Skip to content

Latest commit

 

History

History

rolling-update

README.md

Rolling Update

A ramped deployment updates pods in a rolling update fashion, a secondary ReplicaSet is created with the new version of the application, then the number of replicas of the old version is decreased and the new version is increased until the correct number of replicas is reached.

rolling-update

Create namespace

kubectl create namespace rolling-update

Deploy App

cd ~/Workshop-K8S/k8s/rolling-update
kubectl -n rolling-update apply -f hello-world.yaml
kubectl -n rolling-update rollout status deployment hello-world -w
kubectl -n rolling-update get events
kubectl -n rolling-update get pods -o wide

Create Service (ClusterIP, NodePort, LoadBalancer)

kubectl -n rolling-update apply -f hello-world-service.yaml
kubectl -n rolling-update get service

Create Ingress

kubectl -n rolling-update apply -f hello-world-ingress.yaml

Test

Note down the Public IP adres of the load balancer.

kubectl get svc -A

Test with curl:

curl http://[public ip adress loadbalancer]/api

Or in browser

http://[PUBLIC IP loadbalancer]

Self healing

Find the right pod name in the rolling-update namespace and delete the pod:

kubectl -n rolling-update get pods
kubectl -n rolling-update delete pod <hello-world pod name>

See what happened and if a new pod is started:

kubectl -n rolling-update get events
kubectl -n rolling-update get pods

Healtcheck

Check health check:

curl [public ip adress load balancer]/health

Marks server as down:

curl -X POST [public ip adress load balancer]/down

Check health check:

curl [public ip adress load balancer]/health

Check that health check is failed:

kubectl -n rolling-update get events -w
kubectl -n rolling-update get pods -o wide

Update & Rollback

Update App to use new version

kubectl -n rolling-update set image deployments/hello-world hello-world=avthart/hello-app:1.0.1

This results in a new pod with status broken version (status CrashLoopBackOff). The new version is broken :-(

Roll out status

kubectl -n rolling-update rollout status deployments/hello-world
kubectl -n rolling-update get pods
kubectl -n rolling-update describe pod hello-world

Which version is running? Stil alive?

curl [public ip adress load balancer]/api

Rollback App

kubectl -n rolling-update rollout history deployments/hello-world
kubectl -n rolling-update rollout history deployments/hello-world --revision=2
kubectl -n rolling-update rollout undo deployments/hello-world

Update to working version:

kubectl -n rolling-update set image deployments/hello-world hello-world=avthart/hello-app:1.0.3
kubectl -n rolling-update rollout status deployments/hello-world

Which version is running? Was there any downtime?

curl [public ip adress load balancer]/api

Scaling

Scale app to 3 instances

kubectl -n rolling-update scale --replicas=15 deployment/hello-world && \
kubectl -n rolling-update get pods -w

Check that request is handled by all instances:

while sleep 0.5; do curl [public ip adress load balancer]/api; done

Stop with CONTROL-C

Clean up

kubectl delete ns rolling-update