Advanced Pod operations

Advanced operations with pods

Adding resource requests and limits

Check the list of namespaces:

kubectl get ns

Create a new benchmark Namespace for the Pod instance:

kubectl create ns benchmark

Create a Pod template manifest with a stress tool included:

kubectl run -n benchmark stress --image=vish/stress --dry-run=client -o yaml \
                                -- -cpus 1 -mem-total 350Mi -mem-alloc-size 100Mi -mem-alloc-sleep 5s > pod-stress.yaml

Edit the pod-stress.yaml manifest to get:

apiVersion: v1
kind: Pod
metadata:
  name: stress
  namespace: benchmark
spec:
  containers:
  - name: stress
    args: 
    - -cpus
    - "1"
    - -mem-total
    - "350Mi"
    - -mem-alloc-size
    - "100Mi" 
    - -mem-alloc-sleep
    - "5s"
    image: vish/stress
    resources:
      requests:
        cpu: "500m"
        memory: "100Mi"
      limits:
        cpu: "1"
        memory: "300Mi"

Apply the Pod manifest and create it:

kubectl apply -f pod-stress.yaml

View detailed information about the Pod:

kubectl get pod -n benchmark stress -o wide

Check resource usage after a few seconds:

kubectl top pods -n benchmark

Check the Pod status:

kubectl get pods -n benchmark

Output:

NAME     READY   STATUS    RESTARTS      AGE
stress   1/1     Running   1 (25s ago)   95s

Now, check the Pod details and try to find out why it was restarted:

kubectl describe pod -n benchmark stress

Output:

Name:         stress
Namespace:    benchmark

...

    State:          Running
      Started:      Wed, 28 Dec 2022 12:12:52 +0000
    Last State:     Terminated
      Reason:       OOMKilled
      Exit Code:    1
      Started:      Wed, 28 Dec 2022 12:11:42 +0000
      Finished:     Wed, 28 Dec 2022 12:12:50 +0000

...

Clean up the environment:

kubectl delete ns benchmark

Add a named port to Pod

Add a named port for the container in Pod manifests:

Create the myapp namespace:

kubectl create ns myapp

Create the myapp Pod manifest:

pod-myapp.yaml

apiVersion: v1
kind: Pod
metadata:
  name: myapp
  namespace: myapp
spec:
  containers:
  - name: myapp
    image: ghcr.io/go4clouds/myapp:v1.0
    resources:
      limits:
        cpu: "500m"
        memory: "200Mi"
      requests:
        cpu: "150m"
        memory: "100Mi"
    ports:
    - name: http
      containerPort: 8081
      protocol: TCP

Apply the myapp Pod manifest to the cluster:

kubectl apply -f pod-myapp.yaml

Check the Pod status and wait until the status changes to Running:

kubectl get pod -n myapp myapp -o wide

Check the logs of the myapp Pod container:

kubectl logs -n myapp myapp

Open a tunnel connection to the myapp pod container:

kubectl port-forward -n myapp pod/myapp 8081:8081

Test the connection in another terminal:

curl http://127.0.0.1:8081

Add liveness and readiness probes

We would like to extend the previous example and add liveness and readiness probes to it.

Update the myapp Pod manifest example with a liveness probe:

pod-myapp.yaml

apiVersion: v1
kind: Pod
metadata:
  name: myapp
  namespace: myapp
spec:
  containers:
  - name: myapp
    image: ghcr.io/go4clouds/myapp:v1.0
    resources:
      limits:
        cpu: "500m"
        memory: "200Mi"
      requests:
        cpu: "150m"
        memory: "100Mi"
    ports:
    - name: http
      containerPort: 8081
      protocol: TCP
    livenessProbe:
      httpGet:
          path: /
          port: 8080
      failureThreshold: 1
      initialDelaySeconds: 3
      periodSeconds: 5

Save the changes and clean up the already running Pod instance:

kubectl delete -f pod-myapp.yaml

Apply the changes and start a new Pod:

kubectl apply -f pod-myapp.yaml

Wait a few seconds and check the Pod status:

kubectl describe pod -n myapp myapp

Output:

...

Events:
  Type     Reason     Age                From               Message
  ----     ------     ----               ----               -------
  Normal   Scheduled  52s                default-scheduler  Successfully assigned myapp/myapp to worker1
  Normal   Pulled     17s (x2 over 51s)  kubelet            Container image "ghcr.io/go4clouds/myapp:v1.0" already present on machine
  Normal   Created    17s (x2 over 51s)  kubelet            Created container myapp
  Normal   Started    17s (x2 over 51s)  kubelet            Started container myapp
  Warning  Unhealthy  12s (x2 over 47s)  kubelet            Liveness probe failed: Get "http://192.168.235.155:8080/": dial tcp 192.168.235.155:8080: connect: connection refused
  Normal   Killing    12s (x2 over 47s)  kubelet            Container myapp failed liveness probe, will be restarted

The pod was restarted because the check couldn't succeed since the application is listening on port 8081.

Let's fix the configuration:

pod-myapp.yaml

apiVersion: v1
kind: Pod
metadata:
  name: myapp
  namespace: myapp
spec:
  containers:
  - name: myapp
    image: ghcr.io/go4clouds/myapp:v1.0
    resources:
      limits:
        cpu: "500m"
        memory: "200Mi"
      requests:
        cpu: "150m"
        memory: "100Mi"
    ports:
    - name: http
      containerPort: 8081
      protocol: TCP
    livenessProbe:
      httpGet:
          path: /
          port: http
      failureThreshold: 1
      initialDelaySeconds: 3
      periodSeconds: 5

Save the changes and clean up the already running Pod instance:

kubectl delete -f pod-myapp.yaml

Apply the changes and start a new Pod:

kubectl apply -f pod-myapp.yaml

Now check the Pod status:

kubectl get pods -n myapp
kubectl describe pod -n myapp myapp

Clean up the myapp Pod:

kubectl delete ns myapp