[libvirt-users] libvirt restarting vms on restart?

Recent versions of libvirt seem to have taken to starting new copies of my vms when they restart. Any vm marked to start when libvirt is started is started, even if it is already running, which leads to multiple running vms trying to use the same disk image. I think this probably started with the upgrade from 0.7.4 to 0.7.5 but I can't be entirely sure as I only realised yesterday that I had somehow managed to get multiple instances of a VM running. It definitely happened this morning when the updated from 0.7.5.-1 to 0.7.5-3 in the F12 virt-preview repo caused a restart. Tom -- Tom Hughes (tom@compton.nu) http://www.compton.nu/

On Fri, Jan 15, 2010 at 09:45:13AM +0000, Tom Hughes wrote:
Recent versions of libvirt seem to have taken to starting new copies of my vms when they restart.
Any vm marked to start when libvirt is started is started, even if it is already running, which leads to multiple running vms trying to use the same disk image.
I think this probably started with the upgrade from 0.7.4 to 0.7.5 but I can't be entirely sure as I only realised yesterday that I had somehow managed to get multiple instances of a VM running. It definitely happened this morning when the updated from 0.7.5.-1 to 0.7.5-3 in the F12 virt-preview repo caused a restart.
It sounds like you are probably hitting this bug https://bugzilla.redhat.com/show_bug.cgi?id=550400 the trouble is I can't reproduce it myself. When you manually start a VM, can you check whether the /var/run/libvirt/qemu/$GUEST.xml file contains the monitor path info. Daniel -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|

On 15/01/10 11:03, Daniel P. Berrange wrote:
It sounds like you are probably hitting this bug
https://bugzilla.redhat.com/show_bug.cgi?id=550400
the trouble is I can't reproduce it myself. When you manually start a VM, can you check whether the /var/run/libvirt/qemu/$GUEST.xml file contains the monitor path info.
It doesn't, no. After I noticed the problem this morning I quickly killed both qemus with -9 to minimise disk access then restarted the VM from virt-manager and the xml file currently looks like: <domstatus state='running' pid='25809'> <domain type='kvm' id='2'> ... </domain> </domstatus> Tom -- Tom Hughes (tom@compton.nu) http://www.compton.nu/

On Fri, Jan 15, 2010 at 11:08:21AM +0000, Tom Hughes wrote:
On 15/01/10 11:03, Daniel P. Berrange wrote:
It sounds like you are probably hitting this bug
https://bugzilla.redhat.com/show_bug.cgi?id=550400
the trouble is I can't reproduce it myself. When you manually start a VM, can you check whether the /var/run/libvirt/qemu/$GUEST.xml file contains the monitor path info.
It doesn't, no. After I noticed the problem this morning I quickly killed both qemus with -9 to minimise disk access then restarted the VM from virt-manager and the xml file currently looks like:
<domstatus state='running' pid='25809'> <domain type='kvm' id='2'> ... </domain> </domstatus>
FYI, I've posted a fix for this issue http://www.redhat.com/archives/libvir-list/2010-January/msg00545.html Daniel -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|
participants (2)
-
Daniel P. Berrange
-
Tom Hughes