| 2024-03-04 08:16:49 |
Cyprian Kleist |
description |
In case of OpenStack deployment with multiple host aggregates (e.g. computes with local disk nova backend and rbd nova backend) and flavors specific for each aggregate, the `test_unshelve_to_specific_host` randomly fails. The reason is that method `get_host_other_than` pick hypervisor randomly, and sometimes it pick one from other aggregate than original. As a consequence the unshelving fails, because hypervisor does not match flavor of VM.
The expected behaviour is that other host is picked from the same aggregate and test passes every time.
Steps to reproduce:
* Create OpenStack deployment with two host aggregates (A and B) and corresponding flavors (flavA and flavB)
* Use in tempest config flavor (flavA) corresponding with one of the aggregates (A)
* run `test_unshelve_to_specific_host` to observe random failing
* disable all computes from aggregate B
* run test again to observe that now it's passing every time |
In case of OpenStack deployment with multiple host aggregates (e.g. computes with local disk nova backend and rbd nova backend) and flavors specific for each aggregate, the `test_unshelve_to_specific_host` randomly fails. The reason is that method `get_host_other_than` pick hypervisor randomly, and sometimes it pick one from other aggregate than original. As a consequence the unshelving fails, because hypervisor does not match flavor of VM.
The expected behaviour is that other host is picked from the same aggregate and test passes every time.
Steps to reproduce:
* Create OpenStack deployment with two host aggregates (A and B) and corresponding flavors (flavA and flavB)
* Use in tempest config flavor (flavA) corresponding with one of the aggregates (A)
* run `test_unshelve_to_specific_host` to observe random failing
* disable all computes from aggregate B
* run test again to observe that now it's passing every time
Versions:
* tempest 36.0.0
* openstack 2023.2 bobcat |
|
