member subnet handling bug in amphorav1
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
octavia |
Confirmed
|
Medium
|
Gregory Thiemonge |
Bug Description
While working on
815313: Add member test with multiple subnets on same network |
https:/
it appears that the new test fails on stable/wallaby
After some investigations, it fails only in amphorav1, but not in amphorav2.
The cause of the failure is that a subnet is not unplugged from the amphora when a member is deleted.
It happens because the provisioning_status of the member is not correct in the octavia objects:
The member object is pulled from the loadbalancer DB object that is injected by taskflow.
The provisioning_status of the member is updated by a previous task but the loadbalancer still contains the old object.
We need to reload the loadbalancer object in CalculateAmphor
Changed in octavia: | |
assignee: | nobody → Gregory Thiemonge (gthiemonge) |
importance: | Undecided → Medium |
status: | New → Confirmed |
Patch in 2023.1: https:/ /review. opendev. org/c/openstack /octavia/ +/888669