Kvm + Libvirt + Virt-manager
Hi list, I have updated C7 to 7.1 and get some issue (I don’t know if problems depend from upgrade). My first problem is on virt-manager that crash after some time. From system messages I get that virt-manager is crashed with signal sigsegv and this problem is related to python (python get sigsegv). Anyone get similar behaviour after upgrade?
Another problem, I can’t say if it showed before/after upgrade, is related to libvirt when I shutdown my system. On my host I have different vm runned on KVM, where only 2 have autostart enabled. When shutting down the system I get from the console an error like this:
Suspending test1: Done Suspending test2: … error: Failed to save domain $someid state error: internal error: unable to execute QEMU command ‘migrate’: State blocked by non-migratable device ‘0000:00:04.8/ich9_ahci’
Vm test2 has a disk attached as raw device and not as image file. This message, with same error, is showed also for 2 other vm with autostart disabled and disk on qcow2 images, but only when them are online during shutdown.
When system starts, all vms are running without problem.
I can’t figure out with this error…
Why libvirt-guest.sh run migrate command when I shutdown the system?
Hope in help.
Thanks in advance
2 thoughts on - Kvm + Libvirt + Virt-manager
I also see this problem. It happened on CentOS 7.0 as well. If I leave virt-
manager running, then after 1 to 3 days it will crash. The virtual machines remain running. In CentOS 7.0 I could not restart virt-manager unless I
rebooted the computer. In 7.1 virt-manager will restart.
The host computer is a Q9440 processor with 8 gig of RAM and a single hard drive. I use 64-bit CentOS on it.
Bill Gee
This issue seems to be reported by ABRT in 2 bug reports (#8472 and
#8592 .. links from the below abrt report)
This ABRT bug is tracked here:
https://retrace.fedoraproject.org/faf/problems/796335/
Red Hat Engineers know that CentOS now dumps abrt information to this location and they do now use that to help them prioritize bugs with high bug counts that also apply to RHEL. Obviously, CentOS does not have any SLA or priority, but chances are much better now that there are integrated ABRT reports that things like this might be fixed.