Description of problem:
=======================

In one of the automation case, executed rm -rf from the master mount. It cleared all the content from master but slave complained for the following errors:

[2016-11-21 23:01:51.599591] W [syncdutils(slave):506:errno_wrap] <top>: reached maximum retries (['ebdcd4fb-3c7e-406e-87b0-b94bdc858055', '.gfid/0091d587-073c-40bb-b415-1f6c3e3424b8/level80', '.gfid/0091d587-073c-40bb-b415-1f6c3e3424b8/level80'])...[Errno 39] Directory not empty: '.gfid/0091d587-073c-40bb-b415-1f6c3e3424b8/level80'
[2016-11-21 23:01:51.602191] W [resource(slave):733:entry_ops] <top>: Recursive remove ebdcd4fb-3c7e-406e-87b0-b94bdc858055 => .gfid/0091d587-073c-40bb-b415-1f6c3e3424b8/level80failed: Directory not empty
[2016-11-21 23:01:51.631956] W [syncdutils(slave):506:errno_wrap] <top>: reached maximum retries (['ebdcd4fb-3c7e-406e-87b0-b94bdc858055', '.gfid/0091d587-073c-40bb-b415-1f6c3e3424b8/level80', '.gfid/0091d587-073c-40bb-b415-1f6c3e3424b8/level80'])...[Errno 39] Directory not empty: '.gfid/0091d587-073c-40bb-b415-1f6c3e3424b8/level80'
[2016-11-21 23:01:51.632200] W [resource(slave):733:entry_ops] <top>: Recursive remove ebdcd4fb-3c7e-406e-87b0-b94bdc858055 => .gfid/0091d587-073c-40bb-b415-1f6c3e3424b8/level80failed: Directory not empty
[2016-11-21 23:01:57.232242] W [syncdutils(slave):506:errno_wrap] <top>: reached maximum retries (['0091d587-073c-40bb-b415-1f6c3e3424b8', '.gfid/471f647f-cc13-41ed-b25a-a909882490a5/level70', '.gfid/471f647f-cc13-41ed-b25a-a909882490a5/level70'])...[Errno 39] Directory not empty: '.gfid/471f647f-cc13-41ed-b25a-a909882490a5/level70/level80'
[2016-11-21 23:01:57.232630] W [resource(slave):733:entry_ops] <top>: Recursive remove 0091d587-073c-40bb-b415-1f6c3e3424b8 => .gfid/471f647f-cc13-41ed-b25a-a909882490a5/level70failed: Directory not empty

Following was the directory structure remaining at the slave client:

#ls /mnt/slave/
thread3
#ls /mnt/slave/thread3/
level00
#ls /mnt/slave/thread3/level00/
level10
#ls /mnt/slave/thread3/level00/level10/
level20
#ls /mnt/slave/thread3/level00/level10/level20/
level30
#ls /mnt/slave/thread3/level00/level10/level20/level30/
level40
#ls /mnt/slave/thread3/level00/level10/level20/level30/level40/
level50
#ls /mnt/slave/thread3/level00/level10/level20/level30/level40/level50/
level60
#ls /mnt/slave/thread3/level00/level10/level20/level30/level40/level50/level60/
level70
#ls /mnt/slave/thread3/level00/level10/level20/level30/level40/level50/level60/level70/
level80
#ls /mnt/slave/thread3/level00/level10/level20/level30/level40/level50/level60/level70/level80/
#

Re-executing rm -rf * directly on the slave client resolves the issue as:

[root@dhcp37-64 slave]# rm -rf *
[root@dhcp37-64 slave]# ls -l
total 0
[root@dhcp37-64 slave]# ls -lR
.:
total 0
[root@dhcp37-64 slave]# 


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

glusterfs-3.8.4-5.el6rhs.x86_64


How reproducible:
=================

Have run the same case multiple times on RHEL7 based platform and havent seen this. Will run once again on RHEL6 based platform.


Steps Carried:
==============

Carried automated suit (34) of geo-replication