[Gluster-users] stat-prefetch error and possible memory leak

[Zack Lesko]

Hi,

I am seeing the following error message logged on one of two replicated
storage servers (CentOS 5.6 x64, two bricks, one volume). This server has
Symantec Backup Exec client installed and I am backing it up using this. The
error seems to coincide with the time the backup begins, which also seems to
coincide with a dramatic increase in glusterfs process memory usage which is
not freed with drop_caches.

The other replicated storage server is using rsync to an internet backup
service, and I have not noticed any errors or memory leaks on it recently.

[2011-07-22 09:01:03.585923] E
[stat-prefetch.c:695:sp_remove_caches_from_all_fds_opened]
(-->/opt/glusterfs/3.2.2/lib64/glusterfs/3.2.2/xlator/mount/fuse.so(fuse_setattr_resume+0x157)
[0x2af23d326b77]
(-->/opt/glusterfs/3.2.2/lib64/glusterfs/3.2.2/xlator/debug/io-stats.so(io_stats_setattr+0x139)
[0x2aaaac0896f9]
(-->/opt/glusterfs/3.2.2/lib64/glusterfs/3.2.2/xlator/performance/stat-prefetch.so(sp_setattr+0x7e)
[0x2aaaabe7667e]))) 0-data-stat-prefetch: invalid argument: inode

Thanks,
Zack
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20110722/5030f7bd/attachment.html>