[Bugs] [Bug 1283036] New: Index entries are not being purged in case of file does not exist

Wed Nov 18 05:01:38 UTC 2015

https://bugzilla.redhat.com/show_bug.cgi?id=1283036

            Bug ID: 1283036
           Summary: Index entries are not being purged in case of file
                    does not exist
           Product: GlusterFS
           Version: 3.7.6
         Component: replicate
          Assignee: bugs at gluster.org
          Reporter: aspandey at redhat.com
                CC: atalur at redhat.com, bugs at gluster.org,
                    gluster-bugs at redhat.com, pkarampu at redhat.com,
                    ravishankar at redhat.com
        Depends On: 1270668

+++ This bug was initially created as a clone of Bug #1270668 +++

Description of problem:
If a file does not exist on a brick but the index entry is present in
XXX/.glusterfs/indices/xattrop, this entry is not getting removed even after 
running index heal.

Version-Release number of selected component (if applicable):

[root at aspandey:/home/blr/aspandey/glusterfs]# glusterfs --version
glusterfs 3.8dev built on Oct 12 2015 10:40:32
Repository revision: git://git.gluster.com/glusterfs.git
Copyright (c) 2006-2013 Red Hat, Inc. <http://www.redhat.com/>
GlusterFS comes with ABSOLUTELY NO WARRANTY.
It 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.
[root at aspandey:/home/blr/aspandey/glusterfs]# 

How reproducible:
100%

Steps to Reproduce:
1. Create a replicate volume, mount it and create some files on mount point.
2. add bricks and before executing rebalance, kill one of the new add bricks
3. execute rebalance.
4. add brick again and execute rebalance (some files from written on bricks in
[2] will be rebalanced to bricks added in [4]) 
5. start the volume with force
6. start heal
7. heal info shows nothing to be healed  
8. Look into XXX/.glusterfs/indices/xattrop in one of the brick added in [2].
Index entries would still be present.

Actual results:
Indices are present for the file which does not exist on volume 

Expected results:

No indices should be present.

Additional info:

[root at aspandey:/home/blr/aspandey/glusterfs]# gluster v heal repvol info
Brick aspandey:/brick/gluster/R1
Number of entries: 0

Brick aspandey:/brick/gluster/R2
Number of entries: 0

Brick aspandey:/brick/gluster/R11
Number of entries: 0

Brick aspandey:/brick/gluster/R12
Number of entries: 0

Brick aspandey:/brick/gluster/R111
Number of entries: 0

Brick aspandey:/brick/gluster/R112
Number of entries: 0

[root at aspandey:/home/blr/aspandey/glusterfs]# ll
/brick/gluster/R11/.glusterfs/indices/xattrop/
total 0
----------. 4 root root 0 Oct 12 10:48 26b306f3-7204-461b-9b61-3147651ee181
----------. 4 root root 0 Oct 12 10:48 2e37a37f-ce9f-4ccc-8c9c-fb4eefded44a
----------. 4 root root 0 Oct 12 10:48 2ea426f0-0e7a-43a3-ae1c-b9463be4062e
----------. 4 root root 0 Oct 12 10:48
xattrop-e5af794a-3622-47be-8b9d-65a9f211975d
[root at aspandey:/home/blr/aspandey/glusterfs]# gluster v info

Volume Name: repvol
Type: Distributed-Replicate
Volume ID: e33f7dc2-3a21-4b0f-9965-f039f6dbb003
Status: Started
Number of Bricks: 3 x 2 = 6
Transport-type: tcp
Bricks:
Brick1: aspandey:/brick/gluster/R1
Brick2: aspandey:/brick/gluster/R2
Brick3: aspandey:/brick/gluster/R11
Brick4: aspandey:/brick/gluster/R12
Brick5: aspandey:/brick/gluster/R111
Brick6: aspandey:/brick/gluster/R112
Options Reconfigured:
performance.readdir-ahead: on

--- Additional comment from Vijay Bellur on 2015-10-12 04:24:19 EDT ---

REVIEW: http://review.gluster.org/12336 (cluster/afr : Remove index entries if
corresponding               file does not exist.) posted (#1) for review on
master by Ashish Pandey (aspandey at redhat.com)

--- Additional comment from Vijay Bellur on 2015-10-16 04:28:12 EDT ---

REVIEW: http://review.gluster.org/12336 (cluster/afr : Remove index entries if
corresponding               file does not exist.) posted (#2) for review on
master by Ashish Pandey (aspandey at redhat.com)

--- Additional comment from Vijay Bellur on 2015-10-29 12:05:12 EDT ---

REVIEW: http://review.gluster.org/12336 (cluster/afr : Remove stale indices)
posted (#3) for review on master by Ashish Pandey (aspandey at redhat.com)

--- Additional comment from Vijay Bellur on 2015-11-16 03:20:12 EST ---

COMMIT: http://review.gluster.org/12336 committed in master by Pranith Kumar
Karampuri (pkarampu at redhat.com) 
------
commit 92e3bbbad803688a4dbcbab6bcd35867aa055da1
Author: Ashish Pandey <aspandey at redhat.com>
Date:   Mon Oct 12 13:14:08 2015 +0530

    cluster/afr : Remove stale indices

    Change-Id: Iba23338a452b49dc9fe6ae7b4ca108ebc377fe42
    BUG: 1270668
    Signed-off-by: Ashish Pandey <aspandey at redhat.com>
    Reviewed-on: http://review.gluster.org/12336
    Tested-by: Gluster Build System <jenkins at build.gluster.com>
    Reviewed-by: Pranith Kumar Karampuri <pkarampu at redhat.com>

Referenced Bugs:

https://bugzilla.redhat.com/show_bug.cgi?id=1270668
[Bug 1270668] Index entries are not being purged in case of file does not
exist
-- 
You are receiving this mail because:
You are on the CC list for the bug.
You are the assignee for the bug.