[Gluster-users] Gluster peers disconnecting

[James]

Dear Gluster,

I'm running Gluster 3.3 on a four host setup. (Two bricks per host.)

I'm attempting to use this as an rsnapshot backup system. Periodically,
the rsync's seem to fail, and I think this is due to underlying gluster
failures. I notice this in /var/log/messages:

# Jul 19 23:47:15 annex1 GlusterFS[26815]: [2012-07-19 23:47:15.424909]
C [client-handshake.c:126:rpc_client_ping
_timer_expired] 0-gbackups-client-2: server 172.16.1.83:24009 has not
responded in the last 42 seconds, disconnecting.

I notice that df shows a smaller than normal volume size, which returns
to normal after a remount.

$ df -h
172.16.1.80:/gbackups
55T 1.6T 54T 3% /mnt/gbackups
 
$ umount /mnt/gbackups
$ df -h
172.16.1.80:/gbackups
73T 1.9T 71T 3% /mnt/gbackups

One rsnapshot error that I get is:
# [19/Jul/2012:23:50:01] /usr/bin/rsnapshot
-c /etc/rsnapshot/rsnapshot-sftp/rsnapshot.conf sync: ERROR: /bin/cp
-al /mnt/gbackups/rsnapshot/sftp/daily.0 /mnt/gbackups/rsnapshot/sftp/.sync failed (result 256, exit status 1). Perhaps your cp does not support -al options?

I've manually tested the cp -al and it's working fine. It's most likely
due to a node that's dropping out temporarily.

Doing a gluster peer status shows all nodes "connected".

Any ideas what is causing this please?
Thanks in advance,
James


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20120720/3988e210/attachment.sig>