Poor VMware driver exception handling in compute manager

[Édouard Thuleau]

Another opened bug (https://bugs.launchpad.net/nova/+bug/788550) reveals new bug.
The VMware driver raise exception when we try to pause an instance.
After pause request, the instance state passed to 'pausing' but it stays in this state indefinitely. It didn't come back to the 'running' state.

While with libvirt driver, if we try to pause an instance, the driver raise an APIError exception (it's not implemented). The instance passed to 'pausing' but few second after it came back to 'running' state and log files says:

2011-05-26 15:43:24,014 INFO nova.compute.manager [-] Updating host status
2011-05-26 15:43:24,047 INFO nova.compute.manager [-] DB/VM state mismatch. Changing state from '0' to '1'

[vivek.ys (vivekys)]

Once the bug
https://bugs.launchpad.net/bugs/790174 gets resolved then adding pause and resume functionality in vmops.py of vmwareapi will resolve this issue. Am I right?

[Édouard Thuleau (ethuleau)]

Yes it will resolve the pause and resume cases. But if we try to unpause or resume a running instance, the instance pass to 'unpausing' or 'resuming' state and stay in this state definitively.

No process synchronize DB and compute driver. With the libvirt driver, this process exist and a log explains that :

2011-05-31 11:30:18,675 INFO nova.compute.manager [-] Updating host status
2011-05-31 11:30:18,715 INFO nova.compute.manager [-] DB/VM state mismatch. Changing state from '0' to '1'

[Thierry Carrez (ttx)]

@Vivek.ys: any chance you can address the comments on your merge proposal and propose for merging again ?

[Mark McLoughlin (markmc)]

This has gotten very stale, so I think it makes sense to de-assign

Of course, please feel free to pick it up again Vivek

[Sean Dague (sdague)]

it's been a year since there was active work on the bug, and 6 months since the last comment. Closing out. Please reopen if this persists