The OCP 4.13.0 MNO (3+5) environment are not usable for the other testing. All nodes are printing in the BMC console/terminal "Memory cgroup out of memory" When login to the master node via SSH can't even execute ls -ltr command also kubectl won't work. When rebooting all the nodes from BMC the cluster started work again. This looks like memory leak issue.

Please use the deployment file that I have used and you will be able to reproduce the issue. Try to increase number of replicas to 40 and slowly nodes will start to go down

1. Create deployment.yaml filewith below content:
~~~
apiVersion: apps/v1
kind: Deployment
metadata:
  name: stress-ng-test-limit
  labels:
    k8s-app: stress-ng-test-limit
spec:
  replicas: 1
  selector:
    matchLabels:
      app: stress-ng-test-limit
  template:
    metadata:
      name: stress-ng-test-limit
      labels:
        app: stress-ng-test-limit
    spec:
      serviceAccount: stress-ng-sa
      containers:
      - name: stress-ng-test-container-limit
        image: quay.io/dmoessne/stress-ng-test:0.2
        command: [ "/bin/bash", "-c", "--" ]
        args: [ "stress-ng -c 1 --vm 32 --vm-bytes 100% --vm-method all --madvise 2"]
        #command: ["sleep", "infinity"]
        resources:
           requests:
             memory: 100Mi
           limits:
             memory: "1G"
        securityContext:
          seccompProfile:
            type: RuntimeDefault
          capabilities:
            drop:
            - ALL
          privileged: true
~~~
2. [quickcluster@upi-0 ~]$ oc create -f deployment.yaml 
deployment.apps/stress-ng-test-limit created
 3. [quickcluster@upi-0 ~]$ oc get all
NAME                                        READY   STATUS    RESTARTS   AGE
pod/stress-ng-test-limit-59c59dbf65-j6z84   1/1     Running   0          59sNAME                                   READY   UP-TO-DATE   AVAILABLE   AGE
deployment.apps/stress-ng-test-limit   1/1     1            1           59sNAME                                              DESIRED   CURRENT   READY   AGE
replicaset.apps/stress-ng-test-limit-59c59dbf65   1         1         1       59s

4. Increase replicas to 40

oc scale --replicas=40 deployment.apps/stress-ng-test-limit

5. After sometime nodes will go down , openshift web console will stop working. Operators will start behaving abnormally. 

node goes down

During stress testing , node should not go down.