Bug 761857 (GLUSTER-125) - stat information not returned from the same subvolume always
Summary: stat information not returned from the same subvolume always
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: GLUSTER-125
Product: GlusterFS
Classification: Community
Component: replicate
Version: mainline
Hardware: All
OS: Linux
low
medium
Target Milestone: ---
Assignee: Vikas Gorur
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks: GLUSTER-106
TreeView+ depends on / blocked
 
Reported: 2009-07-09 17:36 UTC by Vikas Gorur
Modified: 2009-11-23 07:38 UTC (History)
2 users (show)

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

Attachments (Terms of Use)
Add an attachment (proposed patch, testcase, etc.)

Description Vikas Gorur 2009-07-09 17:36:13 UTC 
replicate returns stat info from one child once and from another child later. Can be reproduced by running the pjd-fstest POSIX compliance test suite, tests/open/00.t. If the clocks on the servers are not in sync, the test will fail.
Comment 1 Vikas Gorur 2009-07-11 09:06:11 UTC 
The bug is in the various inode-write calls in replicate. They also return stat info, but do not follow the read-subvolume logic.
Comment 2 Vikas Gorur 2009-07-14 09:20:00 UTC 
We can easily fix this by returning stat info from the read child in all the inode-write ops too. Doing this, however, means we will lose the "quick unwind" feature, as we will have to wait until the call has returned from the read child to unwind. Is it worth it?
Comment 3 Vikas Gorur 2009-07-14 10:36:39 UTC 
Patch submitted: http://patches.gluster.com/patch/744/
Comment 4 Anand Avati 2009-07-16 04:37:40 UTC 
PATCH: http://patches.gluster.com/patch/744 in master (Return stat info from read-child in all the inode-write ops)
Comment 5 Anand Avati 2009-07-27 13:32:37 UTC 
PATCH: http://patches.gluster.com/patch/744 in release-2.0 (Return stat info from read-child in all the inode-write ops)
Note You need to log in before you can comment on or make changes to this bug.