RHEVM: VM started booted as a different VM.

Latest response

Hi everyone

RHEVM 3.1.0-43

I experienced something very odd and rather worrying whilst starting a RHEL VM in RHEV today.

I started up a VM and montiored the console and noticed that it was booting a a completely different VM !

It was immediately noticeable because I was expecting it to boot a RHEL5 VM but instead booted a completely different RHEL6 VM.

The title on the Spice console screen reflected the name of my expected RHEL5 VM , so I know I didn't inadvertantly start the wrong one.

Fortunately the RHEL6 VM it booted was not already online so did not corruption the data. I powered off the VM.

I suspected that somewhere in the database a disk was being incorrectly referenced, perhaps corrupted and so took a dump to compare against the previous night postgres backup.  On first inspection it all checks out as fine, no duplicate disk ID's or corrupted entries. All appears to check out as to what is being displayed under the Disks tab in webadmin. Puzzled, I restarted the VM and this time it booted the correct VM.  I had changed nothing in between times.

Has anyone experienced this ?   Can anyone explain what may have happened ?  

I've opened a support case.

Responses

Could it have been a browser issue ?

The reason I ask is becuase we suffer from the bug in 3.1.0-43 which leaves open postgres connections, requiring a regular restart of the ovirt-engine and postgresql services to clear.

We log out tof webadmin when carrying out the restart but did not restart our browsers (Firefox).  Wondering if upon login, it does not clear some sort of cache, leaving further browsing in an indetermined state.

(We'll be updating to 3.1.0-50 which fixes the open connection issue this coming weekend.)