Looking for Best Practices Upgrading/Downgrading Dynatrace Operator in k8s

[StrangerThing]

We're trying to standardize and build pipelines for our Dynatrace deployments inside Kubernetes, whether it be Azure Kubernetes Service (AKS) or on-prem with OpenShift. I haven't found too much on the process and best practices for upgrading, but am also looking for proper rollback procedures if we need to downgrade for some reason. We've tried downgrading in a test cluster and saw some issues, so I figured before we got too deep into the weeds I could ask here to see if anyone has a good guide on the upgrade/downgrade process(es). Didn't find too much in the docs for this. Thanks in advance!

[gopher · Accepted solution]

Hi @StrangerThing 

For the general approach of making things easiest to deploy, I would suggest using the OCI image approach. 
This is compatible with most of your CD tools, and will make life easier when it comes to upgrade / lifecycle.

Downgrade:  uninstall, bounce all nodes, reinstall, bounce all nodes.    Painful, but less painful than the alternative of troubleshooting afterwards.

Upgrade: OCI version update, let helm handle it.   if you need to update the CRD API version (only if you need new features), then you can manually push this in after the OCI Image is updated.   

Fresh install: OCI deployment and CRD post install. 

This is the command structure: Use a private registry — Dynatrace Docs