Is the use of patch to update the memory correct?

mostly looks right, though I don't think you should change the labels at all. that could be causing an issue. here's an example from one of our clusters:

- kind: Deployment
    patch:
      spec:
        template:
          spec:
            containers:
            - name: spicedb
              resources:
                limits:
                  cpu: "3"
                  memory: 15Gi
                requests:
                  cpu: "3"
                  memory: 15Gi

got it thanks

if removing the selector doesn't work, do you maybe have some webhooks installed (like kyverno or gatekeeper or something?)

On a separate note, we found we have VPA running in a specific env, which could be downscaling the pods, we are trying to turn that off and see if that was the cause of the issue

ah

but thank you for the pointers