[Bugs] [Bug 1402297] New: Enabling Quota on volumes leads inconsistent data

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1402297

            Bug ID: 1402297
           Summary: Enabling Quota on volumes leads inconsistent data
           Product: GlusterFS
           Version: 3.7.16
         Component: quota
          Severity: low
          Assignee: bugs at gluster.org
          Reporter: pankaj at quadeyesecurities.com
                CC: bugs at gluster.org



Description of problem:

After enabling quota, mounted data are not showing on client side. We have 8
volume of varying sizes. We are enabling quota on 3 volumes of 4TB each in
sizes.
We have installed GlusterFS community version 3.7.16 on AWS. But once we mount
GlusterFS Volumes to the client side with quota enabled then we are not able to
get consistent data. Sometime it shows and sometime it doesn't. If shown then
we are not able to create or write on the files. Please help us to resolve this
problem 

Version-Release number of selected component (if applicable):
3.7.16

How reproducible:
Install glusterfs on AWS and create few volumes having lots of varying size of
data. Enable quota and then mount your Volumes on client. after few moment or
days filesystem becomes inconsistent. 

Steps to Reproduce:
1. Install GlusterFS on AWS 2 ec2-instances replicated with transport type
socket
2. Now enable quota on few volumes and mount them onto client
3. After few moment it will become inconsistent. 

Actual results:
Data on client side become inconsistent over the time of period. 

Expected results:
It should work perfectly fine without any inconsistency in data at client side
by enabling quota on volumes.

Additional info:
We keep monitoring directory based quota limit and all are under our control
but still we are facing this issue. Some strange behavior regarding file system
attributes. We see some duplicate quota attributes attached on our files under
same volume which have quota enabled. If we clear those particular attributes
by "setfattr -x " command then data become consistent. But we can't clear
thousand of file manually on both glusetrfs server nodes.  Please help us to
resolve this issue or tell us if we have to do some explicit configuration for
same. We are following GlusterFS documentation for configuration.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
You are the assignee for the bug.