Hello,
i use glusterfs version 3.6.1 on 2 bricks in 1 distribute volume. the volume is
mounted on 1 client manually by mount -t nfs server1:vol1 /mnt/gfs. if i copy
smal files e.g. up to 500MB, its no problem. But if i copy large files e.g.
>30gb nfs gets slower until its stops and the system hangs. then i could only
hard power down by powerswitch.
in syslog i found:
kernel: task mc blocked for more than 120 seconds...
not tainted 3.16-0.bpo.2-amd64
..
call trace.
....
nfs_file_fsync...
nfs_file_write...
new_sync_write...
...
and so on. The System has 11 disk in raid6 with xfs and 8GB RAM. No other
services are running. Searching in web i found glusterfs bug 977492. i saw that
volume default options nfs.outstanding-rpc-limit and nfs.drc has changed since
a few version, but it seems that this doesnt work for my simple workload. But i
tried option nfs.outstanding-rpc-limit: 8 instead of default 16. the result i
could copy some larger files 40gb at once, but as the second starts i got the
same problem again.
I hope you can help. thx
_______________________________________________
Gluster-users mailing list
[email protected]
http://www.gluster.org/mailman/listinfo/gluster-users
