Upgrading of dynatrace operator, any gotchas / feedback

[sivart_89]

We are planning to upgrade the dynatrace operator to v0.12.1 (coming from 0.9.0) because it is the recommended version based off the kubernetes version we are on (1.21), per the doc below. This is the first I am doing it and was wondering if there are any things to look out for here. Will any monitoring be impacted here?

https://docs.dynatrace.com/docs/shortlink/support-model-k8s-ocp

Looking more into the doc above I see the below noted for operator version 0.12.1. So should I actually be upgrading to operator version 0.13.0 instead? I would prefer not to go to 0.13.0 and implement the workaround steps but if that is the preferred version based off the kubernetes version of 1.21 then I don't mind. However, the doc to me is still reading that 0.12.1 is the preferred version.

A new Go version used in Dynatrace Operator is incompatible with the CRI-O version of OpenShift 4.8 and Kubernetes 1.21. See the required manual workaround in Dynatrace Operator release notes version 0.13.0.

[marina_pollehn · Accepted solution]

I would make sure that your operator matches your CSI driver manifest version. We had some issues when there was an auto update and the operator and CSI driver manifest version did not align anymore. It was not allowing the pods to start which obviously impacted performance. Now we use version 1.0.0 for both and the problems disappeared (before the operator was 1.0.0 and the CSI driver was 0.15.0 - so a mismatch due to the automatic update).

    labels:
      app.kubernetes.io/component: operator
      app.kubernetes.io/name: dynatrace-operator
      app.kubernetes.io/version: 1.0.0

        labels:
          app.kubernetes.io/component: csi-driver
          app.kubernetes.io/name: dynatrace-operator
          app.kubernetes.io/version: 1.0.0   

Then again, just my personal experience with two clusters :).