(Feel free to update this bug's summary to be more specific.)
Component Readiness has found a potential regression in the following test:

[sig-cli] Kubectl Port forwarding Shutdown client connection while the remote stream is writing data to the port-forward connection port-forward should keep working after detect broken connection [Suite:openshift/conformance/parallel] [Suite:k8s]

Extreme regression detected.
Fishers Exact probability of a regression: 100.00%.
Test pass rate dropped from 99.11% to 78.26%.

Sample (being evaluated) Release: 4.20
Start Time: 2025-09-08T00:00:00Z
End Time: 2025-09-15T08:00:00Z
Success Rate: 78.26%
Successes: 18
Failures: 5
Flakes: 0
Base (historical) Release: 4.19
Start Time: 2025-05-18T00:00:00Z
End Time: 2025-06-17T23:59:59Z
Success Rate: 99.11%
Successes: 101
Failures: 1
Flakes: 10

View the test details report for additional context.

This was a k8s test that we can see was flaky in past releases, but lost the ability to retry in 4.20 and thus now can show up regressed as it fails. Because we're so close to GA, the ability to retry the test was restored temporarily so a fix can be pursued in 4.21. This bug is for that long term fix.

See this link for a list of the jobs where the test fails, rosa-sts is a common one, but there are a number of other more fringe configurations where it shows up.

Once fixed, we should remove the retry for this test, being restored in the pr linked to the original issue.
Filed by: dgoodwin@redhat.com