diff options
author | Shyam <srangana@redhat.com> | 2014-08-12 10:48:27 -0400 |
---|---|---|
committer | Kaleb KEITHLEY <kkeithle@redhat.com> | 2014-10-20 07:54:59 -0700 |
commit | 7a1e42e0d4be42fa8aa0c7a430b4fbc3ab509705 (patch) | |
tree | ca1780e2f1c5e5fc139c2e0a625ff8456426582b /tests | |
parent | 91175b38c9264676d75a275c16add45f7c64f4c1 (diff) |
cluster/dht: Fix dht_access treating directory like files
When the cluster topology changes due to add-brick, all sub
volumes of DHT will not contain the directories till a rebalance
is completed. Till the rebalance is run, if a caller bypasses
lookup and calls access due to saved/cached inode information
(like NFS server does) then, dht_access misreads the error
(ESTALE/ENOENT) from the new subvolumes and incorrectly tries
to handle the inode as a file. This results in the directories
in memory state in DHT to be corrupted and not heal even post
a rebalance.
This commit fixes the problem in dht_access thereby preventing
DHT from misrepresenting a directory as a file in the case
presented above.
Change-Id: Idcdaa3837db71c8fe0a40ec0084a6c3dbe27e772
BUG: 1139997
Signed-off-by: Shyam <srangana@redhat.com>
Reviewed-on: http://review.gluster.org/8462
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Vijay Bellur <vbellur@redhat.com>
Reviewed-on: http://review.gluster.org/8678
Reviewed-by: Kaleb KEITHLEY <kkeithle@redhat.com>
Diffstat (limited to 'tests')
-rwxr-xr-x | tests/bugs/bug-1125824.t | 95 | ||||
-rw-r--r-- | tests/include.rc | 1 |
2 files changed, 96 insertions, 0 deletions
diff --git a/tests/bugs/bug-1125824.t b/tests/bugs/bug-1125824.t new file mode 100755 index 00000000000..c51ef9c16e1 --- /dev/null +++ b/tests/bugs/bug-1125824.t @@ -0,0 +1,95 @@ +#!/bin/bash + +. $(dirname $0)/../include.rc +. $(dirname $0)/../volume.rc +. $(dirname $0)/../nfs.rc + +create_files () { + for i in {1..10}; do + orig=$(printf %s/file%04d $1 $i) + echo "This is file $i" > $orig + done + for i in {1..10}; do + mkdir $(printf %s/dir%04d $1 $i) + done + sync +} + +create_dirs () { + for i in {1..10}; do + mkdir $(printf %s/dir%04d $1 $i) + create_files $(printf %s/dir%04d $1 $i) + done + sync +} + +stat_files () { + for i in {1..10}; do + orig=$(printf %s/file%04d $1 $i) + stat $orig + done + for i in {1..10}; do + stat $(printf %s/dir%04d $1 $i) + done + sync +} + +stat_dirs () { + for i in {1..10}; do + stat $(printf %s/dir%04d $1 $i) + stat_files $(printf %s/dir%04d $1 $i) + done + sync +} + +cleanup; + +TEST glusterd +TEST pidof glusterd +TEST $CLI volume info; + +TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{1,2,3,4}; + +EXPECT "$V0" volinfo_field $V0 'Volume Name'; +EXPECT 'Created' volinfo_field $V0 'Status'; +EXPECT '4' brick_count $V0 + +TEST $CLI volume start $V0; +EXPECT 'Started' volinfo_field $V0 'Status'; + +EXPECT_WITHIN $NFS_EXPORT_TIMEOUT "1" is_nfs_export_available; +TEST mount_nfs $H0:/$V0 $N0 + +# Create and poulate the NFS inode tables +TEST create_dirs $N0 +TEST stat_dirs $N0 + +# add-bricks changing the state of the volume where some bricks +# would have some directories and others would not +TEST $CLI volume add-brick $V0 replica 2 $H0:$B0/${V0}{5,6,7,8} + +# Post this dht_access was creating a mess for directories which is fixed +# with this commit. The issues could range from getting ENOENT or +# ESTALE or entries missing to directories not having complete +# layouts. +TEST cd $N0 +TEST ls -lR + +TEST $CLI volume rebalance $V0 start force +EXPECT_WITHIN $REBALANCE_TIMEOUT "completed" rebalance_status_field $V0 + +# tests to check post rebalance if layouts and entires are fine and +# accessible by NFS to clear the volume +TEST ls -lR +rm -rf ./* + +# Get out of the mount, so that umount can work +TEST cd / + +EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $N0 +TEST $CLI volume stop $V0; +EXPECT 'Stopped' volinfo_field $V0 'Status'; +TEST $CLI volume delete $V0; +TEST ! $CLI volume info $V0; + +cleanup; diff --git a/tests/include.rc b/tests/include.rc index 54427dbda50..141e6581c5d 100644 --- a/tests/include.rc +++ b/tests/include.rc @@ -9,6 +9,7 @@ DEBUG=${DEBUG:=0} # turn on debugging? NFS_EXPORT_TIMEOUT=20 UMOUNT_TIMEOUT=5 +REBALANCE_TIMEOUT=200 statedumpdir=`gluster --print-statedumpdir`; # Default directory for statedump |