-
Scale the number of replicas of your Hello World service by running the following commands:
kubectl get deployment NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE helloworld-service-v1 1 1 1 1 1m
kubectl scale deployment helloworld-service-v1 --replicas=4
kubectl get deployment NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE helloworld-service-v1 4 4 4 4 1m
kubectl get pods NAME READY STATUS RESTARTS AGE helloworld-service-v1-... 1/1 Running 0 1m helloworld-service-v1-... 1/1 Running 0 1m helloworld-service-v1-... 1/1 Running 0 1m helloworld-service-v1-... 1/1 Running 0 2m
-
Try scaling out further.
kubectl scale deployment helloworld-service-v1 --replicas=17
If you look at the pod status, some of the pods will show a Pending state. That is because we have cordoned one worker node, leaving only two available for scheduling. And the underlying infrastructure has run out of capacity to run the containers with the requested resources.
-
Pick a pod name that has a
Pendingstate to confirm the lack of resources in the detailed status.kubectl describe pod helloworld-service... -
Uncordon the worker to be available for scheduling.
a. Open another terminal, do the
export KUBECONFIG ...command in Lab 1 and run:``` kubectl get po -w -o wide ``` This will monitor the recovering process.b. Go back the working terminal. Get the name of the worker that has status "not-ready".
``` kubectl get nodes ```c. Undordon the worker to make it available for scheduling.
``` kubectl uncordon [name] ``` -
Verify all three workers are available and the pending pod is rescheduled.
kubectl get nodes kubectl get pods -o wide
Clean up by deleting all deployed services from the cluster.
kubectl delete all --all