Subscription stuck at Upgrading

[operator-framework]

Subscription stuck at Upgrading

[njhale]

@eivantsov When you first create a Subscription, an InstallPlan resource should be generated. You should be able run kubectl -n openshift-operators get installplans to see if any have been created. You can also check if any CSVs have been created in openshift-operators by running kubectl -n openshift-operators get csvs.

@njhale thanks

somehow, I have got 1000s of install plans - oc delete --all installplans -n=openshift-operators is deleting them 10 mins already :) Not sure how I ended up with that many installplans objects. Re-provisioning a cluster sounds like the best way for me to proceed.

@njhale now on a fresh cluster, it worked for my custom operator.

Now, I have installed Prometheus operator, and this time status is unknown. No installplan is created at all. There is no ClusterServiceVersion too.

Not sure if it's relevant but logs from pod catalog-operator-859bf87cd5-cm5zp say:

E0207 17:47:56.669006       1 queueinformer_operator.go:155] Sync "openshift-operators" failed: {che alpha {installed-custom-global-operators openshift-operators}} not found: CatalogSource {installed-custom-global-operators openshift-operators} not found

And indeed:

eugene@ivantsoft ~/go/src/github.com/operator-framework/operator-marketplace/scripts (master) $ oc get CatalogSource -n=openshift-operators
NAME                                   AGE
installed-community-global-operators   15m

Che is a custom operator that I have added by creating a new OperatorSource:

apiVersion: marketplace.redhat.com/v1alpha1
kind: OperatorSource
metadata:
  name: che-operator
  namespace: openshift-marketplace
spec:
  type: appregistry
  endpoint: https://quay.io/cnr
  registryNamespace: eivantsov
  displayName: "Community Operators"
  publisher: "Red Hat"

And the operator was successfully installed and works as expected.

I did not delete or modify any CRs. Just added a custom operator with the above yaml.

[njhale]

@njhale thanks

somehow, I have got 1000s of install plans - oc delete --all installplans -n=openshift-operators is deleting them 10 mins already :) Not sure how I ended up with that many installplans objects. Re-provisioning a cluster sounds like the best way for me to proceed.

We merged a fix for this in OLM though I'm not sure it's made its way to the installer yet. You could run one of the oc admin release commands to figure out what OLM images are included.

Now, I have installed Prometheus operator, and this time status is unknown. No installplan is created at all. There is no ClusterServiceVersion too.

Could you run oc get packagemanifests -n openshift-operators -o yaml ?

[ecordell]

This bug has been fixed and is available in all of the latest releases (upstream and ocp). Closing, but feel free to re-open if there's still an issue somewhere.