Merge lp://qastaging/~rvb/maas/bug-1442059 into lp://qastaging/~maas-committers/maas/trunk
Status: | Merged | ||||
---|---|---|---|---|---|
Approved by: | Raphaël Badin | ||||
Approved revision: | no longer in the source branch. | ||||
Merged at revision: | 3834 | ||||
Proposed branch: | lp://qastaging/~rvb/maas/bug-1442059 | ||||
Merge into: | lp://qastaging/~maas-committers/maas/trunk | ||||
Diff against target: |
15 lines (+2/-2) 1 file modified
src/provisioningserver/rpc/power.py (+2/-2) |
||||
To merge this branch: | bzr merge lp://qastaging/~rvb/maas/bug-1442059 | ||||
Related bugs: |
|
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Gavin Panella (community) | Approve | ||
Review via email: mp+257039@code.qastaging.launchpad.net |
Commit message
Set the timeout used by twisted to cancel the power action to 5 minutes. There is already a timeout that depends on the retry policy (how power actions are retried at the Python level) and what the power template does. In the case of AMT, where the template itself does some retrying, the original timeout of 2 minutes wasn't enough (yes, AMT is that flaky).
Description of the change
Another fix would have been to watch the PXE requests and immediately mark the power action as successful when the node MAAS is trying to power up PXE boots. I decided against it because this is both racy and involved (it would tie together two fairly independent systems in MAAS: the power mechanism and the PXE request endpoint).
When we port the power templates to Python, we can rationalize how we deal with timeouts: right now each layer has its own timeout.