[Gluster-devel] Spurious disconnects

Vijay Bellur vbellur at redhat.com
Wed Dec 26 05:58:26 UTC 2012


On 12/25/2012 08:58 AM, Emmanuel Dreyfus wrote:
> Hi
>
> I still see spurious disconnections on 3.4.0qaX, but I have trouble to
> understand what is going on. I am not even sure this is always the same
> problem, and it is really annoying since it always end up in total
> failure, with the volume getting unusable beacause all bricks are down.

With 3.4.0, brick processes make use of ports from 49152 onwards 
(instead of 24009). Do you per chance have any firewall rules that can 
cause disconnects?

>
> Anyone can help reading the following logs? They are produced by four
> concurent tar -xzf on a volume called gfs33:
>
> Client log:
> http://ftp.espci.fr/shadow/manu/gfs33.log
>
> Server bricks logs:
> http://ftp.espci.fr/shadow/manu/client-0-export-wd3a.log
> http://ftp.espci.fr/shadow/manu/client-1-export-wd3a.log
> http://ftp.espci.fr/shadow/manu/client-2-export-wd1a.log
> http://ftp.espci.fr/shadow/manu/client-3-export-wd1a.log

Nothing much to glean from the logs apart from the fact that there are a 
bunch of annoying ENOENT messages in the bricks. Can you change the log 
level to DEBUG and see if you notice anything amiss there?

Thanks,
Vijay





More information about the Gluster-devel mailing list