Operator pod is killed because it is out of memory - OOMKilled

When you install IBM Cloud Pak foundational services, some operator pods crash due out-of-memory issues.

Symptom

The operator pod continuously shows the CrashLoopBackOff status.

The description of pod indicates that the Reason is OOMKilled.

1. Get the pod name and status.

   oc get pods -n ibm-common-services
   

2. Get details of the pod that is crashing.

   oc describe pod <name-of-the-pod-that-is-crashing-> -n ibm-common-services
   

   Following is a sample output:

   State:          Waiting
    Reason:       CrashLoopBackOff
   Last State:     Terminated
    Reason:       OOMKilled
    Exit Code:    137
    Started:      Wed, 23 Sep 2020 10:53:24 -0700
    Finished:     Wed, 23 Sep 2020 10:54:24 -0700
   Ready:          False
   

Cause

The configured memory limit of the operator pod is not sufficient.

Resolving the problem

Increase the resource limits of the operator pods that are crashing.

Complete these steps for each pod that is crashing:

1. Get the ClusterServiceVersion(CSV) name of the operator.

   oc -n ibm-common-services get csv | grep <operator-name>
   

2. Edit the CSV of the operator.

   oc -n ibm-common-services edit csv <operator-CSV-name>
   

3. Increase the value of the memory limit parameter in the .spec.install.spec.deployments[0].spec.template.spec.container[0].resources.limits.memory section.

   spec:
    containers:
    - command:
      - operator name
      resources:
        limits:
          cpu: 50m
          memory: 128Mi
        requests:        
          cpu: 20m
          memory: 64Mi