Bug 763672 (GLUSTER-1940) - Directory corruption
Summary: Directory corruption
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: GLUSTER-1940
Product: GlusterFS
Classification: Community
Component: core
Version: 3.0.5
Hardware: All
OS: Linux
low
high
Target Milestone: ---
Assignee: Amar Tumballi
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2010-10-13 02:17 UTC by Joe Julian
Modified: 2015-12-01 16:45 UTC (History)
2 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed:
Regression: ---
Mount Type: fuse
Documentation: ---
CRM:
Verified Versions:


Attachments (Terms of Use)
client and server logs and xattrs. (469.35 KB, application/gzip)
2010-10-12 23:20 UTC, Joe Julian
no flags Details

Description Joe Julian 2010-10-12 23:20:29 UTC
Created attachment 352

Comment 1 Joe Julian 2010-10-13 02:17:07 UTC
I'm unable to delete a directory as I'm getting the error, "Directory not empty"

See the attached configuration and log files

To produce this result, I attempted to build through prep the centos kernel (rpmbuild -bp kernel-2.6.spec) then rm -rf kernel-2.6.18 (from the BUILD directory).

Comment 2 Amar Tumballi 2010-10-13 03:16:17 UTC
Can you run a 'find $mountpoint' and then attempt directory deletion? 

Also, was there any server disconnects while you were running kernel untar process?

Comment 3 Amar Tumballi 2010-10-13 03:19:39 UTC
Joe, Btw, I see that you are using 3.0.5. Can you try with our latest release 3.1.0, and see if this issue is fixed?

Comment 4 Joe Julian 2010-10-13 03:32:03 UTC
The xattr dump was done using find. Additionally, I ran ls -laR in the {mountpoint}/home/jjulian/dev/rpmbuild/BUILD the subfolder of which is the one with the problem.

I cannot run 3.1 yet as I don't see a way to disable client-side performance translators for just my mysql host (mysql tables corrupt with client side performance translators last time I tested it).

Comment 5 Joe Julian 2010-10-15 20:34:09 UTC
Vijay had me try patch 5504 with 'option use-readdirp yes' which seems to have solved the directory corruption I was having.

Comment 6 Amar Tumballi 2010-10-20 03:41:04 UTC
Joe,
Can we close this bug as fixed then?? We have this patch in 3.1 codebase, and now committed it to 3.0.x branch also.. 

Regards,


Note You need to log in before you can comment on or make changes to this bug.