Bug 810046
| Summary: | GlusterFS replica mount hangs forever on "toxic" file healing. | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Community] GlusterFS | Reporter: | Jeff Byers <jbyers> | ||||
| Component: | replicate | Assignee: | Pranith Kumar K <pkarampu> | ||||
| Status: | CLOSED DEFERRED | QA Contact: | |||||
| Severity: | high | Docs Contact: | |||||
| Priority: | unspecified | ||||||
| Version: | 3.2.5 | CC: | agk, bugs, gluster-bugs, jdarcy, natasa.cyrus, sheggodu | ||||
| Target Milestone: | --- | ||||||
| Target Release: | --- | ||||||
| Hardware: | x86_64 | ||||||
| OS: | Linux | ||||||
| Whiteboard: | |||||||
| Fixed In Version: | Doc Type: | Bug Fix | |||||
| Doc Text: | Story Points: | --- | |||||
| Clone Of: | Environment: | ||||||
| Last Closed: | 2014-12-14 19:40:33 UTC | Type: | Bug | ||||
| Regression: | --- | Mount Type: | --- | ||||
| Documentation: | --- | CRM: | |||||
| Verified Versions: | Category: | --- | |||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||
| Embargoed: | |||||||
| Attachments: |
|
||||||
|
Description
Jeff Byers
2012-04-05 01:15:05 UTC
It seems that this problem is in part related to the use of the SGID (Set Group ID) dir/file mode attributes, that were added to the Samba 'smb.conf' file to resolve an unrelated GlusterFS issue Bug 802423 -GlusterFS does not work well with MS Office 2010 and Samba "posix locking = yes". https://bugzilla.redhat.com/show_bug.cgi?id=802423 [Gluster-users] Using Samba and MSOffice files http://gluster.org/pipermail/gluster-users/2011-November/009127.html The thought that the SGID mode bit was somehow related to the problem was because the 2-way replica toxic files showed the SGID bit set on one copy, but not the other. However, AFAIK, the Samba host was not ever changing the SGID bit or creating files without it, so the case where the SGID bit was lost must be related to replica healing in some way. When I changed the Samba 'smb.conf' settings from: directory mask = 2770 force directory mode = 2770 create mask = 2770 force create mode = 2770 force security mode = 2770 force directory security mode = 2770 to: directory mask = 770 force directory mode = 770 create mask = 770 force create mode = 770 force security mode = 770 force directory security mode = 770 the "toxic" file healing problem does not occur, where it occured most of the time otherwise. This can act as a work-around, but it should be possible to use the SGID file/directory mode bits if desired. Jeff,
>>I have tared one of the problematic directories from each brick,
>> and saved the extended attributes and reproduce the problem with
>>these files on a different system. If it would be helpful, I
>>could upload these files.
We would like to recreate the problem in our setup. Could you provide the xattrs which result in this scenario.
Pranith
Created attachment 591616 [details]
The extended file attributes dumps as requested.
Jeff Byers,
What are the .dmp files? How to open them?. If I just create such a directory hierarchy and assign the xattrs I should be able to see the issue?. Should it be done only on samba or it can happen even on fuse client?.
Pranith
The .dmp files are the output of 'getfattr' with the '-d' option, and is in ascii text format. This format can be used to restore attributes to files using 'setfattr'. Not sure if you could just restore and reproduce, it has been almost 2.5 months. I was surprised that I still had the files. Since Samba only accesses through the GlusterFS fuse client, so it probably shouldn't be necessary. The version that this bug has been reported against, does not get any updates from the Gluster Community anymore. Please verify if this report is still valid against a current (3.4, 3.5 or 3.6) release and update the version, or close this bug. If there has been no update before 9 December 2014, this bug will get automatocally closed. |