When on projects is used the openshift.io/node-selector option in the annotation pointing to the default "node-role.kubernetes.io/worker=" key and a deployment with a running replicaset/PODS is set to use a different role then the default one the scheduler enter in an infinite loop of POD creation.

Tested on 4.11

1. Create a project via [oc create/apply] with annotation: openshift.io/node-selector: node-role.kubernetes.io/worker= 
2. Create a deployment that creates a running POD 
3. Edit the deployment and add the nodeName: option pointing to a different role than the worker one

Infinite POD creation: 
❯ oc get po
NAME                                     READY   STATUS         RESTARTS   AGE
infinite-pod-creation-7458cbbd88-98zpn   1/1     Running        0          8m40s
infinite-pod-creation-7688f685c7-2grmh   0/1     NodeAffinity   0          1s
infinite-pod-creation-7688f685c7-4g7dd   0/1     NodeAffinity   0          2s
infinite-pod-creation-7688f685c7-59zr6   0/1     NodeAffinity   0          1s
infinite-pod-creation-7688f685c7-5l5xl   0/1     NodeAffinity   0          2s
infinite-pod-creation-7688f685c7-5nw22   0/1     NodeAffinity   0          2s
infinite-pod-creation-7688f685c7-5qr7z   0/1     NodeAffinity   0          1s
infinite-pod-creation-7688f685c7-5wp2q   0/1     NodeAffinity   0          2s
infinite-pod-creation-7688f685c7-6kxjg   0/1     NodeAffinity   0          1s
infinite-pod-creation-7688f685c7-74d7m   0/1     NodeAffinity   0          2s
infinite-pod-creation-7688f685c7-78hzm   0/1     NodeAffinity   0          1s 
....
...
..
.

The scheduler should be able to find a conflict and throw an error 

apiVersion: project.openshift.io/v1
kind: Project
metadata:
  annotations:
    openshift.io/description: ""
    openshift.io/display-name: ""
    openshift.io/node-selector: node-role.kubernetes.io/worker=
    openshift.io/requester: kube:admin
    openshift.io/sa.scc.mcs: s0:c29,c19
    openshift.io/sa.scc.supplemental-groups: 1000850000/10000
    openshift.io/sa.scc.uid-range: 1000850000/10000
  labels:
    kubernetes.io/metadata.name: infinite-pod-creation
    pod-security.kubernetes.io/audit: restricted
    pod-security.kubernetes.io/audit-version: v1.24
    pod-security.kubernetes.io/warn: restricted
    pod-security.kubernetes.io/warn-version: v1.24
  name: infinite-pod-creation
spec:
  finalizers:
  - kubernetes

======================================================

apiVersion: apps/v1
kind: Deployment
metadata:
  annotations:
  labels:
    app: infinite-pod-creation
    app.kubernetes.io/component: infinite-pod-creation
    app.kubernetes.io/instance: infinite-pod-creation
  name: infinite-pod-creation
spec:
  progressDeadlineSeconds: 600
  replicas: 1
  revisionHistoryLimit: 10
  selector:
    matchLabels:
      deployment: infinite-pod-creation
  strategy:
    rollingUpdate:
      maxSurge: 25%
      maxUnavailable: 25%
    type: RollingUpdate
  template:
    metadata:
      annotations:
      creationTimestamp: null
      labels:
        deployment: infinite-pod-creation
        app: infinite-pod-creation
    spec:
      topologySpreadConstraints:
      - maxSkew: 1
        topologyKey: kubernetes.io/hostname
        whenUnsatisfiable: DoNotSchedule
        labelSelector:
          matchLabels:
            app: infinite-pod-creation
      containers:
      - image: quay.io/openshift-release-dev/ocp-v4.0-art-dev@sha256:a9db83f67aa4389811bad29af878d038e18bc39f63673fe77fe30f9bf1bd97de
        imagePullPolicy: IfNotPresent
        name: infinite-pod-creation
        ports:
        - containerPort: 8080
          protocol: TCP
        - containerPort: 8888
          protocol: TCP
        resources: {}
        terminationMessagePath: /dev/termination-log
        terminationMessagePolicy: File
      dnsPolicy: ClusterFirst
      restartPolicy: Always
      schedulerName: default-scheduler
      securityContext: {}
      terminationGracePeriodSeconds: 30
 
============================================================

oc patch deployment infinite-pod-creation -p '{"spec":{"template":{"spec":{"nodeName": "$NODE-NAME-DIFFERENT-FROM-WORKER-ROLE"}}}}'