session-65.txt

Cluster creation and upgrade
-----------------------------
admin activity

aws eks update-kubeconfig --region us-east-1 --name expense-dev

cluster is running, few application are running inside

Cluster upgrade is planned
--------------------------
no downtime to the existing apps but no new release and no new deployments in the upgrade time...
communications sent about cluster upgrade activity is started...

1. first we need to get another node group called green.
2. taint the green nodes, so that they should not get any pods scheduled
kubectl taint nodes ip-10-0-11-151.ec2.internal project=expense:NoSchedule
3. now upgrade your control plane, do it from console

4. upgrade green node group also to 1.30

5. shift the workloads from 1.29 node group to 1.30 nodegroup

6. cordon blue nodes

7. untaint green nodes

8. drain blue nodes

inform all stake holders, application teams. perform sanity testing

close the activity