Red Hat Bugzilla – Bug 866578
engine: cannot stop vm that was suspended after qemu stopped responding and was re initiated
Last modified: 2015-10-27 19:56:36 EDT
Created attachment 627551 [details]
Description of problem:
I ran a vm and started suspending it. while it was in saving state I ran kill -STOP <qemu pid>
after the vm becamse non-responding I ran kill -CONT <qemu pid>
the vm returns to up state while somewhere in the db its still suspended.
as a result when we try to stop the vm we do not send destroy to the vm's pid and only delete the image group.
vm is not stopped when we try to stop it.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. start suspending a vm
2. run in host kill -STOP <qemu pid>
3. when vm becomes none-responsive run kill -CONT <qemu pid>
4. when vm returns to up state try to stop it
vm is not stopped.
if vm becomes non-responsive during saving state we should roll back on suspend.
if we also restart libvirt while qemu is down vm will not be shut down unless we kill its pid.
Dafna, what is the real life use case of this scenario?
I fail to understand how a user will get unintentionally to this scenario and why RHEV should handle it if it was done on purpose.
if qemu crashes during hibernate.
may not happen a lot but its possible.
(In reply to comment #2)
> if qemu crashes during hibernate.
> may not happen a lot but its possible.
Qemu crash is not the same as:
1.run in host kill -STOP <qemu pid>
2. when vm becomes none-responsive run kill -CONT <qemu pid>
SIGCONT 19,18,25 Cont Continue if stopped
SIGSTOP 17,19,23 Stop Stop process
I don't see these happen in normal workflow.
Does it happen when you kill -9?