3scale operator backend-listener restart multiple times

Setup:

• OCP cluster-version: 4.12, 4.13
• 3scale-operator-bundle-container:
  • v4.12: registry-proxy.engineering.redhat.com/rh-osbs/iib:490901
  • v4.13: registry-proxy.engineering.redhat.com/rh-osbs/iib:490917
• 3scale-apicast-operator-bundle-container:
  • v4.12: registry-proxy.engineering.redhat.com/rh-osbs/iib:490924
  • v4.13: registry-proxy.engineering.redhat.com/rh-osbs/iib:490948
• 3scale version: 2.13.3
• Reproduction steps:
  • Deploy 3scale operator and apicast operator on OCP cluster
  • Check deployment status using $oc get pods -n 3scale

Expected result:

• backend-listener pod status should be in running state

• It is observed that backend-listener pod is restarting multiple times

Further Observation:

1. Using $oc describe pod backend_listener_pod -n 3scale, below messages are observed.

 Last State:     Terminated

      Reason:       OOMKilled

      Exit Code:    137

      Started:      Mon, 15 May 2023 03:58:06 +0000

      Finished:     Mon, 15 May 2023 06:18:25 +0000

    Ready:          True

    Restart Count:  57

    Limits:

      cpu:     1

      memory:  700Mi

    Requests:

      cpu:      500m

      memory:   550Mi

2. Running backend-listener pod also shows last state where it has terminated and restarted, reason as OOMKilled (exit code: 137), pod used more memory than allowed.

Workaround:

• Updated spec field for APIManager for ResourceRequirementsEnabled=false, issue is not observed further.
• Reference: https://github.com/3scale/3scale-operator/blob/master/doc/apimanager-reference.md