Bug 1179640

+++ This bug was initially created as a clone of Bug #1178590 +++

Description of problem:
I create a disperse 3 redundancy 1 volume, start, then enable quota. I down one brick, copy some files and directories to mountpooint, then restart thart brick. exec 
find /cluster2/test/ -d -exec getfattr -h -n trusted.ec.heal {} \; (I have mager this patch)
All REG files is OK, trusted.ec.heal="Good: 111, Bad: 000".
but All DIR is failed, trusted.ec.heal="Good: 110, Bad: 000"
exec find /cluster2/test/ -d -exec getfattr -h -n trusted.ec.heal {} \; again, DIRs also can not be healed.

Then I disable quota, exec  find /cluster2/test/ -d -exec getfattr -h -n trusted.ec.heal {} \;, All DIRs are OK.

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


How reproducible:


Steps to Reproduce:
1.create a disperse 3 redundancy 1 volume
2.enable quota default.
3.down one brick.
4.copy some files and directories to mountpoint.
5.restart that brick.
6.find /cluster2/test/ -d -exec getfattr -h -n trusted.ec.heal {} \;
7.All REGs are OK, but All DIRs are failed. 

Actual results:
Heal DIR is failed when quota is enbale.

Expected results:
Heal DIR is OK when quota is enbale.


Additional info:

--- Additional comment from jiademing on 2015-01-04 21:51:08 EST ---

Bricks error log:

[2014-12-26 10:51:40.116480] E [quota.c:3303:quota_setxattr] 0-test-quota: attempt to set internal xattr: trusted.digioceanfs.quota*: Operation not permitted

--- Additional comment from Jeff Darcy on 2015-01-06 08:11:01 EST ---

jiademing, I can't find the patch you mentioned.  Could you please post it to review.gluster.org or (less optimally) attach it here?  Thanks.

This might relate to (or interact with) http://review.gluster.org/#/c/9385/

--- Additional comment from Anand Avati on 2015-01-07 04:02:17 EST ---

REVIEW: http://review.gluster.org/9401 (cluster/ec: Do not modify quota, selinux xattrs in healing) posted (#1) for review on master by Pranith Kumar Karampuri (pkarampu)

--- Additional comment from Pranith Kumar K on 2015-01-07 04:03:26 EST ---

Patch above fixes the issue reported in the 'steps'. There is still more to the bug where quota-limit xattr is not allowed to be healed unless the pid of the process doing it is -ve i.e. internal process. I will be sending that patch once I discuss the solution with Xavi.

--- Additional comment from jiademing on 2015-01-07 04:12:04 EST ---

(In reply to Jeff Darcy from comment #2)
> jiademing, I can't find the patch you mentioned.  Could you please post it
> to review.gluster.org or (less optimally) attach it here?  Thanks.
> 
> This might relate to (or interact with) http://review.gluster.org/#/c/9385/

I meger this patch http://review.gluster.org/#/c/9072/, so I can use "find /cluster2/test/ -d -exec getfattr -h -n trusted.ec.heal {} \;" to heal the files and dirs.

I view that patch http://review.gluster.org/#/c/9385/, it reject user to modify the xattrs of ec, but can not reslove above problem.

ec want to heal xattrs of DIRS, but the quota translater doesn't allow ec to modify some xattr (like trusted.glusterfs.quota.dirty), so lead to heal xattrs failed.

Bricks error log:
[2014-12-26 10:51:40.116480] E [quota.c:3303:quota_setxattr] 0-test-quota: attempt to set internal xattr: trusted.glusterfs.quota*: Operation not permitted