summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorPranith Kumar K <pkarampu@redhat.com>2013-12-15 08:05:04 +0530
committerPranith Kumar Karampuri <pkarampu@redhat.com>2014-10-15 01:22:02 -0700
commit7e8eefca2caaaa61e1b31b747384f660c595d9c9 (patch)
treea6e862f2caad4097a41dd8b260a2333e90db3b4a
parent7808d4b7892ccd80bcabb1253ee359a64c51c612 (diff)
heal: glfs-heal implementation
Thanks a lot to Niels for helping me to get build stuff right. Change-Id: I634f24d90cd856ceab3cc0c6e9a91003f443403e BUG: 1147462 Signed-off-by: Pranith Kumar K <pkarampu@redhat.com> Reviewed-on: http://review.gluster.org/6529 Reviewed-by: Krutika Dhananjay <kdhananj@redhat.com> Tested-by: Gluster Build System <jenkins@build.gluster.com>
-rw-r--r--Makefile.am3
-rw-r--r--cli/src/cli-cmd-volume.c33
-rw-r--r--configure.ac2
-rw-r--r--glusterfs.spec.in1
-rw-r--r--heal/Makefile.am3
-rw-r--r--heal/src/Makefile.am35
-rw-r--r--heal/src/glfs-heal.c610
-rw-r--r--tests/basic/afr/self-heald.t161
-rw-r--r--tests/basic/self-heald.t44
-rw-r--r--tests/bugs/bug-880898.t4
-rw-r--r--xlators/cluster/afr/src/afr-self-heal-common.c21
-rw-r--r--xlators/cluster/afr/src/afr-self-heal.h7
-rw-r--r--xlators/cluster/afr/src/afr-self-heald.c2
-rw-r--r--xlators/cluster/afr/src/afr-self-heald.h6
14 files changed, 870 insertions, 62 deletions
diff --git a/Makefile.am b/Makefile.am
index 7bb4b96da2f..187bfd5ed16 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -10,9 +10,8 @@ EXTRA_DIST = autogen.sh \
$(shell find $(top_srcdir)/tests -type f -print)
SUBDIRS = $(ARGP_STANDALONE_DIR) libglusterfs rpc api xlators glusterfsd \
- $(FUSERMOUNT_SUBDIR) doc extras cli @SYNCDAEMON_SUBDIR@ \
+ $(FUSERMOUNT_SUBDIR) doc extras cli heal @SYNCDAEMON_SUBDIR@ \
@UMOUNTD_SUBDIR@
-
pkgconfigdir = @pkgconfigdir@
pkgconfig_DATA = glusterfs-api.pc libgfchangelog.pc
diff --git a/cli/src/cli-cmd-volume.c b/cli/src/cli-cmd-volume.c
index 15f448b5ba2..a31b82011e0 100644
--- a/cli/src/cli-cmd-volume.c
+++ b/cli/src/cli-cmd-volume.c
@@ -16,6 +16,7 @@
#include <sys/socket.h>
#include <netdb.h>
#include <sys/types.h>
+#include <sys/wait.h>
#include <netinet/in.h>
#ifndef _CONFIG_H
@@ -1903,6 +1904,10 @@ cli_cmd_volume_heal_cbk (struct cli_state *state, struct cli_cmd_word *word,
dict_t *options = NULL;
xlator_t *this = NULL;
cli_local_t *local = NULL;
+ int heal_op = 0;
+ runner_t runner = {0};
+ char buff[PATH_MAX] = {0};
+ char *out = NULL;
this = THIS;
frame = create_frame (this, this->ctx->pool);
@@ -1921,13 +1926,33 @@ cli_cmd_volume_heal_cbk (struct cli_state *state, struct cli_cmd_word *word,
parse_error = 1;
goto out;
}
+ ret = dict_get_int32 (options, "heal-op", &heal_op);
+ if (ret < 0)
+ goto out;
- proc = &cli_rpc_prog->proctable[GLUSTER_CLI_HEAL_VOLUME];
+ if (heal_op == GF_AFR_OP_INDEX_SUMMARY) {
+ runinit (&runner);
+ runner_add_args (&runner, SBIN_DIR"/glfsheal", words[2], NULL);
+ runner_redir (&runner, STDOUT_FILENO, RUN_PIPE);
+ ret = runner_start (&runner);
+ if (ret == -1)
+ goto out;
+ while ((out = fgets(buff, sizeof(buff),
+ runner_chio (&runner, STDOUT_FILENO)))) {
+ printf ("%s", out);
+ }
- CLI_LOCAL_INIT (local, words, frame, options);
+ ret = runner_end (&runner);
+ ret = WEXITSTATUS (ret);
+ }
+ else {
+ proc = &cli_rpc_prog->proctable[GLUSTER_CLI_HEAL_VOLUME];
- if (proc->fn) {
- ret = proc->fn (frame, THIS, options);
+ CLI_LOCAL_INIT (local, words, frame, options);
+
+ if (proc->fn) {
+ ret = proc->fn (frame, THIS, options);
+ }
}
out:
diff --git a/configure.ac b/configure.ac
index cc03b571fb1..3757c33df72 100644
--- a/configure.ac
+++ b/configure.ac
@@ -210,6 +210,8 @@ AC_CONFIG_FILES([Makefile
geo-replication/Makefile
geo-replication/src/Makefile
geo-replication/syncdaemon/Makefile
+ heal/Makefile
+ heal/src/Makefile
glusterfs.spec])
AC_CANONICAL_HOST
diff --git a/glusterfs.spec.in b/glusterfs.spec.in
index 1ba37c1eb16..2caecddce97 100644
--- a/glusterfs.spec.in
+++ b/glusterfs.spec.in
@@ -995,6 +995,7 @@ fi
%endif
# binaries
%{_sbindir}/glusterd
+%{_sbindir}/glfsheal
%{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/storage*
%{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/features/posix*
%{_libdir}/glusterfs/%{version}%{?prereltag}/xlator/protocol/server*
diff --git a/heal/Makefile.am b/heal/Makefile.am
new file mode 100644
index 00000000000..a985f42a877
--- /dev/null
+++ b/heal/Makefile.am
@@ -0,0 +1,3 @@
+SUBDIRS = src
+
+CLEANFILES =
diff --git a/heal/src/Makefile.am b/heal/src/Makefile.am
new file mode 100644
index 00000000000..65093df35cc
--- /dev/null
+++ b/heal/src/Makefile.am
@@ -0,0 +1,35 @@
+sbin_PROGRAMS = glfsheal
+
+glfsheal_SOURCES = glfs-heal.c
+
+glfsheal_LDADD = $(top_builddir)/libglusterfs/src/libglusterfs.la $(GF_LDADD)\
+ $(RLLIBS) $(top_builddir)/rpc/xdr/src/libgfxdr.la \
+ $(top_builddir)/rpc/rpc-lib/src/libgfrpc.la \
+ $(top_builddir)/api/src/libgfapi.la \
+ $(top_builddir)/xlators/cluster/afr/src/afr.la \
+ $(GF_GLUSTERFS_LIBS) $(XML_LIBS)
+
+glfsheal_LDFLAGS = $(GF_LDFLAGS)
+
+AM_CPPFLAGS = $(GF_CPPFLAGS) \
+ -I$(top_srcdir)/xlators/lib/src\
+ -I$(top_srcdir)/xlators/cluster/afr/src\
+ -I$(top_srcdir)/libglusterfs/src -I$(top_srcdir)/rpc/rpc-lib/src\
+ -I$(top_srcdir)/rpc/xdr/src\
+ -I$(top_srcdir)/api/src\
+ -I$(top_srcdir)/contrib/argp-standalone\
+ -DDATADIR=\"$(localstatedir)\" \
+ -DCONFDIR=\"$(sysconfdir)/glusterfs\" \
+ -DGSYNCD_PREFIX=\"$(libexecdir)/glusterfs\"\
+ -DSYNCDAEMON_COMPILE=$(SYNCDAEMON_COMPILE) -DSBIN_DIR=\"$(sbindir)\"\
+ $(XML_CPPFLAGS)
+
+AM_CFLAGS = -Wall $(GF_GLUSTERFS_CFLAGS)
+
+CLEANFILES =
+
+$(top_builddir)/libglusterfs/src/libglusterfs.la:
+ $(MAKE) -C $(top_builddir)/libglusterfs/src/ all
+
+$(top_builddir)/xlators/cluster/afr/src/afr.la:
+ $(MAKE) -C $(top_builddir)/xlators/cluster/afr/src/ all
diff --git a/heal/src/glfs-heal.c b/heal/src/glfs-heal.c
new file mode 100644
index 00000000000..d02858642ab
--- /dev/null
+++ b/heal/src/glfs-heal.c
@@ -0,0 +1,610 @@
+/*
+ Copyright (c) 2014 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.
+*/
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <errno.h>
+#include "glfs.h"
+#include "glfs-handles.h"
+#include "glfs-internal.h"
+#include "syncop.h"
+#include <string.h>
+#include <time.h>
+#include "afr.h"
+#include "afr-self-heal.h"
+#include "afr-self-heald.h"
+
+#define DEFAULT_HEAL_LOG_FILE_DIRECTORY DATADIR "/log/glusterfs"
+
+int
+glfsh_link_inode_update_loc (loc_t *loc, struct iatt *iattr)
+{
+ inode_t *link_inode = NULL;
+ int ret = -1;
+
+ link_inode = inode_link (loc->inode, NULL, NULL, iattr);
+ if (link_inode == NULL)
+ goto out;
+
+ inode_unref (loc->inode);
+ loc->inode = link_inode;
+ ret = 0;
+out:
+ return ret;
+}
+
+int
+glfsh_get_index_dir_loc (loc_t *rootloc, xlator_t *xl, loc_t *dirloc,
+ int32_t *op_errno)
+{
+ void *index_gfid = NULL;
+ int ret = 0;
+ dict_t *xattr = NULL;
+ struct iatt iattr = {0};
+ struct iatt parent = {0};
+
+ ret = syncop_getxattr (xl, rootloc, &xattr, GF_XATTROP_INDEX_GFID);
+ if (ret < 0) {
+ *op_errno = -ret;
+ goto out;
+ }
+
+ ret = dict_get_ptr (xattr, GF_XATTROP_INDEX_GFID, &index_gfid);
+ if (ret < 0) {
+ *op_errno = EINVAL;
+ goto out;
+ }
+
+ uuid_copy (dirloc->gfid, index_gfid);
+ dirloc->path = "";
+ dirloc->inode = inode_new (rootloc->inode->table);
+ ret = syncop_lookup (xl, dirloc, NULL,
+ &iattr, NULL, &parent);
+ dirloc->path = NULL;
+ if (ret < 0) {
+ *op_errno = -ret;
+ goto out;
+ }
+ ret = glfsh_link_inode_update_loc (dirloc, &iattr);
+ if (ret)
+ goto out;
+ glfs_loc_touchup (dirloc);
+
+ ret = 0;
+out:
+ if (xattr)
+ dict_unref (xattr);
+ return ret;
+}
+
+int
+afr_selfheal_locked_metadata_inspect (call_frame_t *frame, xlator_t *this,
+ inode_t *inode,
+ gf_boolean_t *metadata_selfheal)
+{
+ int ret = -1;
+ unsigned char *locked_on = NULL;
+ afr_private_t *priv = this->private;
+
+ locked_on = alloca0 (priv->child_count);
+
+ ret = afr_selfheal_inodelk (frame, this, inode, this->name,
+ LLONG_MAX - 1, 0, locked_on);
+ {
+ if (ret == 0) {
+ /* Not a single lock */
+ ret = -afr_final_errno (frame->local, priv);
+ if (ret == 0)
+ ret = -ENOTCONN;/* all invalid responses */
+ goto out;
+ }
+ ret = afr_selfheal_unlocked_inspect (frame, this, inode->gfid,
+ NULL, NULL,
+ metadata_selfheal, NULL);
+ }
+ afr_selfheal_uninodelk (frame, this, inode, this->name,
+ LLONG_MAX - 1, 0, locked_on);
+out:
+ return ret;
+}
+
+int
+afr_selfheal_locked_data_inspect (call_frame_t *frame, xlator_t *this,
+ inode_t *inode, gf_boolean_t *data_selfheal)
+{
+ int ret = -1;
+ afr_private_t *priv = NULL;
+ unsigned char *locked_on = NULL;
+ unsigned char *data_lock = NULL;
+
+ priv = this->private;
+ locked_on = alloca0 (priv->child_count);
+ data_lock = alloca0 (priv->child_count);
+
+ ret = afr_selfheal_tryinodelk (frame, this, inode, priv->sh_domain,
+ 0, 0, locked_on);
+ {
+ if (ret == 0) {
+ ret = -afr_final_errno (frame->local, priv);
+ if (ret == 0)
+ ret = -ENOTCONN;/* all invalid responses */
+ goto out;
+ }
+ ret = afr_selfheal_inodelk (frame, this, inode, this->name,
+ 0, 0, data_lock);
+ {
+ if (ret == 0) {
+ ret = -afr_final_errno (frame->local, priv);
+ if (ret == 0)
+ ret = -ENOTCONN;
+ /* all invalid responses */
+ goto unlock;
+ }
+ ret = afr_selfheal_unlocked_inspect (frame, this,
+ inode->gfid, NULL,
+ data_selfheal,
+ NULL, NULL);
+ }
+ afr_selfheal_uninodelk (frame, this, inode, this->name, 0, 0,
+ data_lock);
+ }
+unlock:
+ afr_selfheal_uninodelk (frame, this, inode, priv->sh_domain, 0, 0,
+ locked_on);
+out:
+ return ret;
+}
+
+int
+afr_selfheal_locked_entry_inspect (call_frame_t *frame, xlator_t *this,
+ inode_t *inode, gf_boolean_t *entry_selfheal)
+{
+ int ret = -1;
+ afr_private_t *priv = NULL;
+ unsigned char *locked_on = NULL;
+ unsigned char *data_lock = NULL;
+
+ priv = this->private;
+ locked_on = alloca0 (priv->child_count);
+ data_lock = alloca0 (priv->child_count);
+
+ ret = afr_selfheal_tryentrylk (frame, this, inode, priv->sh_domain,
+ NULL, locked_on);
+ {
+ if (ret == 0) {
+ ret = -afr_final_errno (frame->local, priv);
+ if (ret == 0)
+ ret = -ENOTCONN;/* all invalid responses */
+ goto out;
+ }
+
+ ret = afr_selfheal_entrylk (frame, this, inode, this->name,
+ NULL, data_lock);
+ {
+ if (ret == 0) {
+ ret = -afr_final_errno (frame->local, priv);
+ if (ret == 0)
+ ret = -ENOTCONN;
+ /* all invalid responses */
+ goto unlock;
+ }
+ ret = afr_selfheal_unlocked_inspect (frame, this,
+ inode->gfid,
+ NULL, NULL, NULL,
+ entry_selfheal);
+ }
+ afr_selfheal_unentrylk (frame, this, inode, this->name, NULL,
+ data_lock);
+ }
+unlock:
+ afr_selfheal_unentrylk (frame, this, inode, priv->sh_domain, NULL,
+ locked_on);
+out:
+ return ret;
+}
+
+int
+afr_selfheal_locked_inspect (call_frame_t *frame, xlator_t *this, uuid_t gfid,
+ inode_t **inode,
+ gf_boolean_t *data_selfheal,
+ gf_boolean_t *metadata_selfheal,
+ gf_boolean_t *entry_selfheal)
+{
+ int ret = -1;
+ gf_boolean_t dsh = _gf_false;
+ gf_boolean_t msh = _gf_false;
+ gf_boolean_t esh = _gf_false;
+
+ ret = afr_selfheal_unlocked_inspect (frame, this, gfid, inode,
+ &dsh, &msh, &esh);
+ if (ret)
+ goto out;
+
+ /* For every heal type hold locks and check if it indeed needs heal */
+
+ if (msh) {
+ msh = _gf_false;
+ ret = afr_selfheal_locked_metadata_inspect (frame, this,
+ *inode, &msh);
+ if (msh || ret < 0)
+ goto out;
+ }
+
+ if (dsh) {
+ dsh = _gf_false;
+ ret = afr_selfheal_locked_data_inspect (frame, this, *inode,
+ &dsh);
+ if (dsh || ret < 0)
+ goto out;
+ }
+
+ if (esh) {
+ esh = _gf_false;
+ ret = afr_selfheal_locked_entry_inspect (frame, this, *inode,
+ &esh);
+ }
+
+out:
+ if (entry_selfheal)
+ *entry_selfheal = esh;
+ if (data_selfheal)
+ *data_selfheal = dsh;
+ if (metadata_selfheal)
+ *metadata_selfheal = msh;
+ return ret;
+}
+
+static xlator_t*
+_get_afr_ancestor (xlator_t *xl)
+{
+ if (!xl || !xl->parents)
+ return NULL;
+
+ while (xl->parents) {
+ xl = xl->parents->xlator;
+ if (!xl)
+ break;
+ if (strcmp (xl->type, "cluster/replicate") == 0)
+ return xl;
+ }
+
+ return NULL;
+}
+
+static int
+glfsh_process_entries (xlator_t *xl, fd_t *fd, gf_dirent_t *entries,
+ off_t *offset, uint64_t *num_entries)
+{
+ gf_dirent_t *entry = NULL;
+ gf_dirent_t *tmp = NULL;
+ int ret = 0;
+ char *path = NULL;
+ uuid_t gfid = {0};
+ inode_t *inode = NULL;
+ call_frame_t *frame = NULL;
+ xlator_t *this = NULL;
+ gf_boolean_t data_selfheal = _gf_false;
+ gf_boolean_t metadata_selfheal = _gf_false;
+ gf_boolean_t entry_selfheal = _gf_false;
+
+ this = THIS;
+ frame = afr_frame_create (this);
+ if (!frame) {
+ ret = -1;
+ goto out;
+ }
+
+ list_for_each_entry_safe (entry, tmp, &entries->list, list) {
+ *offset = entry->d_off;
+ if ((strcmp (entry->d_name, ".") == 0) ||
+ (strcmp (entry->d_name, "..") == 0))
+ continue;
+
+ uuid_clear (gfid);
+ GF_FREE (path);
+ path = NULL;
+ if (inode) {
+ inode_forget (inode, 1);
+ inode_unref (inode);
+ inode = NULL;
+ }
+
+ AFR_STACK_RESET (frame);
+ uuid_parse (entry->d_name, gfid);
+ ret = afr_selfheal_locked_inspect (frame, this, gfid, &inode,
+ &data_selfheal,
+ &metadata_selfheal,
+ &entry_selfheal);
+ if (ret == 0) {
+ if (!entry_selfheal && !metadata_selfheal &&
+ !data_selfheal)
+ continue;
+ }
+
+ ret = afr_shd_gfid_to_path (this, xl, gfid, &path);
+
+ if (ret == -ENOENT || ret == -ESTALE) {
+ afr_shd_index_purge (xl, fd->inode, entry->d_name);
+ ret = 0;
+ continue;
+ }
+
+ (*num_entries)++;
+ printf ("%s\n", path ? path : uuid_utoa (inode->gfid));
+ }
+ ret = 0;
+out:
+ if (inode) {
+ inode_forget (inode, 1);
+ inode_unref (inode);
+ }
+
+ GF_FREE (path);
+ return ret;
+}
+
+static int
+glfsh_crawl_directory (xlator_t *readdir_xl, fd_t *fd, loc_t *loc)
+{
+ off_t offset = 0;
+ gf_dirent_t entries;
+ int ret = 0;
+ gf_boolean_t free_entries = _gf_false;
+ uint64_t num_entries = 0;
+
+ INIT_LIST_HEAD (&entries.list);
+
+ while (1) {
+ ret = syncop_readdir (readdir_xl, fd, 131072, offset, &entries);
+ if (ret <= 0)
+ break;
+ ret = 0;
+ free_entries = _gf_true;
+
+ if (list_empty (&entries.list))
+ goto out;
+
+ ret = glfsh_process_entries (readdir_xl, fd, &entries, &offset,
+ &num_entries);
+ if (ret < 0)
+ goto out;
+
+ gf_dirent_free (&entries);
+ free_entries = _gf_false;
+ }
+ ret = 0;
+out:
+ if (free_entries)
+ gf_dirent_free (&entries);
+ if (ret < 0) {
+ printf ("Failed to complete gathering info. "
+ "Number of entries so far: %"PRIu64"\n", num_entries);
+ }
+ else {
+ printf ("Number of entries: %"PRIu64"\n", num_entries);
+ }
+ return ret;
+}
+
+static int
+glfsh_print_brick (xlator_t *xl, loc_t *rootloc)
+{
+ int ret = 0;
+ dict_t *xattr = NULL;
+ char *pathinfo = NULL;
+ char *brick_start = NULL;
+ char *brick_end = NULL;
+
+ ret = syncop_getxattr (xl, rootloc, &xattr, GF_XATTR_PATHINFO_KEY);
+ if (ret < 0)
+ goto out;
+
+ ret = dict_get_str (xattr, GF_XATTR_PATHINFO_KEY, &pathinfo);
+ if (ret < 0)
+ goto out;
+
+ brick_start = strchr (pathinfo, ':') + 1;
+ brick_end = pathinfo + strlen (pathinfo) - 1;
+ *brick_end = 0;
+ printf ("Brick %s\n", brick_start);
+
+out:
+ if (xattr)
+ dict_unref (xattr);
+ return ret;
+}
+
+void
+glfsh_print_brick_from_xl (xlator_t *xl)
+{
+ char *remote_host = NULL;
+ char *remote_subvol = NULL;
+ int ret = 0;
+
+ ret = dict_get_str (xl->options, "remote-host", &remote_host);
+ if (ret < 0)
+ goto out;
+
+ ret = dict_get_str (xl->options, "remote-subvolume", &remote_subvol);
+ if (ret < 0)
+ goto out;
+out:
+ if (ret < 0)
+ printf ("Brick - Not able to get brick information\n");
+ else
+ printf ("Brick %s:%s\n", remote_host, remote_subvol);
+}
+
+void
+glfsh_print_pending_heals (xlator_t *xl, loc_t *rootloc)
+{
+ int ret = 0;
+ loc_t dirloc = {0};
+ fd_t *fd = NULL;
+ int32_t op_errno = 0;
+
+ ret = glfsh_print_brick (xl, rootloc);
+ if (ret < 0) {
+ glfsh_print_brick_from_xl (xl);
+ printf ("Status: %s\n", strerror (-ret));
+ goto out;
+ }
+
+ ret = glfsh_get_index_dir_loc (rootloc, xl, &dirloc, &op_errno);
+ if (ret < 0) {
+ if (op_errno == ESTALE || op_errno == ENOENT)
+ printf ("Number of entries: 0\n");
+ else
+ printf ("Status: %s\n", strerror (op_errno));
+ goto out;
+ }
+
+ fd = fd_anonymous (dirloc.inode);
+ ret = glfsh_crawl_directory (xl, fd, &dirloc);
+ if (fd)
+ fd_unref (fd);
+ if (ret < 0)
+ printf ("Failed to find entries with pending self-heal\n");
+out:
+ loc_wipe (&dirloc);
+ return;
+}
+
+static int
+glfsh_validate_replicate_volume (xlator_t *xl)
+{
+ xlator_t *afr_xl = NULL;
+ int ret = -1;
+
+ while (xl->next)
+ xl = xl->next;
+
+ while (xl) {
+ if (strcmp (xl->type, "protocol/client") == 0) {
+ afr_xl = _get_afr_ancestor (xl);
+ if (afr_xl) {
+ ret = 0;
+ break;
+ }
+ }
+
+ xl = xl->prev;
+ }
+
+ return ret;
+}
+
+int
+main (int argc, char **argv)
+{
+ glfs_t *fs = NULL;
+ int ret = 0;
+ char *volname = NULL;
+ xlator_t *top_subvol = NULL;
+ xlator_t *xl = NULL;
+ loc_t rootloc = {0};
+ char logfilepath[PATH_MAX] = {0};
+ xlator_t *old_THIS = NULL;
+ xlator_t *afr_xl = NULL;
+
+ if (argc != 2) {
+ printf ("Usage: %s <volname>\n", argv[0]);
+ ret = -1;
+ goto out;
+ }
+ volname = argv[1];
+
+ fs = glfs_new (volname);
+ if (!fs) {
+ ret = -1;
+ printf ("Not able to initialize volume '%s'\n", volname);
+ goto out;
+ }
+
+ ret = glfs_set_volfile_server (fs, "tcp", "localhost", 24007);
+ snprintf (logfilepath, sizeof (logfilepath),
+ DEFAULT_HEAL_LOG_FILE_DIRECTORY"/glfsheal-%s.log", volname);
+ ret = glfs_set_logging(fs, logfilepath, GF_LOG_INFO);
+ if (ret < 0) {
+ ret = -1;
+ printf ("Not able to initialize volume '%s'\n", volname);
+ goto out;
+ }
+
+ ret = glfs_init (fs);
+ if (ret < 0) {
+ ret = -1;
+ if (errno == ENOENT) {
+ printf ("Volume %s does not exist\n", volname);
+ }
+ else {
+ printf ("%s: Not able to fetch volfile from "
+ "glusterd\n", volname);
+ }
+ goto out;
+ }
+
+ sleep (2);
+ __glfs_entry_fs (fs);
+ top_subvol = glfs_active_subvol (fs);
+ if (!top_subvol) {
+ ret = -1;
+ if (errno == ENOTCONN) {
+ printf ("Volume %s is not started (Or) All the bricks "
+ "are not running.\n", volname);
+ }
+ else {
+ printf ("%s: Not able to mount the volume, %s\n",
+ volname, strerror (errno));
+ }
+ goto out;
+ }
+
+ ret = glfsh_validate_replicate_volume (top_subvol);
+ if (ret < 0) {
+ printf ("Volume %s is not of type replicate\n", volname);
+ goto out;
+ }
+ rootloc.inode = inode_ref (top_subvol->itable->root);
+ glfs_loc_touchup (&rootloc);
+
+ xl = top_subvol;
+ while (xl->next)
+ xl = xl->next;
+
+ while (xl) {
+ if (strcmp (xl->type, "protocol/client") == 0) {
+ afr_xl = _get_afr_ancestor (xl);
+ if (afr_xl) {
+ old_THIS = THIS;
+ THIS = afr_xl;
+ glfsh_print_pending_heals (xl, &rootloc);
+ THIS = old_THIS;
+ printf("\n");
+ }
+ }
+
+ xl = xl->prev;
+ }
+
+ loc_wipe (&rootloc);
+ glfs_subvol_done (fs, top_subvol);
+ glfs_fini (fs);
+
+ return 0;
+out:
+ if (fs && top_subvol)
+ glfs_subvol_done (fs, top_subvol);
+ loc_wipe (&rootloc);
+ if (fs)
+ glfs_fini (fs);
+
+ return ret;
+}
diff --git a/tests/basic/afr/self-heald.t b/tests/basic/afr/self-heald.t
new file mode 100644
index 00000000000..8500a14d8c3
--- /dev/null
+++ b/tests/basic/afr/self-heald.t
@@ -0,0 +1,161 @@
+#!/bin/bash
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../volume.rc
+
+cleanup;
+
+function disconnected_brick_count {
+ local vol=$1
+ $CLI volume heal $vol info | grep -i transport | wc -l
+}
+
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{0,1,2,3,4,5}
+TEST $CLI volume set $V0 cluster.background-self-heal-count 0
+TEST $CLI volume set $V0 cluster.eager-lock off
+TEST $CLI volume start $V0
+TEST glusterfs --volfile-id=/$V0 --volfile-server=$H0 $M0 --attribute-timeout=0 --entry-timeout=0
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST kill_brick $V0 $H0 $B0/${V0}2
+TEST kill_brick $V0 $H0 $B0/${V0}4
+cd $M0
+HEAL_FILES=0
+for i in {1..10}
+do
+ dd if=/dev/urandom of=f bs=1M count=10 2>/dev/null
+ HEAL_FILES=$(($HEAL_FILES+1)) #+1 for data/metadata self-heal of 'f'
+ mkdir a; cd a;
+ #+3 for metadata self-heal of 'a' one per subvolume of DHT
+ HEAL_FILES=$(($HEAL_FILES+3))
+done
+#+3 represents entry sh on "/", one per subvolume of DHT?
+HEAL_FILES=$(($HEAL_FILES + 3))
+
+cd ~
+EXPECT "$HEAL_FILES" afr_get_pending_heal_count $V0
+
+#When bricks are down, it says Transport End point Not connected for them
+EXPECT "3" disconnected_brick_count $V0
+
+#Create some stale indices and verify that they are not counted in heal info
+#TO create stale index create and delete files when one brick is down in
+#replica pair.
+for i in {11..20}; do echo abc > $M0/$i; done
+HEAL_FILES=$(($HEAL_FILES + 10)) #count extra 10 files
+EXPECT "$HEAL_FILES" afr_get_pending_heal_count $V0
+#delete the files now, so that stale indices will remain.
+for i in {11..20}; do rm -f $M0/$i; done
+#After deleting files they should not appear in heal info
+HEAL_FILES=$(($HEAL_FILES - 10))
+EXPECT "$HEAL_FILES" afr_get_pending_heal_count $V0
+
+
+TEST ! $CLI volume heal $V0
+TEST $CLI volume set $V0 cluster.self-heal-daemon off
+TEST ! $CLI volume heal $V0
+TEST ! $CLI volume heal $V0 full
+TEST $CLI volume start $V0 force
+TEST $CLI volume set $V0 cluster.self-heal-daemon on
+EXPECT_WITHIN 20 "Y" glustershd_up_status
+EXPECT_WITHIN 20 "1" afr_child_up_status_in_shd $V0 0
+EXPECT_WITHIN 20 "1" afr_child_up_status_in_shd $V0 2
+EXPECT_WITHIN 20 "1" afr_child_up_status_in_shd $V0 4
+TEST $CLI volume heal $V0
+sleep 5 #Until the heal-statistics command implementation
+#check that this heals the contents partially
+TEST [ $HEAL_FILES -gt $(afr_get_pending_heal_count $V0) ]
+
+TEST $CLI volume heal $V0 full
+EXPECT_WITHIN 30 "0" afr_get_pending_heal_count $V0
+
+#Test that ongoing IO is not considered as Pending heal
+(dd if=/dev/zero of=$M0/file1 bs=1K 2>/dev/null 1>/dev/null)&
+back_pid1=$!;
+(dd if=/dev/zero of=$M0/file2 bs=1K 2>/dev/null 1>/dev/null)&
+back_pid2=$!;
+(dd if=/dev/zero of=$M0/file3 bs=1K 2>/dev/null 1>/dev/null)&
+back_pid3=$!;
+(dd if=/dev/zero of=$M0/file4 bs=1K 2>/dev/null 1>/dev/null)&
+back_pid4=$!;
+(dd if=/dev/zero of=$M0/file5 bs=1K 2>/dev/null 1>/dev/null)&
+back_pid5=$!;
+EXPECT 0 afr_get_pending_heal_count $V0
+kill -SIGTERM $back_pid1;
+kill -SIGTERM $back_pid2;
+kill -SIGTERM $back_pid3;
+kill -SIGTERM $back_pid4;
+kill -SIGTERM $back_pid5;
+wait >/dev/null 2>&1;
+
+#Test that volume heal info reports files even when self-heal
+#options are disabled
+TEST touch $M0/f
+TEST mkdir $M0/d
+#DATA
+TEST $CLI volume set $V0 cluster.data-self-heal off
+EXPECT "off" volume_option $V0 cluster.data-self-heal
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST kill_brick $V0 $H0 $B0/${V0}2
+TEST kill_brick $V0 $H0 $B0/${V0}4
+echo abc > $M0/f
+EXPECT 1 afr_get_pending_heal_count $V0
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN 20 "Y" glustershd_up_status
+EXPECT_WITHIN 20 "1" afr_child_up_status_in_shd $V0 0
+EXPECT_WITHIN 20 "1" afr_child_up_status_in_shd $V0 2
+EXPECT_WITHIN 20 "1" afr_child_up_status_in_shd $V0 4
+TEST $CLI volume heal $V0
+EXPECT_WITHIN 30 "0" afr_get_pending_heal_count $V0
+TEST $CLI volume set $V0 cluster.data-self-heal on
+
+#METADATA
+TEST $CLI volume set $V0 cluster.metadata-self-heal off
+EXPECT "off" volume_option $V0 cluster.metadata-self-heal
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST kill_brick $V0 $H0 $B0/${V0}2
+TEST kill_brick $V0 $H0 $B0/${V0}4
+TEST chmod 777 $M0/f
+EXPECT 1 afr_get_pending_heal_count $V0
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN 20 "Y" glustershd_up_status
+EXPECT_WITHIN 20 "1" afr_child_up_status_in_shd $V0 0
+EXPECT_WITHIN 20 "1" afr_child_up_status_in_shd $V0 2
+EXPECT_WITHIN 20 "1" afr_child_up_status_in_shd $V0 4
+TEST $CLI volume heal $V0
+EXPECT_WITHIN 30 "0" afr_get_pending_heal_count $V0
+TEST $CLI volume set $V0 cluster.metadata-self-heal on
+
+#ENTRY
+TEST $CLI volume set $V0 cluster.entry-self-heal off
+EXPECT "off" volume_option $V0 cluster.entry-self-heal
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST kill_brick $V0 $H0 $B0/${V0}2
+TEST kill_brick $V0 $H0 $B0/${V0}4
+TEST touch $M0/d/a
+EXPECT 2 afr_get_pending_heal_count $V0
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN 20 "Y" glustershd_up_status
+EXPECT_WITHIN 20 "1" afr_child_up_status_in_shd $V0 0
+EXPECT_WITHIN 20 "1" afr_child_up_status_in_shd $V0 2
+EXPECT_WITHIN 20 "1" afr_child_up_status_in_shd $V0 4
+TEST $CLI volume heal $V0
+EXPECT_WITHIN 30 "0" afr_get_pending_heal_count $V0
+TEST $CLI volume set $V0 cluster.entry-self-heal on
+
+#Negative test cases
+#Fail volume does not exist case
+TEST ! $CLI volume heal fail info
+
+#Fail volume stopped case
+TEST $CLI volume stop $V0
+TEST ! $CLI volume heal $V0 info
+
+#Fail non-replicate volume info
+TEST $CLI volume delete $V0
+TEST $CLI volume create $V0 $H0:$B0/${V0}{6}
+TEST $CLI volume start $V0
+TEST ! $CLI volume heal $V0 info
+
+cleanup
diff --git a/tests/basic/self-heald.t b/tests/basic/self-heald.t
deleted file mode 100644
index 0e4d104cd13..00000000000
--- a/tests/basic/self-heald.t
+++ /dev/null
@@ -1,44 +0,0 @@
-#!/bin/bash
-
-. $(dirname $0)/../include.rc
-. $(dirname $0)/../volume.rc
-
-cleanup;
-
-TEST glusterd
-TEST pidof glusterd
-TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{0,1,2,3,4,5}
-TEST $CLI volume set $V0 cluster.background-self-heal-count 0
-TEST $CLI volume set $V0 cluster.eager-lock off
-TEST $CLI volume start $V0
-TEST $GFS --volfile-id=/$V0 --volfile-server=$H0 $M0
-TEST kill_brick $V0 $H0 $B0/${V0}0
-TEST kill_brick $V0 $H0 $B0/${V0}2
-TEST kill_brick $V0 $H0 $B0/${V0}4
-cd $M0
-HEAL_FILES=0
-for i in {1..10}
-do
- dd if=/dev/urandom of=f bs=1024k count=10 2>/dev/null
- HEAL_FILES=$(($HEAL_FILES+1))
- mkdir a; cd a;
- HEAL_FILES=$(($HEAL_FILES+3)) #As many times as distribute subvols
-done
-HEAL_FILES=$(($HEAL_FILES + 3)) #Count the brick root dir
-
-cd ~
-EXPECT "$HEAL_FILES" afr_get_pending_heal_count $V0
-TEST ! $CLI volume heal $V0
-TEST $CLI volume set $V0 cluster.self-heal-daemon off
-TEST ! $CLI volume heal $V0 info
-TEST ! $CLI volume heal $V0
-TEST $CLI volume start $V0 force
-TEST $CLI volume set $V0 cluster.self-heal-daemon on
-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 2
-EXPECT_WITHIN $CHILD_UP_TIMEOUT "1" afr_child_up_status_in_shd $V0 4
-
-TEST $CLI volume heal $V0 full
-EXPECT_WITHIN $HEAL_TIMEOUT "0" afr_get_pending_heal_count $V0
-cleanup
diff --git a/tests/bugs/bug-880898.t b/tests/bugs/bug-880898.t
index 8ba35ec1a59..4b9fb50a522 100644
--- a/tests/bugs/bug-880898.t
+++ b/tests/bugs/bug-880898.t
@@ -17,7 +17,7 @@ do
fi
done
-gluster volume heal $V0 info | grep "Status: self-heal-daemon is not running on $uuid";
-EXPECT "0" echo $?
+#Command execution should fail reporting that the bricks are not running.
+TEST ! $CLI volume heal $V0 info
cleanup;
diff --git a/xlators/cluster/afr/src/afr-self-heal-common.c b/xlators/cluster/afr/src/afr-self-heal-common.c
index 14a514beffa..6198d4cf72c 100644
--- a/xlators/cluster/afr/src/afr-self-heal-common.c
+++ b/xlators/cluster/afr/src/afr-self-heal-common.c
@@ -932,13 +932,14 @@ afr_selfheal_unlocked_inspect (call_frame_t *frame, xlator_t *this,
if (replies[i].op_ret == -1)
continue;
- if (afr_is_data_set (this, replies[i].xdata))
+ if (data_selfheal && afr_is_data_set (this, replies[i].xdata))
*data_selfheal = _gf_true;
- if (afr_is_metadata_set (this, replies[i].xdata))
+ if (metadata_selfheal &&
+ afr_is_metadata_set (this, replies[i].xdata))
*metadata_selfheal = _gf_true;
- if (afr_is_entry_set (this, replies[i].xdata))
+ if (entry_selfheal && afr_is_entry_set (this, replies[i].xdata))
*entry_selfheal = _gf_true;
valid_cnt ++;
@@ -967,7 +968,8 @@ afr_selfheal_unlocked_inspect (call_frame_t *frame, xlator_t *this,
priv->children[i]->name,
uuid_utoa (replies[i].poststat.ia_gfid));
- *metadata_selfheal = _gf_true;
+ if (metadata_selfheal)
+ *metadata_selfheal = _gf_true;
}
if (!IA_EQUAL (first, replies[i].poststat, gid)) {
@@ -978,7 +980,8 @@ afr_selfheal_unlocked_inspect (call_frame_t *frame, xlator_t *this,
priv->children[i]->name,
uuid_utoa (replies[i].poststat.ia_gfid));
- *metadata_selfheal = _gf_true;
+ if (metadata_selfheal)
+ *metadata_selfheal = _gf_true;
}
if (!IA_EQUAL (first, replies[i].poststat, prot)) {
@@ -989,7 +992,8 @@ afr_selfheal_unlocked_inspect (call_frame_t *frame, xlator_t *this,
priv->children[i]->name,
uuid_utoa (replies[i].poststat.ia_gfid));
- *metadata_selfheal = _gf_true;
+ if (metadata_selfheal)
+ *metadata_selfheal = _gf_true;
}
if (IA_ISREG(first.ia_type) &&
@@ -1001,11 +1005,12 @@ afr_selfheal_unlocked_inspect (call_frame_t *frame, xlator_t *this,
priv->children[i]->name,
uuid_utoa (replies[i].poststat.ia_gfid));
- *data_selfheal = _gf_true;
+ if (data_selfheal)
+ *data_selfheal = _gf_true;
}
}
- if (valid_cnt > 0) {
+ if (valid_cnt > 0 && link_inode) {
*link_inode = afr_inode_link (inode, &first);
if (!*link_inode) {
ret = -EINVAL;
diff --git a/xlators/cluster/afr/src/afr-self-heal.h b/xlators/cluster/afr/src/afr-self-heal.h
index f208e6bc813..bb3f0e7029f 100644
--- a/xlators/cluster/afr/src/afr-self-heal.h
+++ b/xlators/cluster/afr/src/afr-self-heal.h
@@ -198,4 +198,11 @@ afr_mark_active_sinks (xlator_t *this, unsigned char *sources,
gf_boolean_t
afr_does_witness_exist (xlator_t *this, uint64_t *witness);
+
+int
+afr_selfheal_unlocked_inspect (call_frame_t *frame, xlator_t *this,
+ uuid_t gfid, inode_t **link_inode,
+ gf_boolean_t *data_selfheal,
+ gf_boolean_t *metadata_selfheal,
+ gf_boolean_t *entry_selfheal);
#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 10e19964ed9..a341015303c 100644
--- a/xlators/cluster/afr/src/afr-self-heald.c
+++ b/xlators/cluster/afr/src/afr-self-heald.c
@@ -40,8 +40,6 @@
#define NTH_INDEX_HEALER(this, n) &((((afr_private_t *)this->private))->shd.index_healers[n])
#define NTH_FULL_HEALER(this, n) &((((afr_private_t *)this->private))->shd.full_healers[n])
-int afr_shd_gfid_to_path (xlator_t *this, xlator_t *subvol, uuid_t gfid, char **path_p);
-
char *
afr_subvol_name (xlator_t *this, int subvol)
{
diff --git a/xlators/cluster/afr/src/afr-self-heald.h b/xlators/cluster/afr/src/afr-self-heald.h
index 59f06b79cff..02b26b8061f 100644
--- a/xlators/cluster/afr/src/afr-self-heald.h
+++ b/xlators/cluster/afr/src/afr-self-heald.h
@@ -67,4 +67,10 @@ afr_selfheal_daemon_init (xlator_t *this);
int
afr_xl_op (xlator_t *this, dict_t *input, dict_t *output);
+int
+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);
#endif /* !_AFR_SELF_HEALD_H */