secondary-scheduler pod gets OOMKilled on large clusters:

$ oc get po secondary-scheduler-7d9bf885fc-xqn2z -o yaml
[...]
  containerStatuses:
  - containerID: cri-o://e2675fbeef0f7f3993bbb2ca63d815d19538434bc2c80e61e2c49d12ba894b0a
    image: registry.redhat.io/openshift4/noderesourcetopology-scheduler-rhel9@sha256:a633cf3d9ae757c9316eae69b1e41cd27c4c3c114cc979ae973201a4328532ed
    imageID: registry.redhat.io/openshift4/noderesourcetopology-scheduler-rhel9@sha256:a633cf3d9ae757c9316eae69b1e41cd27c4c3c114cc979ae973201a4328532ed
    lastState:
      terminated:
        containerID: cri-o://87481f8a582a7360d4b370975f397558499c185a5b00d4c34ded03ed1f702539
        exitCode: 137
        finishedAt: "2025-10-08T05:19:02Z"
        reason: OOMKilled
        startedAt: "2025-10-08T05:18:56Z"
    name: secondary-scheduler
    ready: true
    restartCount: 262
    started: true
    state:
      running:
        startedAt: "2025-10-08T05:19:44Z"

$ oc get deploy -n openshift-numaresources secondary-scheduler -o jsonpath='{.spec.template.spec.containers[].resources}' | jq
{
  "limits": {
    "cpu": "600m",
    "memory": "1200Mi"
  },
  "requests": {
    "cpu": "600m",
    "memory": "1200Mi"
  }
}

    numaresources-operator.v4.16.3

    Observable on a large cluster, affected one has 90 nodes: 3 master + 18 storage + 69 worker

    1.
    2.
    3.
    

secondary-scheduler pod gets OOMKilled

secondary pod should be stable

    It would be a good idea to introduce the ability to tune limits and requests of this pod, perhaps in the NUMAResourcesScheduler CR