DHT_MSG_REBAL_STRUCT_SET, DHT_MSG_HAS_MIGINFO, DHT_MSG_SETTLE_HASH_FAILED,
DHT_MSG_DEFRAG_PROCESS_DIR_FAILED, DHT_MSG_FD_CTX_SET_FAILED,
DHT_MSG_STALE_LOOKUP, DHT_MSG_PARENT_LAYOUT_CHANGED,
DHT_MSG_LOCK_MIGRATION_FAILED, DHT_MSG_LOCK_INODE_UNREF_FAILED,
@@ -96,15 +94,13 @@ GLFS_MSGID(
DHT_MSG_UNLOCK_FILE_FAILED, DHT_MSG_REMOVE_XATTR_FAILED,
DHT_MSG_DATA_MIGRATE_ABORT, DHT_MSG_DEFRAG_NULL, DHT_MSG_PARENT_NULL,
DHT_MSG_GFID_NOT_PRESENT, DHT_MSG_CHILD_LOC_FAILED,
- DHT_MSG_SET_LOOKUP_FAILED, DHT_MSG_DIR_REMOVED,
- DHT_MSG_TIER_FIX_LAYOUT_STARTED, DHT_MSG_FIX_NOT_COMP,
- DHT_MSG_REMOVE_TIER_FAILED, DHT_MSG_SUBVOL_DETER_FAILED,
- DHT_MSG_LOCAL_SUBVOL, DHT_MSG_NODE_UUID, DHT_MSG_SIZE_FILE,
- DHT_MSG_GET_DATA_SIZE_FAILED, DHT_MSG_PTHREAD_JOIN_FAILED,
- DHT_MSG_COUNTER_THREAD_CREATE_FAILED, DHT_MSG_MIGRATION_INIT_QUEUE_FAILED,
- DHT_MSG_PAUSED_TIMEOUT, DHT_MSG_WOKE, DHT_MSG_ABORT_REBALANCE,
- DHT_MSG_CREATE_TASK_REBAL_FAILED, DHT_MSG_REBAL_ESTIMATE_NOT_AVAIL,
- DHT_MSG_MIG_TIER_PAUSED, DHT_MSG_ADD_CHOICES_ERROR,
+ DHT_MSG_SET_LOOKUP_FAILED, DHT_MSG_DIR_REMOVED, DHT_MSG_FIX_NOT_COMP,
+ DHT_MSG_SUBVOL_DETER_FAILED, DHT_MSG_LOCAL_SUBVOL, DHT_MSG_NODE_UUID,
+ DHT_MSG_SIZE_FILE, DHT_MSG_GET_DATA_SIZE_FAILED,
+ DHT_MSG_PTHREAD_JOIN_FAILED, DHT_MSG_COUNTER_THREAD_CREATE_FAILED,
+ DHT_MSG_MIGRATION_INIT_QUEUE_FAILED, DHT_MSG_PAUSED_TIMEOUT, DHT_MSG_WOKE,
+ DHT_MSG_ABORT_REBALANCE, DHT_MSG_CREATE_TASK_REBAL_FAILED,
+ DHT_MSG_REBAL_ESTIMATE_NOT_AVAIL, DHT_MSG_ADD_CHOICES_ERROR,
DHT_MSG_GET_CHOICES_ERROR, DHT_MSG_PREPARE_STATUS_ERROR,
DHT_MSG_SET_CHOICE_FAILED, DHT_MSG_SET_HASHED_SUBVOL_FAILED,
DHT_MSG_XATTR_HEAL_NOT_POSS, DHT_MSG_LINKTO_FILE_FAILED,
@@ -180,7 +176,6 @@ GLFS_MSGID(
"adding bricks"
#define DHT_MSG_NEW_TARGET_FOUND_STR "New target found for file"
#define DHT_MSG_INSUFF_MEMORY_STR "insufficient memory"
-#define DHT_MSG_MIG_TIER_PAUSED_STR "Migrate file paused"
#define DHT_MSG_SET_XATTR_FAILED_STR "failed to set xattr"
#define DHT_MSG_SET_MODE_FAILED_STR "failed to set mode"
#define DHT_MSG_FILE_EXISTS_IN_DEST_STR "file exists in destination"
@@ -222,17 +217,14 @@ GLFS_MSGID(
#define DHT_MSG_GFID_NOT_PRESENT_STR "gfid not present"
#define DHT_MSG_CHILD_LOC_FAILED_STR "Child loc build failed"
#define DHT_MSG_SET_LOOKUP_FAILED_STR "Failed to set lookup"
-#define DHT_MSG_LOG_TIER_STATUS_STR "lookup to cold tier on attach heal failed"
#define DHT_MSG_DIR_LOOKUP_FAILED_STR "lookup failed"
#define DHT_MSG_DIR_REMOVED_STR "Dir renamed or removed. Skipping"
#define DHT_MSG_READDIR_ERROR_STR "readdir failed, Aborting fix-layout"
#define DHT_MSG_SETTLE_HASH_FAILED_STR "Settle hash failed"
#define DHT_MSG_DEFRAG_PROCESS_DIR_FAILED_STR "gf_defrag_process_dir failed"
-#define DHT_MSG_TIER_FIX_LAYOUT_STARTED_STR "Tiering fix layout started"
#define DHT_MSG_FIX_NOT_COMP_STR \
"Unable to retrieve fixlayout xattr. Assume background fix layout not " \
"complete"
-#define DHT_MSG_REMOVE_TIER_FAILED_STR "Failed removing tier fix layout xattr"
#define DHT_MSG_SUBVOL_DETER_FAILED_STR \
"local subvolume determination failed with error"
#define DHT_MSG_LOCAL_SUBVOL_STR "local subvol"
@@ -248,8 +240,6 @@ GLFS_MSGID(
#define DHT_MSG_MIGRATION_INIT_QUEUE_FAILED_STR \
"Failed to initialise migration queue"
#define DHT_MSG_REBALANCE_STOPPED_STR "Received stop command on rebalance"
-#define DHT_MSG_TIER_RESUME_STR "Pause end. Resume tiering"
-#define DHT_MSG_TIER_PAUSED_STR "Pause tiering"
#define DHT_MSG_PAUSED_TIMEOUT_STR "Request pause timer timeout"
#define DHT_MSG_WOKE_STR "woken"
#define DHT_MSG_ABORT_REBALANCE_STR "Aborting rebalance"
diff --git a/xlators/cluster/dht/src/dht-rebalance.c b/xlators/cluster/dht/src/dht-rebalance.c
index 4f63e76aad4..8ba8082bd86 100644
--- a/xlators/cluster/dht/src/dht-rebalance.c
+++ b/xlators/cluster/dht/src/dht-rebalance.c
@@ -16,8 +16,8 @@
#include "glusterfs/compat-errno.h" // for ENODATA on BSD
#define GF_DISK_SECTOR_SIZE 512
-#define DHT_REBALANCE_PID 4242 /* Change it if required */
-#define DHT_REBALANCE_BLKSIZE (1024 * 1024) /* 1 MB */
+#define DHT_REBALANCE_PID 4242 /* Change it if required */
+#define DHT_REBALANCE_BLKSIZE 1048576 /* 1 MB */
#define MAX_MIGRATE_QUEUE_COUNT 500
#define MIN_MIGRATE_QUEUE_COUNT 200
#define MAX_REBAL_TYPE_SIZE 16
@@ -45,7 +45,10 @@ gf_defrag_free_dir_dfmeta(struct dir_dfmeta *meta, int local_subvols_cnt)
if (meta) {
for (i = 0; i < local_subvols_cnt; i++) {
- gf_dirent_free(&meta->equeue[i]);
+ if (meta->equeue)
+ gf_dirent_free(&meta->equeue[i]);
+ if (meta->lfd && meta->lfd[i])
+ fd_unref(meta->lfd[i]);
}
GF_FREE(meta->equeue);
@@ -53,6 +56,7 @@ gf_defrag_free_dir_dfmeta(struct dir_dfmeta *meta, int local_subvols_cnt)
GF_FREE(meta->iterator);
GF_FREE(meta->offset_var);
GF_FREE(meta->fetch_entries);
+ GF_FREE(meta->lfd);
GF_FREE(meta);
}
}
@@ -143,75 +147,6 @@ dht_strip_out_acls(dict_t *dict)
}
}
-static int
-dht_write_with_holes(xlator_t *to, fd_t *fd, struct iovec *vec, int count,
- int32_t size, off_t offset, struct iobref *iobref,
- int *fop_errno)
-{
- int i = 0;
- int ret = -1;
- int start_idx = 0;
- int tmp_offset = 0;
- int write_needed = 0;
- int buf_len = 0;
- int size_pending = 0;
- char *buf = NULL;
-
- /* loop through each vector */
- for (i = 0; i < count; i++) {
- buf = vec[i].iov_base;
- buf_len = vec[i].iov_len;
-
- for (start_idx = 0; (start_idx + GF_DISK_SECTOR_SIZE) <= buf_len;
- start_idx += GF_DISK_SECTOR_SIZE) {
- if (mem_0filled(buf + start_idx, GF_DISK_SECTOR_SIZE) != 0) {
- write_needed = 1;
- continue;
- }
-
- if (write_needed) {
- ret = syncop_write(
- to, fd, (buf + tmp_offset), (start_idx - tmp_offset),
- (offset + tmp_offset), iobref, 0, NULL, NULL);
- /* 'path' will be logged in calling function */
- if (ret < 0) {
- gf_log(THIS->name, GF_LOG_WARNING, "failed to write (%s)",
- strerror(-ret));
- *fop_errno = -ret;
- ret = -1;
- goto out;
- }
-
- write_needed = 0;
- }
- tmp_offset = start_idx + GF_DISK_SECTOR_SIZE;
- }
-
- if ((start_idx < buf_len) || write_needed) {
- /* This means, last chunk is not yet written.. write it */
- ret = syncop_write(to, fd, (buf + tmp_offset),
- (buf_len - tmp_offset), (offset + tmp_offset),
- iobref, 0, NULL, NULL);
- if (ret < 0) {
- /* 'path' will be logged in calling function */
- gf_log(THIS->name, GF_LOG_WARNING, "failed to write (%s)",
- strerror(-ret));
- *fop_errno = -ret;
- ret = -1;
- goto out;
- }
- }
-
- size_pending = (size - buf_len);
- if (!size_pending)
- break;
- }
-
- ret = size;
-out:
- return ret;
-}
-
/*
return values:
-1 : failure
@@ -619,7 +554,7 @@ out:
static int
__dht_rebalance_create_dst_file(xlator_t *this, xlator_t *to, xlator_t *from,
loc_t *loc, struct iatt *stbuf, fd_t **dst_fd,
- int *fop_errno)
+ int *fop_errno, int file_has_holes)
{
int ret = -1;
int ret2 = -1;
@@ -674,26 +609,23 @@ __dht_rebalance_create_dst_file(xlator_t *this, xlator_t *to, xlator_t *from,
goto out;
}
- if (!!dht_is_tier_xlator(this)) {
- xdata = dict_new();
- if (!xdata) {
- *fop_errno = ENOMEM;
- ret = -1;
- gf_msg(this->name, GF_LOG_ERROR, ENOMEM,
- DHT_MSG_MIGRATE_FILE_FAILED, "%s: dict_new failed)",
- loc->path);
- goto out;
- }
+ xdata = dict_new();
+ if (!xdata) {
+ *fop_errno = ENOMEM;
+ ret = -1;
+ gf_msg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_MIGRATE_FILE_FAILED,
+ "%s: dict_new failed)", loc->path);
+ goto out;
+ }
- ret = dict_set_int32(xdata, GF_CLEAN_WRITE_PROTECTION, 1);
- if (ret) {
- *fop_errno = ENOMEM;
- ret = -1;
- gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
- "%s: failed to set dictionary value: key = %s ", loc->path,
- GF_CLEAN_WRITE_PROTECTION);
- goto out;
- }
+ ret = dict_set_int32_sizen(xdata, GF_CLEAN_WRITE_PROTECTION, 1);
+ if (ret) {
+ *fop_errno = ENOMEM;
+ ret = -1;
+ gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
+ "%s: failed to set dictionary value: key = %s ", loc->path,
+ GF_CLEAN_WRITE_PROTECTION);
+ goto out;
}
ret = syncop_lookup(to, loc, &new_stbuf, NULL, xdata, NULL);
@@ -788,7 +720,7 @@ __dht_rebalance_create_dst_file(xlator_t *this, xlator_t *to, xlator_t *from,
/* No need to bother about 0 byte size files */
if (stbuf->ia_size > 0) {
- if (conf->use_fallocate) {
+ if (conf->use_fallocate && !file_has_holes) {
ret = syncop_fallocate(to, fd, 0, 0, stbuf->ia_size, NULL, NULL);
if (ret < 0) {
if (ret == -EOPNOTSUPP || ret == -EINVAL || ret == -ENOSYS) {
@@ -815,9 +747,7 @@ __dht_rebalance_create_dst_file(xlator_t *this, xlator_t *to, xlator_t *from,
goto out;
}
}
- }
-
- if (!conf->use_fallocate) {
+ } else {
ret = syncop_ftruncate(to, fd, stbuf->ia_size, NULL, NULL, NULL,
NULL);
if (ret < 0) {
@@ -1068,22 +998,90 @@ __dht_rebalance_migrate_data(xlator_t *this, gf_defrag_info_t *defrag,
int ret = 0;
int count = 0;
off_t offset = 0;
+ off_t data_offset = 0;
+ off_t hole_offset = 0;
struct iovec *vector = NULL;
struct iobref *iobref = NULL;
uint64_t total = 0;
size_t read_size = 0;
+ size_t data_block_size = 0;
dict_t *xdata = NULL;
dht_conf_t *conf = NULL;
conf = this->private;
+
/* if file size is '0', no need to enter this loop */
while (total < ia_size) {
- read_size = (((ia_size - total) > DHT_REBALANCE_BLKSIZE)
- ? DHT_REBALANCE_BLKSIZE
- : (ia_size - total));
+ /* This is a regular file - read it sequentially */
+ if (!hole_exists) {
+ read_size = (((ia_size - total) > DHT_REBALANCE_BLKSIZE)
+ ? DHT_REBALANCE_BLKSIZE
+ : (ia_size - total));
+ } else {
+ /* This is a sparse file - read only the data segments in the file
+ */
+
+ /* If the previous data block is fully copied, find the next data
+ * segment
+ * starting at the offset of the last read and written byte, */
+ if (data_block_size <= 0) {
+ ret = syncop_seek(from, src, offset, GF_SEEK_DATA, NULL,
+ &data_offset);
+ if (ret) {
+ if (ret == -ENXIO)
+ ret = 0; /* No more data segments */
+ else
+ *fop_errno = -ret; /* Error occurred */
+
+ break;
+ }
+
+ /* If the position of the current data segment is greater than
+ * the position of the next hole, find the next hole in order to
+ * calculate the length of the new data segment */
+ if (data_offset > hole_offset) {
+ /* Starting at the offset of the last data segment, find the
+ * next hole */
+ ret = syncop_seek(from, src, data_offset, GF_SEEK_HOLE,
+ NULL, &hole_offset);
+ if (ret) {
+ /* If an error occurred here it's a real error because
+ * if the seek for a data segment was successful then
+ * necessarily another hole must exist (EOF is a hole)
+ */
+ *fop_errno = -ret;
+ break;
+ }
+
+ /* Calculate the total size of the current data block */
+ data_block_size = hole_offset - data_offset;
+ }
+ } else {
+ /* There is still data in the current segment, move the
+ * data_offset to the position of the last written byte */
+ data_offset = offset;
+ }
+
+ /* Calculate how much data needs to be read and written. If the data
+ * segment's length is bigger than DHT_REBALANCE_BLKSIZE, read and
+ * write DHT_REBALANCE_BLKSIZE data length and the rest in the
+ * next iteration(s) */
+ read_size = ((data_block_size > DHT_REBALANCE_BLKSIZE)
+ ? DHT_REBALANCE_BLKSIZE
+ : data_block_size);
+
+ /* Calculate the remaining size of the data block - maybe there's no
+ * need to seek for data in the next iteration */
+ data_block_size -= read_size;
+
+ /* Set offset to the offset of the data segment so read and write
+ * will have the correct position */
+ offset = data_offset;
+ }
ret = syncop_readv(from, src, read_size, offset, 0, &vector, &count,
&iobref, NULL, NULL, NULL);
+
if (!ret || (ret < 0)) {
if (!ret) {
/* File was probably truncated*/
@@ -1095,50 +1093,42 @@ __dht_rebalance_migrate_data(xlator_t *this, gf_defrag_info_t *defrag,
break;
}
- if (hole_exists) {
- ret = dht_write_with_holes(to, dst, vector, count, ret, offset,
- iobref, fop_errno);
- } else {
- if (!conf->force_migration && !dht_is_tier_xlator(this)) {
+ if (!conf->force_migration) {
+ if (!xdata) {
+ xdata = dict_new();
if (!xdata) {
- xdata = dict_new();
- if (!xdata) {
- gf_msg("dht", GF_LOG_ERROR, 0,
- DHT_MSG_MIGRATE_FILE_FAILED,
- "insufficient memory");
- ret = -1;
- *fop_errno = ENOMEM;
- break;
- }
+ gf_msg("dht", GF_LOG_ERROR, 0, DHT_MSG_MIGRATE_FILE_FAILED,
+ "insufficient memory");
+ ret = -1;
+ *fop_errno = ENOMEM;
+ break;
+ }
- /* Fail this write and abort rebalance if we
- * detect a write from client since migration of
- * this file started. This is done to avoid
- * potential data corruption due to out of order
- * writes from rebalance and client to the same
- * region (as compared between src and dst
- * files). See
- * https://github.com/gluster/glusterfs/issues/308
- * for more details.
- */
- ret = dict_set_int32(xdata, GF_AVOID_OVERWRITE, 1);
- if (ret) {
- gf_msg("dht", GF_LOG_ERROR, 0, ENOMEM,
- "failed to set dict");
- ret = -1;
- *fop_errno = ENOMEM;
- break;
- }
+ /* Fail this write and abort rebalance if we
+ * detect a write from client since migration of
+ * this file started. This is done to avoid
+ * potential data corruption due to out of order
+ * writes from rebalance and client to the same
+ * region (as compared between src and dst
+ * files). See
+ * https://github.com/gluster/glusterfs/issues/308
+ * for more details.
+ */
+ ret = dict_set_int32_sizen(xdata, GF_AVOID_OVERWRITE, 1);
+ if (ret) {
+ gf_msg("dht", GF_LOG_ERROR, 0, ENOMEM,
+ "failed to set dict");
+ ret = -1;
+ *fop_errno = ENOMEM;
+ break;
}
}
- ret = syncop_writev(to, dst, vector, count, offset, iobref, 0, NULL,
- NULL, xdata, NULL);
- if (ret < 0) {
- *fop_errno = -ret;
- }
}
+ ret = syncop_writev(to, dst, vector, count, offset, iobref, 0, NULL,
+ NULL, xdata, NULL);
if (ret < 0) {
+ *fop_errno = -ret;
break;
}
@@ -1543,21 +1533,6 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
goto out;
}
- /* If defrag is NULL, it should be assumed that migration is triggered
- * from client using the trusted.distribute.migrate-data virtual xattr
- */
- defrag = conf->defrag;
-
- /* migration of files from clients is restricted to non-tiered clients
- * for now */
- if (!defrag && dht_is_tier_xlator(this)) {
- ret = ENOTSUP;
- goto out;
- }
-
- if (defrag && defrag->tier_conf.is_tier)
- log_level = GF_LOG_TRACE;
-
gf_log(this->name, log_level, "%s: attempting to move from %s to %s",
loc->path, from->name, to->name);
@@ -1704,9 +1679,13 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
goto out;
}
+ /* Try to preserve 'holes' while migrating data */
+ if (stbuf.ia_size > (stbuf.ia_blocks * GF_DISK_SECTOR_SIZE))
+ file_has_holes = 1;
+
/* create the destination, with required modes/xattr */
ret = __dht_rebalance_create_dst_file(this, to, from, loc, &stbuf, &dst_fd,
- fop_errno);
+ fop_errno, file_has_holes);
if (ret) {
gf_msg(this->name, GF_LOG_ERROR, 0, 0,
"Create dst failed"
@@ -1750,8 +1729,8 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
* destination. We need to do update this only post migration
* as in case of failure the linkto needs to point to the source
* subvol */
- ret = __dht_rebalance_create_dst_file(this, to, from, loc, &stbuf,
- &dst_fd, fop_errno);
+ ret = __dht_rebalance_create_dst_file(
+ this, to, from, loc, &stbuf, &dst_fd, fop_errno, file_has_holes);
if (ret) {
gf_log(this->name, GF_LOG_ERROR,
"Create dst failed"
@@ -1838,9 +1817,6 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
ret = 0;
goto out;
}
- /* Try to preserve 'holes' while migrating data */
- if (stbuf.ia_size > (stbuf.ia_blocks * GF_DISK_SECTOR_SIZE))
- file_has_holes = 1;
ret = __dht_rebalance_migrate_data(this, defrag, from, to, src_fd, dst_fd,
stbuf.ia_size, file_has_holes,
@@ -2306,14 +2282,12 @@ out:
}
}
- if (!dht_is_tier_xlator(this)) {
- lk_ret = syncop_removexattr(to, loc, GF_PROTECT_FROM_EXTERNAL_WRITES,
- NULL, NULL);
- if (lk_ret && (lk_ret != -ENODATA) && (lk_ret != -ENOATTR)) {
- gf_msg(this->name, GF_LOG_WARNING, -lk_ret, 0,
- "%s: removexattr failed key %s", loc->path,
- GF_PROTECT_FROM_EXTERNAL_WRITES);
- }
+ lk_ret = syncop_removexattr(to, loc, GF_PROTECT_FROM_EXTERNAL_WRITES, NULL,
+ NULL);
+ if (lk_ret && (lk_ret != -ENODATA) && (lk_ret != -ENOATTR)) {
+ gf_msg(this->name, GF_LOG_WARNING, -lk_ret, 0,
+ "%s: removexattr failed key %s", loc->path,
+ GF_PROTECT_FROM_EXTERNAL_WRITES);
}
if (dict)
@@ -2564,10 +2538,10 @@ out:
* all hardlinks.
*/
-int
+gf_boolean_t
gf_defrag_should_i_migrate(xlator_t *this, int local_subvol_index, uuid_t gfid)
{
- int ret = 0;
+ gf_boolean_t ret = _gf_false;
int i = local_subvol_index;
char *str = NULL;
uint32_t hashval = 0;
@@ -2589,12 +2563,11 @@ gf_defrag_should_i_migrate(xlator_t *this, int local_subvol_index, uuid_t gfid)
}
str = uuid_utoa_r(gfid, buf);
- ret = dht_hash_compute(this, 0, str, &hashval);
- if (ret == 0) {
+ if (dht_hash_compute(this, 0, str, &hashval) == 0) {
index = (hashval % entry->count);
if (entry->elements[index].info == REBAL_NODEUUID_MINE) {
/* Index matches this node's nodeuuid.*/
- ret = 1;
+ ret = _gf_true;
goto out;
}
@@ -2607,12 +2580,12 @@ gf_defrag_should_i_migrate(xlator_t *this, int local_subvol_index, uuid_t gfid)
/* None of the bricks in the subvol are up.
* CHILD_DOWN will kill the process soon */
- return 0;
+ return _gf_false;
}
if (entry->elements[index].info == REBAL_NODEUUID_MINE) {
/* Index matches this node's nodeuuid.*/
- ret = 1;
+ ret = _gf_true;
goto out;
}
}
@@ -2661,6 +2634,7 @@ gf_defrag_migrate_single_file(void *opaque)
struct iatt *iatt_ptr = NULL;
gf_boolean_t update_skippedcount = _gf_true;
int i = 0;
+ gf_boolean_t should_i_migrate = 0;
rebal_entry = (struct dht_container *)opaque;
if (!rebal_entry) {
@@ -2715,11 +2689,29 @@ gf_defrag_migrate_single_file(void *opaque)
goto out;
}
+ should_i_migrate = gf_defrag_should_i_migrate(
+ this, rebal_entry->local_subvol_index, entry->d_stat.ia_gfid);
+
gf_uuid_copy(entry_loc.gfid, entry->d_stat.ia_gfid);
gf_uuid_copy(entry_loc.pargfid, loc->gfid);
ret = syncop_lookup(this, &entry_loc, &iatt, NULL, NULL, NULL);
+
+ if (!should_i_migrate) {
+ /* this node isn't supposed to migrate the file. suppressing any
+ * potential error from lookup as this file is under migration by
+ * another node */
+ if (ret) {
+ gf_msg_debug(this->name, -ret,
+ "Ignoring lookup failure: node isn't migrating %s",
+ entry_loc.path);
+ ret = 0;
+ }
+ gf_msg_debug(this->name, 0, "Don't migrate %s ", entry_loc.path);
+ goto out;
+ }
+
if (ret) {
gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_MIGRATE_FILE_FAILED,
"Migrate file failed: %s lookup failed", entry_loc.path);
@@ -2740,12 +2732,6 @@ gf_defrag_migrate_single_file(void *opaque)
goto out;
}
- if (!gf_defrag_should_i_migrate(this, rebal_entry->local_subvol_index,
- entry->d_stat.ia_gfid)) {
- gf_msg_debug(this->name, 0, "Don't migrate %s ", entry_loc.path);
- goto out;
- }
-
iatt_ptr = &iatt;
hashed_subvol = dht_subvol_get_hashed(this, &entry_loc);
@@ -2888,8 +2874,7 @@ gf_defrag_migrate_single_file(void *opaque)
if (defrag->stats == _gf_true) {
gettimeofday(&end, NULL);
- elapsed = (end.tv_sec - start.tv_sec) * 1e6 +
- (end.tv_usec - start.tv_usec);
+ elapsed = gf_tvdiff(&start, &end);
gf_log(this->name, GF_LOG_INFO,
"Migration of "
"file:%s size:%" PRIu64
@@ -3068,9 +3053,9 @@ int static gf_defrag_get_entry(xlator_t *this, int i,
dht_conf_t *conf, gf_defrag_info_t *defrag,
fd_t *fd, dict_t *migrate_data,
struct dir_dfmeta *dir_dfmeta, dict_t *xattr_req,
- int *should_commit_hash, int *perrno)
+ int *perrno)
{
- int ret = -1;
+ int ret = 0;
char is_linkfile = 0;
gf_dirent_t *df_entry = NULL;
struct dht_container *tmp_container = NULL;
@@ -3086,6 +3071,13 @@ int static gf_defrag_get_entry(xlator_t *this, int i,
}
if (dir_dfmeta->fetch_entries[i] == 1) {
+ if (!fd) {
+ dir_dfmeta->fetch_entries[i] = 0;
+ dir_dfmeta->offset_var[i].readdir_done = 1;
+ ret = 0;
+ goto out;
+ }
+
ret = syncop_readdirp(conf->local_subvols[i], fd, 131072,
dir_dfmeta->offset_var[i].offset,
&(dir_dfmeta->equeue[i]), xattr_req, NULL);
@@ -3245,7 +3237,6 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
dict_t *migrate_data, int *perrno)
{
int ret = -1;
- fd_t *fd = NULL;
dht_conf_t *conf = NULL;
gf_dirent_t entries;
dict_t *xattr_req = NULL;
@@ -3266,7 +3257,7 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
int dfc_index = 0;
int throttle_up = 0;
struct dir_dfmeta *dir_dfmeta = NULL;
- int should_commit_hash = 1;
+ xlator_t *old_THIS = NULL;
gf_log(this->name, GF_LOG_INFO, "migrate data called on %s", loc->path);
gettimeofday(&dir_start, NULL);
@@ -3279,28 +3270,53 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
goto out;
}
- fd = fd_create(loc->inode, defrag->pid);
- if (!fd) {
- gf_log(this->name, GF_LOG_ERROR, "Failed to create fd");
+ old_THIS = THIS;
+ THIS = this;
+
+ dir_dfmeta = GF_CALLOC(1, sizeof(*dir_dfmeta), gf_common_mt_pointer);
+ if (!dir_dfmeta) {
+ gf_log(this->name, GF_LOG_ERROR, "dir_dfmeta is NULL");
ret = -1;
goto out;
}
- ret = syncop_opendir(this, loc, fd, NULL, NULL);
- if (ret) {
- gf_msg(this->name, GF_LOG_WARNING, -ret, DHT_MSG_MIGRATE_DATA_FAILED,
- "Migrate data failed: Failed to open dir %s", loc->path);
- *perrno = -ret;
+ dir_dfmeta->lfd = GF_CALLOC(local_subvols_cnt, sizeof(fd_t *),
+ gf_common_mt_pointer);
+ if (!dir_dfmeta->lfd) {
+ gf_smsg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_INSUFF_MEMORY,
+ "for dir_dfmeta", NULL);
ret = -1;
+ *perrno = ENOMEM;
goto out;
}
- fd_bind(fd);
- dir_dfmeta = GF_CALLOC(1, sizeof(*dir_dfmeta), gf_common_mt_pointer);
- if (!dir_dfmeta) {
- gf_log(this->name, GF_LOG_ERROR, "dir_dfmeta is NULL");
- ret = -1;
- goto out;
+ for (i = 0; i < local_subvols_cnt; i++) {
+ dir_dfmeta->lfd[i] = fd_create(loc->inode, defrag->pid);
+ if (!dir_dfmeta->lfd[i]) {
+ gf_smsg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_FD_CREATE_FAILED,
+ NULL);
+ *perrno = ENOMEM;
+ ret = -1;
+ goto out;
+ }
+
+ ret = syncop_opendir(conf->local_subvols[i], loc, dir_dfmeta->lfd[i],
+ NULL, NULL);
+ if (ret) {
+ fd_unref(dir_dfmeta->lfd[i]);
+ dir_dfmeta->lfd[i] = NULL;
+ gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_FAILED_TO_OPEN,
+ "dir: %s", loc->path, "subvol: %s",
+ conf->local_subvols[i]->name, NULL);
+
+ if (conf->decommission_in_progress) {
+ *perrno = -ret;
+ ret = -1;
+ goto out;
+ }
+ } else {
+ fd_bind(dir_dfmeta->lfd[i]);
+ }
}
dir_dfmeta->head = GF_CALLOC(local_subvols_cnt, sizeof(*(dir_dfmeta->head)),
@@ -3335,6 +3351,7 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
ret = -1;
goto out;
}
+
ret = gf_defrag_ctx_subvols_init(dir_dfmeta->offset_var, this);
if (ret) {
gf_log(this->name, GF_LOG_ERROR,
@@ -3347,7 +3364,8 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
dir_dfmeta->fetch_entries = GF_CALLOC(local_subvols_cnt, sizeof(int),
gf_common_mt_int);
if (!dir_dfmeta->fetch_entries) {
- gf_log(this->name, GF_LOG_ERROR, "dir_dfmeta->fetch_entries is NULL");
+ gf_smsg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_INSUFF_MEMORY,
+ "for dir_dfmeta->fetch_entries", NULL);
ret = -1;
goto out;
}
@@ -3417,8 +3435,9 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
ldfq_count <= MAX_MIGRATE_QUEUE_COUNT &&
!dht_dfreaddirp_done(dir_dfmeta->offset_var, local_subvols_cnt)) {
ret = gf_defrag_get_entry(this, dfc_index, &container, loc, conf,
- defrag, fd, migrate_data, dir_dfmeta,
- xattr_req, &should_commit_hash, perrno);
+ defrag, dir_dfmeta->lfd[dfc_index],
+ migrate_data, dir_dfmeta, xattr_req,
+ perrno);
if (defrag->defrag_status == GF_DEFRAG_STATUS_STOPPED) {
goto out;
@@ -3462,27 +3481,19 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
}
gettimeofday(&end, NULL);
- elapsed = (end.tv_sec - dir_start.tv_sec) * 1e6 +
- (end.tv_usec - dir_start.tv_usec);
+ elapsed = gf_tvdiff(&dir_start, &end);
gf_log(this->name, GF_LOG_INFO,
"Migration operation on dir %s took "
"%.2f secs",
loc->path, elapsed / 1e6);
ret = 0;
out:
-
+ THIS = old_THIS;
gf_defrag_free_dir_dfmeta(dir_dfmeta, local_subvols_cnt);
if (xattr_req)
dict_unref(xattr_req);
- if (fd)
- fd_unref(fd);
-
- if (ret == 0 && should_commit_hash == 0) {
- ret = 2;
- }
-
/* It does not matter if it errored out - this number is
* used to calculate rebalance estimated time to complete.
* No locking required as dirs are processed by a single thread.
@@ -3490,6 +3501,7 @@ out:
defrag->num_dirs_processed++;
return ret;
}
+
int
gf_defrag_settle_hash(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
dict_t *fix_layout)
@@ -3504,7 +3516,6 @@ gf_defrag_settle_hash(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
* rebalance is complete.
*/
if (defrag->cmd == GF_DEFRAG_CMD_START_LAYOUT_FIX ||
- defrag->cmd == GF_DEFRAG_CMD_START_DETACH_TIER ||
defrag->cmd == GF_DEFRAG_CMD_DETACH_START) {
return 0;
}
@@ -3550,114 +3561,6 @@ gf_defrag_settle_hash(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
return 0;
}
-/* Function for doing a named lookup on file inodes during an attach tier
- * So that a hardlink lookup heal i.e gfid to parent gfid lookup heal
- * happens on pre-existing data. This is required so that the ctr database has
- * hardlinks of all the exisitng file in the volume. CTR xlator on the
- * brick/server side does db update/insert of the hardlink on a namelookup.
- * Currently the namedlookup is done synchronous to the fixlayout that is
- * triggered by attach tier. This is not performant, adding more time to
- * fixlayout. The performant approach is record the hardlinks on a compressed
- * datastore and then do the namelookup asynchronously later, giving the ctr db
- * eventual consistency
- * */
-int
-gf_fix_layout_tier_attach_lookup(xlator_t *this, loc_t *parent_loc,
- gf_dirent_t *file_dentry)
-{
- int ret = -1;
- dict_t *lookup_xdata = NULL;
- dht_conf_t *conf = NULL;
- loc_t file_loc = {
- 0,
- };
- struct iatt iatt = {
- 0,
- };
-
- GF_VALIDATE_OR_GOTO("tier", this, out);
-
- GF_VALIDATE_OR_GOTO(this->name, parent_loc, out);
-
- GF_VALIDATE_OR_GOTO(this->name, file_dentry, out);
-
- GF_VALIDATE_OR_GOTO(this->name, this->private, out);
-
- if (!parent_loc->inode) {
- gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
- "%s/%s parent is NULL", parent_loc->path, file_dentry->d_name);
- goto out;
- }
-
- conf = this->private;
-
- loc_wipe(&file_loc);
-
- if (gf_uuid_is_null(file_dentry->d_stat.ia_gfid)) {
- gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
- "%s/%s gfid not present", parent_loc->path, file_dentry->d_name);
- goto out;
- }
-
- gf_uuid_copy(file_loc.gfid, file_dentry->d_stat.ia_gfid);
-
- if (gf_uuid_is_null(parent_loc->gfid)) {
- gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
- "%s/%s"
- " gfid not present",
- parent_loc->path, file_dentry->d_name);
- goto out;
- }
-
- gf_uuid_copy(file_loc.pargfid, parent_loc->gfid);
-
- ret = dht_build_child_loc(this, &file_loc, parent_loc, file_dentry->d_name);
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
- "Child loc build failed");
- ret = -1;
- goto out;
- }
-
- lookup_xdata = dict_new();
- if (!lookup_xdata) {
- gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
- "Failed creating lookup dict for %s", file_dentry->d_name);
- goto out;
- }
-
- ret = dict_set_int32(lookup_xdata, CTR_ATTACH_TIER_LOOKUP, 1);
- if (ret) {
- gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
- "Failed to set lookup flag");
- goto out;
- }
-
- gf_uuid_copy(file_loc.parent->gfid, parent_loc->gfid);
-
- /* Sending lookup to cold tier only */
- ret = syncop_lookup(conf->subvolumes[0], &file_loc, &iatt, NULL,
- lookup_xdata, NULL);
- if (ret) {
- /* If the file does not exist on the cold tier than it must */
- /* have been discovered on the hot tier. This is not an error. */
- gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
- "%s lookup to cold tier on attach heal failed", file_loc.path);
- goto out;
- }
-
- ret = 0;
-
-out:
-
- loc_wipe(&file_loc);
-
- if (lookup_xdata)
- dict_unref(lookup_xdata);
-
- return ret;
-}
-
int
gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
dict_t *fix_layout, dict_t *migrate_data)
@@ -3677,7 +3580,6 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
};
inode_t *linked_inode = NULL, *inode = NULL;
dht_conf_t *conf = NULL;
- int should_commit_hash = 1;
int perrno = 0;
conf = this->private;
@@ -3780,16 +3682,6 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
if (!strcmp(entry->d_name, ".") || !strcmp(entry->d_name, ".."))
continue;
if (!IA_ISDIR(entry->d_stat.ia_type)) {
- /* If its a fix layout during the attach
- * tier operation do lookups on files
- * on cold subvolume so that there is a
- * CTR DB Lookup Heal triggered on existing
- * data.
- * */
- if (defrag->cmd == GF_DEFRAG_CMD_START_TIER) {
- gf_fix_layout_tier_attach_lookup(this, loc, entry);
- }
-
continue;
}
loc_wipe(&entry_loc);
@@ -3806,8 +3698,6 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
goto out;
} else {
- should_commit_hash = 0;
-
continue;
}
}
@@ -3870,7 +3760,6 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
ret = -1;
goto out;
} else {
- should_commit_hash = 0;
continue;
}
}
@@ -3883,11 +3772,12 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
ret = gf_defrag_fix_layout(this, defrag, &entry_loc, fix_layout,
migrate_data);
- if (defrag->defrag_status == GF_DEFRAG_STATUS_STOPPED) {
+ if (defrag->defrag_status == GF_DEFRAG_STATUS_STOPPED ||
+ defrag->defrag_status == GF_DEFRAG_STATUS_FAILED) {
goto out;
}
- if (ret && ret != 2) {
+ if (ret) {
gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LAYOUT_FIX_FAILED,
"Fix layout failed for %s", entry_loc.path);
@@ -3918,6 +3808,17 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
*/
ret = syncop_setxattr(this, loc, fix_layout, 0, NULL, NULL);
+
+ /* In case of a race where the directory is deleted just before
+ * layout setxattr, the errors are updated in the layout structure.
+ * We can use this information to make a decision whether the directory
+ * is deleted entirely.
+ */
+ if (ret == 0) {
+ ret = dht_dir_layout_error_check(this, loc->inode);
+ ret = -ret;
+ }
+
if (ret) {
if (-ret == ENOENT || -ret == ESTALE) {
gf_msg(this->name, GF_LOG_INFO, -ret, DHT_MSG_LAYOUT_FIX_FAILED,
@@ -3943,11 +3844,10 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
}
}
- if ((defrag->cmd != GF_DEFRAG_CMD_START_TIER) &&
- (defrag->cmd != GF_DEFRAG_CMD_START_LAYOUT_FIX)) {
+ if (defrag->cmd != GF_DEFRAG_CMD_START_LAYOUT_FIX) {
ret = gf_defrag_process_dir(this, defrag, loc, migrate_data, &perrno);
- if (ret && (ret != 2)) {
+ if (ret) {
if (perrno == ENOENT || perrno == ESTALE) {
ret = 0;
goto out;
@@ -3963,18 +3863,13 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
if (conf->decommission_in_progress) {
goto out;
}
-
- should_commit_hash = 0;
}
- } else if (ret == 2) {
- should_commit_hash = 0;
}
}
gf_msg_trace(this->name, 0, "fix layout called on %s", loc->path);
- if (should_commit_hash &&
- gf_defrag_settle_hash(this, defrag, loc, fix_layout) != 0) {
+ if (gf_defrag_settle_hash(this, defrag, loc, fix_layout) != 0) {
defrag->total_failures++;
gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_SETTLE_HASH_FAILED,
@@ -3998,10 +3893,6 @@ out:
if (fd)
fd_unref(fd);
- if (ret == 0 && should_commit_hash == 0) {
- ret = 2;
- }
-
return ret;
}
@@ -4010,31 +3901,26 @@ dht_init_local_subvols_and_nodeuuids(xlator_t *this, dht_conf_t *conf,
loc_t *loc)
{
dict_t *dict = NULL;
- gf_defrag_info_t *defrag = NULL;
uuid_t *uuid_ptr = NULL;
int ret = -1;
int i = 0;
int j = 0;
- defrag = conf->defrag;
-
- if (defrag->cmd != GF_DEFRAG_CMD_START_TIER) {
- /* Find local subvolumes */
- ret = syncop_getxattr(this, loc, &dict, GF_REBAL_FIND_LOCAL_SUBVOL,
- NULL, NULL);
- if (ret && (ret != -ENODATA)) {
- gf_msg(this->name, GF_LOG_ERROR, -ret, 0,
- "local "
- "subvolume determination failed with error: %d",
- -ret);
- ret = -1;
- goto out;
- }
-
- if (!ret)
- goto out;
+ /* Find local subvolumes */
+ ret = syncop_getxattr(this, loc, &dict, GF_REBAL_FIND_LOCAL_SUBVOL, NULL,
+ NULL);
+ if (ret && (ret != -ENODATA)) {
+ gf_msg(this->name, GF_LOG_ERROR, -ret, 0,
+ "local "
+ "subvolume determination failed with error: %d",
+ -ret);
+ ret = -1;
+ goto out;
}
+ if (!ret)
+ goto out;
+
ret = syncop_getxattr(this, loc, &dict, GF_REBAL_OLD_FIND_LOCAL_SUBVOL,
NULL, NULL);
if (ret) {
@@ -4125,9 +4011,6 @@ dht_file_counter_thread(void *args)
struct timespec time_to_wait = {
0,
};
- struct timeval now = {
- 0,
- };
uint64_t tmp_size = 0;
if (!args)
@@ -4137,9 +4020,8 @@ dht_file_counter_thread(void *args)
dht_build_root_loc(defrag->root_inode, &root_loc);
while (defrag->defrag_status == GF_DEFRAG_STATUS_STARTED) {
- gettimeofday(&now, NULL);
- time_to_wait.tv_sec = now.tv_sec + 600;
- time_to_wait.tv_nsec = 0;
+ timespec_now(&time_to_wait);
+ time_to_wait.tv_sec += 600;
pthread_mutex_lock(&defrag->fc_mutex);
pthread_cond_timedwait(&defrag->fc_wakeup_cond, &defrag->fc_mutex,
@@ -4212,7 +4094,7 @@ gf_defrag_estimates_init(xlator_t *this, loc_t *loc, pthread_t *filecnt_thread)
goto out;
}
- ret = gf_thread_create(filecnt_thread, NULL, &dht_file_counter_thread,
+ ret = gf_thread_create(filecnt_thread, NULL, dht_file_counter_thread,
(void *)defrag, "dhtfcnt");
if (ret) {
@@ -4269,7 +4151,7 @@ gf_defrag_parallel_migration_init(xlator_t *this, gf_defrag_info_t *defrag,
/*Spawn Threads Here*/
while (index < thread_spawn_count) {
- ret = gf_thread_create(&(tid[index]), NULL, &gf_defrag_task,
+ ret = gf_thread_create(&(tid[index]), NULL, gf_defrag_task,
(void *)defrag, "dhtmig%d", (index + 1) & 0x3ff);
if (ret != 0) {
gf_msg("DHT", GF_LOG_ERROR, ret, 0, "Thread[%d] creation failed. ",
@@ -4376,7 +4258,8 @@ gf_defrag_start_crawl(void *data)
if (!defrag)
goto exit;
- gettimeofday(&defrag->start_time, NULL);
+ defrag->start_time = gf_time();
+
dht_build_root_inode(this, &defrag->root_inode);
if (!defrag->root_inode)
goto out;
@@ -4511,14 +4394,13 @@ gf_defrag_start_crawl(void *data)
}
ret = gf_defrag_fix_layout(this, defrag, &loc, fix_layout, migrate_data);
- if (ret && ret != 2) {
+ if (ret) {
defrag->total_failures++;
ret = -1;
goto out;
}
- if (ret != 2 &&
- gf_defrag_settle_hash(this, defrag, &loc, fix_layout) != 0) {
+ if (gf_defrag_settle_hash(this, defrag, &loc, fix_layout) != 0) {
defrag->total_failures++;
ret = -1;
goto out;
@@ -4633,9 +4515,6 @@ gf_defrag_get_estimates_based_on_size(dht_conf_t *conf)
uint64_t total_processed = 0;
uint64_t tmp_count = 0;
uint64_t time_to_complete = 0;
- struct timeval now = {
- 0,
- };
double elapsed = 0;
defrag = conf->defrag;
@@ -4643,8 +4522,7 @@ gf_defrag_get_estimates_based_on_size(dht_conf_t *conf)
if (!g_totalsize)
goto out;
- gettimeofday(&now, NULL);
- elapsed = now.tv_sec - defrag->start_time.tv_sec;
+ elapsed = gf_time() - defrag->start_time;
/* Don't calculate the estimates for the first 10 minutes.
* It is unlikely to be accurate and estimates are not required
@@ -4694,13 +4572,8 @@ gf_defrag_status_get(dht_conf_t *conf, dict_t *dict)
uint64_t lookup = 0;
uint64_t failures = 0;
uint64_t skipped = 0;
- uint64_t promoted = 0;
- uint64_t demoted = 0;
char *status = "";
double elapsed = 0;
- struct timeval end = {
- 0,
- };
uint64_t time_to_complete = 0;
uint64_t time_left = 0;
gf_defrag_info_t *defrag = conf->defrag;
@@ -4717,17 +4590,12 @@ gf_defrag_status_get(dht_conf_t *conf, dict_t *dict)
lookup = defrag->num_files_lookedup;
failures = defrag->total_failures;
skipped = defrag->skipped;
- promoted = defrag->total_files_promoted;
- demoted = defrag->total_files_demoted;
-
- gettimeofday(&end, NULL);
- elapsed = end.tv_sec - defrag->start_time.tv_sec;
+ elapsed = gf_time() - defrag->start_time;
/* The rebalance is still in progress */
- if ((defrag->cmd != GF_DEFRAG_CMD_START_TIER) &&
- (defrag->defrag_status == GF_DEFRAG_STATUS_STARTED)) {
+ if (defrag->defrag_status == GF_DEFRAG_STATUS_STARTED) {
time_to_complete = gf_defrag_get_estimates_based_on_size(conf);
if (time_to_complete && (time_to_complete > elapsed))
@@ -4742,14 +4610,6 @@ gf_defrag_status_get(dht_conf_t *conf, dict_t *dict)
if (!dict)
goto log;
- ret = dict_set_uint64(dict, "promoted", promoted);
- if (ret)
- gf_log(THIS->name, GF_LOG_WARNING, "failed to set promoted count");
-
- ret = dict_set_uint64(dict, "demoted", demoted);
- if (ret)
- gf_log(THIS->name, GF_LOG