Closed ca-scribner closed 2 years ago
Not sure if this happens always or if it just happens maybe during a juju destroy-model
or a --force
on application/model removal? May need to try different scenarios
This should be investigated a bit more, we need to know if it only happens with --force
or always. Either way, if it occurs outside of --force
, it seems like a juju bug that should be filed and linked in this issue.
Closing this bug as currently juju correctly removes the mutating webhook for:
juju remove-application admission-webhook
juju remove-application admission-webhook --force
Likely the cause has been fixed in Juju since this bug was filed
Sometimes (always?) when doing
juju remove-application admission-webhook
,mutatingwebhookconfiguration
/validatingwebhookconfiguration
objects are left on the cluster. This will block all pod creation because the webhooks will fail (as the services/pods they point to are removed). This can be seen in the kubernetes event logs as timeouts on calls toadmission-webhook.kubeflow.org
:Possible resolutions: