large_ops scenario: actual timeout may be larger than specified one
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
tempest |
In Progress
|
Undecided
|
Sean Dague |
Bug Description
def _wait_for_
for server in self.servers:
The chunk of code above iterates over servers and for each server checks whether it becomes active before the timeout expires.
With a default timeout of 196 seconds, this means that if the first server takes 10 seconds to boot, then the function will wait 196 more seconds for the second server, bringing the total timeout to 206, and so on.
Potential fix approaches:
- have status_timeout return the elapsed time and decrease the timeout for the next check
CONS: Non-negligible refactoring
- do the check in parallel
CONS: consistent number of threads spawn; status_timeout will query nova server every build_interval seconds for each thread.
Changed in tempest: | |
assignee: | Salvatore Orlando (salvatore-orlando) → Sean Dague (sdague) |
Fix proposed to branch: master /review. openstack. org/56242
Review: https:/