Bug #5152

VMs are not polling correctly

Added by Kristian Feldsam about 2 years ago. Updated about 2 years ago.

Status:PendingStart date:05/15/2017
Priority:HighDue date:
Assignee:-% Done:

0%

Category:Drivers - VM
Target version:Release 5.6
Resolution: Pull request:
Affected Versions:OpenNebula 5.2

Description

Hello, I experience few times bug with VM monitoring. When I execute VM live migration and it's fail from whatever reason, oned assume that VM is still running on original host with message

Fail to live migrate VM. Assuming that the VM is still RUNNING (will poll VM).

But never poll not running VM. So VM is parctically not running anywhere, but oned shows that it is running on original host. When I manually execute 'virsh destroy one-ID', oned after while gets info about that. So there is some problem after failed migration.


Related issues

Duplicated by Feature #3624: Revisit KVM/libvirt to OpenNebula state mapping Closed 02/19/2015

History

#1 Updated by Ruben S. Montero about 2 years ago

Hi Kristian

Let me understand this better, so the VM is live migrated and the migration fails. So the VM should be running in the original host, but in your case is not?

#2 Updated by Kristian Feldsam about 2 years ago

Hi, I investigate it deeper and in same cases, failed migration leaves VM in state "paused". VMM driver assume, that paused state = running. Driver should try to unpause VM and if it can't them should destroy and create.

#3 Updated by Ruben S. Montero about 2 years ago

  • Target version changed from Release 5.4 to Release 5.6

#4 Updated by Jaime Melis over 1 year ago

  • Duplicated by Feature #3624: Revisit KVM/libvirt to OpenNebula state mapping added

Also available in: Atom PDF