[Users] vdsm unresponsive with python exception

[Dan Kenigsberg]

On Thu, Apr 11, 2013 at 03:51:07PM -0500, Tony Feldmann wrote:
> That was the issue.  Found out yesterday that vdsm.log was somehow changed
> to root:root.  Just now got a chance to put it back on the mailing list.
> How does the ownership of that file get cahnged.  When the issue occurred I
> am certain there was no one on the system.

http://gerrit.ovirt.org/#/c/12940/ (Separating supervdsm log to
supervdsm.log file) solves the issue. unfortunately, only on the master
branch of vdsm.

I think that this is a nasty issue that has to be backported to the
ovirt-3.2 branch as well, and merits to be part of ovirt-3.2.2.

Regards,
Dan.

> 
> 
> On Thu, Apr 11, 2013 at 2:15 PM, Joop <jvdwege at xs4all.nl> wrote:
> 
> > Dan Kenigsberg wrote:
> >
> >> On Wed, Apr 10, 2013 at 08:59:01AM -0500, Tony Feldmann wrote:
> >>
> >>
> >>> I am having a strange issue in my ovirt cluster.  I have 2 hosts, 1
> >>> running
> >>> engine and added as a host and one other system added as a host.  Both
> >>> systems are running gluster across local disks for shared storage.
> >>> Everything was working fine until last night, where my system that is
> >>> also
> >>> running the engine when unresponsive in the admin page.  All vms were
> >>> still
> >>> running that were on the host.  I shut down the vms that were on the host
> >>> from within the guest os as I was not able to do anything to the vm with
> >>> the host in unresponsive state.  After getting the vms off and rebooting
> >>> the host, the vdsmd service says that it is running, but it continually
> >>> restarts the vdsm process and dumps out these messages: detected
> >>> unhandled
> >>> Python exception in '/usr/share/vdsm/vdsm'.  All services say they are up
> >>> and running but the host stays in unresponsive state and the vdsm process
> >>> keeps respawning.  There is also no data in the vdsm.log.  Can anyone
> >>> shed
> >>> any light on this for me?
> >>>
> >>>
> >>
> >> vdsm-devel at fedorahosted.org may be a better place to ask vdsm-specific
> >> questions.
> >>
> >> Could you log into the non-operational host as root, and stop the vdsm
> >> service.
> >>
> >> Then become the vdsm user with
> >>
> >>     su -s /bin/bash - vdsm
> >>
> >> and run /usr/share/vdsm/vdsm manually. Do you see anything in
> >> particular?
> >>
> >>
> >>
> > Please have a look at the permissions/owner of /var/log/vdsm/vdsm.log.
> > Should be vdsm:kvm and not root:root
> >
> > Joop
> >
> >

> _______________________________________________
> Users mailing list
> Users at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users