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
Looking at the generated CSV yaml, under spec.install.spec.deployments, there are two containers: kube-rbac-proxy
and manager. The latter has a livenessProbe and a readinessProbe defined. But the paths look swapped: the /readyz path is on the livenessProbe, not on the readinessProbe as you'd expect from a path named "readyz". The path on the readinessProbe is /healthz. I'm wondering if a copy/paste typo got made here, swapping the two paths.
What did you do?
Generated a helm-based operator and looked at the yaml for the CSV.
BTW, I'm getting an image pull error on that manager container. The image field is set to just controller:latest, without a registry or org specified. When I install my operator, I see the following in the events for the pod for my operator:
Failed to pull image "controller:latest": rpc error: code = Unknown desc = Error reading manifest latest in docker.io/library/controller: errors: denied: requested access to the resource is denied unauthorized: authentication required
Do we have to have our users create an ImagePullSecret in order to get our operators working? If so, is there a way I can get my install plan to make that for the user? Thanks.
@jwalcorn the endpoint mixup was fixed in #4546. The v1.5 migration guide, when released, will discuss swapping them.
The intent is to run make bundle IMG=<registry>/<repo>/your-operator:vX.Y.Z so kustomize can substitute the default image name controller:latest for your remote image. If you still have questions about this, open another issue. Thanks!
Bug Report
Looking at the generated CSV yaml, under spec.install.spec.deployments, there are two containers:
kube-rbac-proxy
and
manager
. The latter has a livenessProbe and a readinessProbe defined. But the paths look swapped: the/readyz
path is on the livenessProbe, not on the readinessProbe as you'd expect from a path named "readyz". The path on the readinessProbe is/healthz
. I'm wondering if a copy/paste typo got made here, swapping the two paths.What did you do?
Generated a helm-based operator and looked at the yaml for the CSV.
What did you expect to see?
What did you see instead? Under which circumstances?
Environment
Operator type:
/language helm
Kubernetes cluster type:
OpenShift
$ operator-sdk version
v1.4.0
$ go version
(if language is Go)$ kubectl version
v1.17.2
Possible Solution
Swap the paths for the readinessProbe and the livenessProbe
Additional context
If the values are somehow correct, perhaps at least generate a comment explaining the confusion
The text was updated successfully, but these errors were encountered: