[libvirt] F11 and KVM Migrations
I have two host machines on the same subnet, and I have /var/lib/libvirt/images shared between the two via NFS. I'm trying to do a "migration" from one host to another and I'm getting an error. virsh # migrate --live Narwhal qemu+ssh://10.1.1.1/system error: operation failed: failed to start listening VM That's from the host running the VM to the new host. I'm assuming you have to do a "push" migrate, not a "pull"? Where can I look next to figure out why it didn't work? -- Scott Baker - Canby Telcom System Administrator - RHCE - 503.266.8253
On Wed, Jul 08, 2009 at 01:21:53PM -0700, Scott Baker wrote:
I have two host machines on the same subnet, and I have /var/lib/libvirt/images shared between the two via NFS. I'm trying to do a "migration" from one host to another and I'm getting an error.
virsh # migrate --live Narwhal qemu+ssh://10.1.1.1/system error: operation failed: failed to start listening VM
That's from the host running the VM to the new host. I'm assuming you have to do a "push" migrate, not a "pull"? Where can I look next to figure out why it didn't work?
Hum that sounds familiar, what versions of libvirt are you running on both nodes, look at /var/log/libvirt/qemu/Narwhal.log on the target for informations. I suggest to make sure that you get up2date and similar versions on both sides to avoid extra troubles. Daniel -- Daniel Veillard | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ daniel@veillard.com | Rpmfind RPM search engine http://rpmfind.net/ http://veillard.com/ | virtualization library http://libvirt.org/
On 07/08/2009 02:02 PM, Daniel Veillard wrote:
On Wed, Jul 08, 2009 at 01:21:53PM -0700, Scott Baker wrote:
I have two host machines on the same subnet, and I have /var/lib/libvirt/images shared between the two via NFS. I'm trying to do a "migration" from one host to another and I'm getting an error.
virsh # migrate --live Narwhal qemu+ssh://10.1.1.1/system error: operation failed: failed to start listening VM
That's from the host running the VM to the new host. I'm assuming you have to do a "push" migrate, not a "pull"? Where can I look next to figure out why it didn't work?
Hum that sounds familiar, what versions of libvirt are you running on both nodes, look at /var/log/libvirt/qemu/Narwhal.log on the target for informations. I suggest to make sure that you get up2date and similar versions on both sides to avoid extra troubles.
Host 1: libvirt-0.6.2-12.fc11.x86_64 Host 2: libvirt-0.6.2-12.fc11.i586 Both machine are 64 bit, Host 2 is just running i386 code. Will that be a problem? If I check Narwhal.log there is nothing in there at the time of the migration error. Nothing in messages either. As far as I can tell it just fails silently. -- Scott Baker - Canby Telcom System Administrator - RHCE - 503.266.8253
On Wed, Jul 08, 2009 at 02:05:11PM -0700, Scott Baker wrote:
On 07/08/2009 02:02 PM, Daniel Veillard wrote:
On Wed, Jul 08, 2009 at 01:21:53PM -0700, Scott Baker wrote:
I have two host machines on the same subnet, and I have /var/lib/libvirt/images shared between the two via NFS. I'm trying to do a "migration" from one host to another and I'm getting an error.
virsh # migrate --live Narwhal qemu+ssh://10.1.1.1/system error: operation failed: failed to start listening VM
That's from the host running the VM to the new host. I'm assuming you have to do a "push" migrate, not a "pull"? Where can I look next to figure out why it didn't work?
Hum that sounds familiar, what versions of libvirt are you running on both nodes, look at /var/log/libvirt/qemu/Narwhal.log on the target for informations. I suggest to make sure that you get up2date and similar versions on both sides to avoid extra troubles.
Host 1: libvirt-0.6.2-12.fc11.x86_64 Host 2: libvirt-0.6.2-12.fc11.i586
Both machine are 64 bit, Host 2 is just running i386 code. Will that be a problem?
yes that can be a problem !
If I check Narwhal.log there is nothing in there at the time of the migration error. Nothing in messages either. As far as I can tell it just fails silently.
Maybe we need to resurrect the idea of the pre-migration check Also is Narwhal still running on Host 1 after the operation failed, based on the error message I assume the error shows up after a fair amount of time spent sending the domain state between both nodes, right ? Daniel -- Daniel Veillard | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ daniel@veillard.com | Rpmfind RPM search engine http://rpmfind.net/ http://veillard.com/ | virtualization library http://libvirt.org/
Scott Baker wrote:
On 07/08/2009 02:02 PM, Daniel Veillard wrote:
On Wed, Jul 08, 2009 at 01:21:53PM -0700, Scott Baker wrote:
I have two host machines on the same subnet, and I have /var/lib/libvirt/images shared between the two via NFS. I'm trying to do a "migration" from one host to another and I'm getting an error.
virsh # migrate --live Narwhal qemu+ssh://10.1.1.1/system error: operation failed: failed to start listening VM
That's from the host running the VM to the new host. I'm assuming you have to do a "push" migrate, not a "pull"? Where can I look next to figure out why it didn't work? Hum that sounds familiar, what versions of libvirt are you running on both nodes, look at /var/log/libvirt/qemu/Narwhal.log on the target for informations. I suggest to make sure that you get up2date and similar versions on both sides to avoid extra troubles.
Host 1: libvirt-0.6.2-12.fc11.x86_64 Host 2: libvirt-0.6.2-12.fc11.i586
Both machine are 64 bit, Host 2 is just running i386 code. Will that be a problem?
Yeah, that's just not going to work. In theory it might work, but it's never been tested, so I'm not surprised it doesn't. In general migration is extremely finicky when it comes to CPU versions, and versions of the software. If you are interested in debugging it more, I would suggest stopping libvirtd on both sides, starting them back up by hand with: LIBVIRT_DEBUG=1 /usr/sbin/libvirtd --verbose --listen And then re-running your test. That will dump lots of debugging information on both machines, and you might be able to get somewhat of a better idea of what is happening. -- Chris Lalancette
On 07/09/2009 01:43 AM, Chris Lalancette wrote:
Yeah, that's just not going to work. In theory it might work, but it's never been tested, so I'm not surprised it doesn't. In general migration is extremely finicky when it comes to CPU versions, and versions of the software.
If you are interested in debugging it more, I would suggest stopping libvirtd on both sides, starting them back up by hand with:
LIBVIRT_DEBUG=1 /usr/sbin/libvirtd --verbose --listen
And then re-running your test. That will dump lots of debugging information on both machines, and you might be able to get somewhat of a better idea of what is happening.
Host 1: [root@vm1 ~]# rpm -q libvirt libvirt-0.6.2-12.fc11.x86_64 [root@vm1 ~]# uname -r 2.6.29.5-191.fc11.x86_64 Host 2: [root@vm2 ~]# rpm -q libvirt libvirt-0.6.2-12.fc11.x86_64 [root@vm2 ~]# uname -r 2.6.29.5-191.fc11.x86_64 Host 2 doesn't want to start with --verbose: 08:16:44.859: debug : virEventInterruptLocked:635 : Skip interrupt, 0 0 08:16:44.859: debug : virEventAddHandleImpl:111 : Add handle 11 13 0x412c70 0x13368a0 08:16:44.859: debug : virEventInterruptLocked:635 : Skip interrupt, 0 0 08:16:44.868: error : Cannot access CA certificate '/etc/pki/CA/cacert.pem': No such file or directory What's weird is, if I start the service it works fine. It only errors about that cacert if I do it raw. I can't find any reference to cacert in my config. Help! -- Scott Baker - Canby Telcom System Administrator - RHCE - 503.266.8253
participants (3)
-
Chris Lalancette -
Daniel Veillard -
Scott Baker