Fix HA tests for scheduling change #1787
Conversation
| 11. Enable the default disk for the host node. | ||
| 12. Wait for volume2 becoming Healthy. | ||
| 13. Verify data content and r/w capability for volume2. | ||
| 12. Enable the default disk for the host node. |
There was a problem hiding this comment.
At this point, the purpose of the test is complete. We successfully recognized a failed replica during attaching and (after replica-replenishment-wait-interval) scheduled a new replica to replace it.
Previously, the test relied on replica-node-soft-anti-affinity == false to ensure the new replica could not schedule anywhere until the default disk on the test node was enabled. Once the default disk was enabled, the failed replica did not prevent scheduling to the node.
Now, the failed replica does prevent scheduling to the node. However, we can rely on replica-disk-soft-anti-affinity == false to continue to ensure it does not get scheduled until we enabled the default disk.
Longhorn 8043 Signed-off-by: Eric Weber <eric.weber@suse.com>
ejweber
force-pushed
the
8043-avoid-scheduling-a-second-replica
branch
from
4bf5ea4
to
00e60eb
Compare
|
With the changes to longhorn/longhorn-manager#2650, we likely do not need this PR. I will keep it open as a draft until that one merges. |
Which issue(s) this PR fixes:
longhorn/longhorn#8043
What this PR does / why we need it:
longhorn/longhorn-manager#2650 disallows scheduling a replica to a node, zone, or disk if it already contains a failed replica (unless the appropriate soft anti-affinity setting is enabled). This behavior requires minor changes to a couple of tests.