Errors to query/control power are not propagated to the UI

[Björn Tillenius]

Currently, power checking and control for IPMI are broken (bug 2042429). Yet the machine listing doesn't show any power related errors at all.

There are also no mention of any failures in the event log.

[Björn Tillenius (bjornt)]

There's one power error in the listing, but it's from 2 weeks ago.

[Jacopo Rota (r00ta)]

Comparison between 3.4 and 3.5:

- 3.4:
  - we fail the power action after ~4 minutes
  - we transition the machine to `Failed XXX` depending on the original action
  - we mark the tests as Failed

- 3.5:
  - we log the first error after ~2 minutes and we keep trying for ~10 minutes. After that we log `Aborting COMMISSIONING and reverting to NEW. Unable to power control the node. Please check power credentials.`
  - we transition the machine back to its original status
  - we mark the tests as Aborted

The behaviour of 3.5 looks reasonable to me. I'm marking this bug as invalid but feel free to reopen if you think 3.5 should do something different.

[Björn Tillenius (bjornt)]

I still feel that this is something that we need to address, but we don't have to address it for 3.5.0.

Already, 4 minutes is quite a long time for any feedback (assuming that you don't actively look at the event log), and increasing that to 10 minutes is going in the wrong direction.