Description of problem: Upgrading Pipelines Operator from v1.2.3 to v1.4.1(stable channel) fails with below error messages: {},"logging.googleapis.com/sourceLocation":{"file":"/opt/app- root/src/go/src/github.com/tektoncd/operator/vendor/knative.dev/pkg/client/injection/kube/reconciler/core/v1/namespace/controller.go","line":"119","function":"knative.dev/pkg/client/injection/kube/reconciler/core/v1/namespace.createRecorder"}} panic: The environment variable "METRICS_DOMAIN" is not set Version-Release number of selected component (if applicable): 1.4.1 How reproducible: Always Steps to Reproduce: 1. Install Pipelines Operator v1.2.3 (stable channel) 2. Upgrade the operator to v1.4.1 3. Actual results: It fails with error Expected results: It should be updated to latest version as selected Additional info: Manually adding the the METRICS_DOMAIN environment variable to the pipeline operator YAML appears to resolve the issue and allows the pipeline operator to deploy successfully ---------------------- spec: containers: - command: - openshift-pipelines-operator env: - name: METRICS_DOMAIN value: tekton.dev/triggers ---------------
As commented in https://bugzilla.redhat.com/show_bug.cgi?id=1991637 I think the workaround here (before we do a bugfix release) is to add the following to the `openshift-pipelines-operator` deployment in `openshift-operators` namespace: ``` env: - name: METRICS_DOMAIN value: tekton.dev/operator ```
Based on https://github.com/operator-framework/operator-lifecycle-manager/blob/master/doc/design/subscription-config.md#configuring-operators-deployed-by-olm I think the "cleanest" workaround here would be to edit the OpenShift Pipelines operator with the following (in spec) ``` apiVersion: operators.coreos.com/v1alpha1 kind: Subscription metadata: # […] name: openshift-pipeline-operator namespace: openshift-operators # […] spec: # […] config: env: - name: METRICS_DOMAIN value: tekton.dev/operator name: openshift-pipelines-operator-rh # […] ```
From customer : I applied the change and delete the pods controlled by deployment “openshift-pipelines-operator” in “openshift-operators” namespace this action did resolve the crash I reported earlier. Also yes we are running “OpenShift-Serverless” on test/validation cluster as well.
Is there any update on this ticket? The cu is looking for a update.
@dbasant is the customer still using OpenShift Pipelines 1.4.1. as @nchoudhu mentioned the suggested workaround has got upgrade moving. Is that good enough. In addition, I shall discuss this with my team and get an update.
@skanniha Thank you. I shall close this bug.