[Gluster-users] Upgrade from 3.5 to 3.6: issue with socket

[Krishnan Parthasarathi]

> Actually I could solve my issue with the mounting of the volume by
> re-installing Gluster 3.6 from scratch. Before that I tried to update the
> volfile as I read in a bug but that did not work. So the only last problem
> which remains is this warning error with the socket as reported.
> 
> As I did indeed disable NFS on all my volumes (currently only one) I suppose
> this is why as you suggest Joe. But this is really annoying having a log
> entry every 3 seconds in gluster's log file. How can I fix or avoid that? I
> would say this might be a bug?

This is a bug. Could you raise a bug and provide the id in this thread?
Meanwhile, I will send out a patch for review on master and release-3.6 branch.
Thanks for reporting this issue.

Thanks Joe Julian for giving the crucial piece of lead which helped me identify
the issue.

~kp

> 
> 
> > Gesendet: Montag, 09. März 2015 um 04:13 Uhr
> > Von: "Joe Julian" <joe at julianfamily.org>
> > An: gluster-users at gluster.org
> > Betreff: Re: [Gluster-users] Upgrade from 3.5 to 3.6: issue with socket
> >
> > I've seen this when the subprocesses running glustershd and nfs were not
> > restarted after the upgrade. I've also seen it if nfs is disabled on all
> > volumes.
> > 
> > On 03/08/2015 09:10 PM, Krishnan Parthasarathi wrote:
> > >> I just upgraded from 3.5.3 to 3.6.2 and have issues mounting my volume
> > >> on a
> > >> client. On the server side I found this error message which might be the
> > >> cause of my issues:
> > > Could you describe the issues you are facing?
> > >
> > >> [2015-03-08 13:22:36.383715] W [socket.c:611:__socket_rwv] 0-management:
> > >> readv on /var/run/6b8f1f2526c6af8a87f1bb611ae5a86f.socket failed
> > >> (Invalid
> > >> argument)
> > >>
> > >> Anyony has any idea what that could be ?
> > > This log message indicates that there was no (server) process listening
> > > on
> > > the socket identified by the path, when glusterd tried to connect(3).
> > > This
> > > happens if the connect(3) happened sooner than the server process starts
> > > listening.
> > > glusterd will attempt a (re)connect within the next 3 seconds and should
> > > successfully
> > > establish connection with the server. This wouldn't result in any sort of
> > > 'service disruption' in the I/O path.
> > >
> > > Could you confirm that this log snippet is from glusterd's log file?
> > > _______________________________________________
> > > Gluster-users mailing list
> > > Gluster-users at gluster.org
> > > http://www.gluster.org/mailman/listinfo/gluster-users
> > 
> > _______________________________________________
> > Gluster-users mailing list
> > Gluster-users at gluster.org
> > http://www.gluster.org/mailman/listinfo/gluster-users
> >
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://www.gluster.org/mailman/listinfo/gluster-users