Gluster users - May 2012 - [REPOST] Connection failed

[Reposting because I still have the same problem]

Hello, all.

I'm trying to get glusterfs working on two machines (so that I can have
replicated storage on both of them) and I'm stuck on getting glusterd
working.

The two machines are Debian 6.0 (Squeeze) and I'm using the glusterfs
packages from the backports repo (3.2.4-1~bpo60+1).

/etc/glusterfs/glusterd.vol on both machines contains:

  volume management
      type mgmt/glusterd
      option working-directory /etc/glusterd
      option transport-type socket
      option transport.socket.keepalive-time 10
      option transport.socket.keepalive-interval 2
  end-volume

When I start glusterd, I get different results on the two servers:

root at silicium:~# gluster peer status
Number of Peers: 1

Hostname: titane
Uuid: 448e3316-74e3-44aa-a495-5b540e7b8927
State: Peer in Cluster (Connected)

root at titane:~# gluster peer status
Connection failed. Please check if gluster daemon is operational.

The output from 'glusterd --debug' is here:

http://people.parinux.org/~seyman/gluster/glusterd.log
or
http://people.parinux.org/~seyman/gluster/glusterd.log.gz

Any comments or suggestions are welcome.

Emmanuel

(Missed cc-ing to the list. Re-adding again)

Hi again.

AFAICT, glusterd is running properly, it must be a problem with cli then.
glusterd listens for cli connections on port 24007. cli tries to connect to
this port before performing any operation. When the connection cannot
happen, cli reports the message you get.
Since, glusterd is running in your case, something must be preventing cli
from connecting to the port, firewall probably.

Regards,
Kaushal


On Tue, May 22, 2012 at 7:19 PM, Emmanuel Seyman <eseyman at edd.fr>
wrote:
> * Kaushal M [22/05/2012 15:30] :
> >
> > Hi Emmanuel.
>
> Hi, Kaushal.
>
> > The log you have provided, is it for titane or silicium? I have a
feeling
> > it is for the latter.
>
> Yes, it is for titane.
>
> > If it is titane's log, the thing I can make out is that glusterd
was
> killed
> > by a SIGINT. This is probably because --debug doesn't daemonize
the
> > glusterd process and you used ctrl-c top stop it.
>
> Correct. The last two lines were written after the 'gluster peer
status'
> command was run so they're not really relevent to the output (I
probably
> should have removed them from the log)
>
> > Could you test again with option '--log-level=DEBUG' instead
and provide
> > the logs?
>
> Done.
> I've replaced the original log files.
>
> http://people.parinux.org/~seyman/gluster/glusterd.log
> http://people.parinux.org/~seyman/gluster/glusterd.log.gz
>
> Thanks for the help.
>
> Emmanuel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://supercolony.gluster.org/pipermail/gluster-users/attachments/20120523/7152ff1b/attachment.html>