Unfortunately the VM-Manager uses the wrong error-code => error-string mapping. The error number 21 is the following: CANNOT_RESTORE_MACHINE. So the problem is that the machine cannot be restored after the new disk image is created. The following lines are taken form the provisioning.log file:
Feb 25 16:35:37 foss-cloud-node-02 provisioning.pl[19015]: KVM: Restoring machine 2ca6c09d-54b6-4550-949b-d09ccaf9e0ba from /var/virtualization/retain/vm-persistent/0f83f084-8080-413e-b558-b678e504836e/2ca6c09d-54b6-4550-949b-d09ccaf9e0ba/20130225T163500Z/2ca6c09d-54b6-4550-949b-d09ccaf9e0ba.state
Feb 25 16:35:58 foss-cloud-node-02 provisioning.pl[19015]: KVM: Error from libvirt (38): libvirt says: Unable to read from monitor: Connection reset by peer.
The same error is thrown when you try to restore the machine manually using the virsh restore command:
foss-cloud-node-02 syslog-ng.conf.d # virsh restore /var/virtualization/retain/vm-persistent/0f83f084-8080-413e-b558-b678e504836e/2ca6c09d-54b6-4550-949b-d09ccaf9e0ba/20130225T163500Z/2ca6c09d-54b6-4550-949b-d09ccaf9e0ba.state
error: Failed to restore domain from /var/virtualization/retain/vm-persistent/0f83f084-8080-413e-b558-b678e504836e/2ca6c09d-54b6-4550-949b-d09ccaf9e0ba/20130225T163500Z/2ca6c09d-54b6-4550-949b-d09ccaf9e0ba.state
error: Unable to read from monitor: Connection reset by peer
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
This seems to be QEMU related bug. We are in the process of building a new release candidate with an updated libvirt and qemu version and we'll try to verify, if the bug still exists and still causes a segfault.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Diff:
We could verify the bug.
Unfortunately the VM-Manager uses the wrong error-code => error-string mapping. The error number 21 is the following: CANNOT_RESTORE_MACHINE. So the problem is that the machine cannot be restored after the new disk image is created. The following lines are taken form the provisioning.log file:
The same error is thrown when you try to restore the machine manually using the virsh restore command:
Don't know what the VM-Manager should do to prevent this
This seems to be QEMU related bug. We are in the process of building a new release candidate with an updated libvirt and qemu version and we'll try to verify, if the bug still exists and still causes a segfault.
The downgrade of qemu to version 1.3.1 fixed the problem.