Pod AntiAffinity with PodAffinity + NotIn does not work

[CrimsonGlory]

certified-kubernetes-administrator/Domain 2 - Workloads & Scheduling/podaffinity-required.yaml

Line 12 in 0b014fa

operator: NotIn

Context: video 47 from CKA course (Pod Affinity and Pod Anti-Affinity). Code use to produce pod antiaffinity.

Using this code (PodAffinity + NotIn to achieve PodAntiAffinity) does not work for me. The pod is scheduled on the same node as frontend app. Label app=frontend is present on frontend app.

$ kubectl get pods -o wide --show-labels
NAME       READY   STATUS    RESTARTS   AGE   IP            NODE                   NOMINATED NODE   READINESS GATES   LABELS
frontend   1/1     Running   0          86m   10.244.1.47   pool-axe79dmr0-8or0s   <none>           <none>            app=frontend,run=frontend
$ md5sum podaffinity-required.yaml 
8e217e52a432bee34d1aab01226d393c  podaffinity-required.yaml
$ kubectl apply -f podaffinity-required.yaml 
pod/kplabs-pod-affinity created
$ kubectl get pods -o wide              
NAME                  READY   STATUS    RESTARTS   AGE   IP            NODE                   NOMINATED NODE   READINESS GATES
frontend              1/1     Running   0          87m   10.244.1.47   pool-axe79dmr0-8or0s   <none>           <none>
kplabs-pod-affinity   1/1     Running   0          11s   10.244.1.3    pool-axe79dmr0-8or0s   <none>           <none>

But using PodAntiAffinity + In does the trick:

$ kubectl delete -f podaffinity-required.yaml
pod "kplabs-pod-affinity" deleted
$ cat podantiaffinity-required.yaml 
apiVersion: v1
kind: Pod
metadata:
  name: kplabs-pod-affinity
spec:
  affinity:
    podAntiAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
      - labelSelector:
          matchExpressions:
          - key: app
            operator: In
            values:
            - frontend
        topologyKey: kubernetes.io/hostname
  containers:
  - name: pod-affinity
    image: nginx
$ kubectl apply -f podantiaffinity-required.yaml 
pod/kplabs-pod-affinity created
$ kubectl get pods -o wide
NAME                  READY   STATUS    RESTARTS   AGE   IP             NODE                   NOMINATED NODE   READINESS GATES
frontend              1/1     Running   0          89m   10.244.1.47    pool-axe79dmr0-8or0s   <none>           <none>
kplabs-pod-affinity   1/1     Running   0          5s    10.244.0.191   worker-node-8ouvr      <none>           <none>

Other user also had this problem. I'm using the Digital Ocean managed kuberentes created two days ago.