[Gluster-users] Memory leak in rebalance

[Emir Imamagic]

Hello,

we are using glusterfs 3.2.5 and have a distributed volume with over 10M 
directories. We recently added new node and initiated rebalance. After 
several days glusterfsd consumed all the memory and was killed by the 
kernel. At that stage it was still doing the layout rebalance and got 
over 9M directories.

Rebalance reports failed status:
# gluster volume rebalance vol status
rebalance failed

In the glusterd log however I found:
  I [glusterd-rebalance.c:473:glusterd_defrag_start] 0-rebalance: 
rebalance on /etc/glusterd/mount/vol complete

I stumbled upon the patch on gluster-devel which mentions memory leak 
related rebalance: 
http://dev.gluster.com/pipermail/glusterfs/2011-June/003369.html. But 
can't figure out if this was included in the 3.2.5 release.

Another question is - is it safe to initiate rebalance of the data?

Thanks in advance
-- 
Emir Imamagic
Sektor za racunalne sustave
Sveuciliste u Zagrebu, Sveucilisni racunski centar (Srce), www.srce.unizg.hr
Emir.Imamagic at srce.hr, tel: +385 1 616 5809, fax: +385 1 616 5559