ServerRescueTest may fail due to RESCUE taking too long
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack Compute (nova) |
Invalid
|
High
|
Unassigned | ||
tempest |
Invalid
|
Undecided
|
Unassigned |
Bug Description
In the grenade test [0] for a bp I'm working on, ServerRescueTestXML rescue_unrescue test failed because the VM did not get into RESCUE state in time. It seems that the test is flacky.
From the tempest log [1] I see the sequence VM ACTIVE, RESCUE issues, WAIT, timeout, DELETE VM.
From the nova cpu log [1], following request ID: req-6c20654c-
The VM is then deleted by the test and when nova tries to process the RESCUE it throws and exception as the VM is not there:
bc-b27a-
bc-b27a-
bc-b27a-
bc-b27a-
bc-b27a-
bc-b27a-
bc-b27a-
bc-b27a-
bc-b27a-
bc-b27a-
bc-b27a-
There may be a problem in nova as well, as RESCUE is held for 3 minutes waiting on a lock.
[0] https:/
[1] http://
[2] http://
description: | updated |
Changed in nova: | |
status: | Incomplete → Invalid |
This looks like the root cause is Nova exploding on the transition. I'm going to mark the Tempest side invalid.