Juju 2.9 failing to create ClusterRoleBinding
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Canonical Juju |
Fix Released
|
High
|
Yang Kelvin Liu |
Bug Description
I am deploying charms as part of this CI run:
https:/
If you examine the output from the `kubectl get pods -A -oyaml` step, you can see that a `jupyter-ui` Pod was created, with `serviceAccount: jupyter-ui`. If you examine the `kubectl get clusterroles -A -oyaml`, you can also see that a `kubeflow-
This only seems to happen sometimes, as the the other run for that commit completed successfully. You can view the errors that are being returned by downloading the `selenium-har` artifact and uploading it to http://
Changed in juju: | |
milestone: | 2.7.9 → 2.9.8 |
Changed in juju: | |
status: | In Progress → Fix Committed |
Changed in juju: | |
status: | Fix Committed → Fix Released |
In the run link there is a link to Artifacts, which has the 'introspection- reports' . Downloading that and expanding it does have a 'juju' subdirectory, which includes the juju controller debug logs.
I don't see a smoking gun in there (of some sort of api call that failed, which we aren't retrying, etc).
Looking at the log I do see a lot of warnings about: 5a6d-44c2- 8b31-752631a38a 3f: controller-0 2021-06-30 14:42:44 INFO juju.kubernetes .klog klog.go:56 apiextensions. k8s.io/ v1beta1 CustomResourceD efinition is deprecated in v1.16+, unavailable in v1.22+; use apie efinition
4abab061-
xtensions.k8s.io/v1 CustomResourceD
However https:/ /bugs.launchpad .net/juju/ +bug/1921553 seems to say that we should be trying to use apiextensions. k8s.io/ v1 first.
(I don't know that this is the cause of the issue, but it certainly would be good to get rid of the noise in the logs to help identify the actual issue.)