-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
4.20
Description of problem:
Some of the PIS regression tests do not appear to be running as intended. For example, the `Invalid PIS leads to degraded MCN in a custom Pool` PinnedImageSet test is consistently only taking ~3 seconds to run "successfully," both when run in isolation locally and in recent payload test runs. This is logically not enough time for the steps of the test, which, at a high level, include creating a custom MCP, adding a node to it, applying a PIS, waiting for the MCN degraded PIS condition to be true, and then returning the cluster to its pre-testing state. Similar behavior has been seen in an original investigation of the `All Nodes in a Custom Pool should have the PinnedImages in PIS` test. In this test, the custom pool never becomes updated with one ready machine.
Version-Release number of selected component (if applicable):
Seen in 4.20, but I have not checked in 4.19.
How reproducible:
Always.
Steps to Reproduce:
Run theĀ `Invalid PIS leads to degraded MCN in a custom Pool` test against a 4.20 cluster and monitor the MCP throughout the run to see no nodes successfully add to the MCP.
Actual results:
So far, it seems like the test fails in the following steps:
- When a node is added to the custom MCP, there is no wait to make sure the node successfully adds.
- It does not seem like the MCN PIS degraded condition is being properly validated.
Expected results:
The PIS test should run as the test name implies.
Additional info:
N/A
- blocks
-
OCPBUGS-61173 PIS tests using custom MCPs are not running as intended
-
- Closed
-
- is cloned by
-
OCPBUGS-61173 PIS tests using custom MCPs are not running as intended
-
- Closed
-
- links to