You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The scheduled Deploy CKF to EKS workflow started failing last week. Looking into it, it looks like oidc-gatekeeper never goes to Blocked as the tests expect and thus they timeout
asyncio.exceptions.TimeoutError: Timed out waiting for model:
oidc-gatekeeper/0 [idle] active:
That being said, it looks like this wasn't always the case since the scheduled tests had been passing for about a month.
Solution
WRT to the tests, they should be modified to expect what is there in the charm's code.
WRT to oidc-gatekeeper, I think we should investigate if/when/how we introduced a change to the charm, since we can see both in the first and the last successful runs of the tests, that oidc-gatekeeper was going to
INFO juju.model:model.py:2597 Waiting for model:
oidc-gatekeeper/0 [idle] blocked: public-url config required
The text was updated successfully, but these errors were encountered:
Looking at oidc-gatekeeper charm's releases, we see that
Revision 205: Released to ckf-1.7/edge at 21:38 UTC on 23 May 2023 which has this line that checks public url and is actually a rewritten sidecar charm
Revision 250: Released to ckf-1.7/edge at 14:12 UTC on 9 Aug 2023 (with Full Changelog: rev170...rev250) which is a PodSpec charm without the check_public_url implemented.
This explains the above as the last successful run of the workflow was in August 8.
In addition, the above tests will not have to change since we will implement checking for public url in track/ckf-1.7 branch.
Problem
The scheduled Deploy CKF to EKS workflow started failing last week. Looking into it, it looks like
oidc-gatekeeper
never goes toBlocked
as the tests expect and thus they timeoutReproduce
Manually run the tests.
Cause
The tests deploy
1.7/edge
bundle while OIDC gatekeeper operator uses revision 250 for1.7/edge
. Taking a look at rev250 code, we see that the charm code has the _check_public_url method, but it seems like it is not called anywhere, unlike main that implements and calls the method. Thus, it is expected foroidc-gatekeeper
to not go toBlocked
.That being said, it looks like this wasn't always the case since the scheduled tests had been passing for about a month.
Solution
WRT to the tests, they should be modified to expect what is there in the charm's code.
WRT to oidc-gatekeeper, I think we should investigate if/when/how we introduced a change to the charm, since we can see both in the first and the last successful runs of the tests, that
oidc-gatekeeper
was going toThe text was updated successfully, but these errors were encountered: