Scheduling Pods on Kubernetes Control plane (Master) Nodes

By default, your Kubernetes Cluster will not schedule pods on the control-plane node for security reasons. It is recommended you keep it this way, but for test environments you may want to schedule Pods on control-plane node to maximize resource usage.

If you want to be able to schedule pods on the Kubernetes control-plane node, you need to remove a taint on the master nodes.

kubectl taint nodes --all node-role.kubernetes.io/master-
kubectl taint nodes --all  node-role.kubernetes.io/control-plane-

The output will look something like:

node/k8smaster01.computingforgeeks.com untainted
taint "node-role.kubernetes.io/master" not found
taint "node-role.kubernetes.io/master" not found
taint "node-role.kubernetes.io/master" not found

This will remove the node-role.kubernetes.io/master taint from any nodes that have it, including the control-plane node, meaning that the scheduler will then be able to schedule pods everywhere.

For single node the command to use is:

kubectl taint nodes  node-role.kubernetes.io/master-

Testing Pod Scheduling on Kubernetes Control plane node(s)

I have a cluster with three worker nodes and one control plane node.

$ kubectl get nodes
NAME                                STATUS   ROLES    AGE   VERSION
k8smaster01.computingpost.com   Ready    master   12h   v1.24.3
k8snode01.computingpost.com     Ready       12h   v1.24.3
k8snode02.computingpost.com     Ready       12h   v1.24.3
k8snode03.computingpost.com     Ready       9h    v1.24.3

Create a demo namespace:

kubectl create namespace demo

Will create a deployment with 5 replicas.

vim nginx-deployment.yaml

It has the data below:

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx
  namespace: demo
  labels:
    app: nginx
    color: green
spec:
  replicas: 5
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
        color: green
    spec:
      containers:
        - name: nginx
          image: nginx:latest
          imagePullPolicy: IfNotPresent
          ports:
            - name: http
              protocol: TCP
              containerPort: 80
          resources:
            limits:
              cpu: "200m"
              memory: "256Mi"
            requests:
              cpu: 100m
              memory: 128Mi
---
apiVersion: v1
kind: Service
metadata:
  annotations:
  name: nginx-demo-service
  namespace: demo
spec:
  ports:
    - port: 80
      targetPort: 80
      protocol: TCP
  selector:
    app: nginx
  sessionAffinity: None
  type: NodePort

Apply manifest:

kubectl apply -f nginx-deployment.yaml

Check if a pod is scheduled to the control node plane.

$ kubectl get pods -n demo -o wide
NAME                     READY   STATUS    RESTARTS   AGE   IP                NODE                                NOMINATED NODE   READINESS GATES
nginx-675bf5bc87-666jg   1/1     Running   0          17m   192.168.213.131   k8snode01.computingpost.com               
nginx-675bf5bc87-mc6px   1/1     Running   0          17m   192.168.94.13     k8smaster01.computingpost.com             
nginx-675bf5bc87-v5q87   1/1     Running   0          17m   192.168.144.129   k8snode03.computingpost.com                
nginx-675bf5bc87-vctqm   1/1     Running   0          17m   192.168.101.195   k8snode02.computingpost.com                
nginx-675bf5bc87-w5pmh   1/1     Running   0          17m   192.168.213.130   k8snode01.computingpost.com               

We can see there is a pod in master node. Confirm service is live.

$ kubectl get svc -n demo
NAME            TYPE       CLUSTER-IP     EXTERNAL-IP   PORT(S)        AGE
nginx-service   NodePort   10.96.184.67           80:31098/TCP   21m

Since we’re using NodePort, we should be able to access the service on any cluster node IP on port 31098.

We can now clean demo objects.

$ kubectl delete -f nginx-deployment.yaml
deployment.apps "nginx" deleted
service "nginx-service" deleted

$ kubectl get pods,svc -n demo
No resources found in demo namespace.

That’s all on how to Schedule Pods on Kubernetes Control plane Node.