Doing rm -rf on the glusterfs mount point errors out saying resource temporarily unavailable and Input/Output error. 

Setup:

3.0.5rc7(latest git pull from 3.0 series), distributed replicate setup. 

4 servers (distribute to 2 which in turn replicated to 2 servers).

2 clients

volgen generated volfile, hence all the performance translators are there.

I started running sanity scripts on one client and linux kernel untarring on other. While doing this I was doing server up-down. After untarring of linux kernel did make defconfig and then  started removing it. While this was going on mounted a 3.1 client accidently and then unmounted it. Fileop was running on 1st client when kernel directory was removed. Stopped sanity tests and then started removing all the contents on the mount point on the 1st client and simultaneously did 'ls -laR .` on another mount point. While this was going on did server up-down. 


After this rm completed but I could still see the directory on the mount point, and now doing rm -rf errored out.



a) Doing rm -rf on 1st client says both "resource temporarily unavailable and Inout/Output error:"

b) Doing rm -rf on the 2nd client says only "resource temporarily unavailable".


The logfiles of the servers and clients and the statedump information is attaches.

used mainline clients and servers in the mentioned config. Did I/O on both clients and took servers down intermittently. This was followed by rm and ls commands. Both succeeded.

Please re-open the issue if it occurs again.

Regards,
Shishir