In Etstur, we provisioned on-premise Kubernetes Cluster using Ansible. The challenge was we want to
minimize risk of upgrading cluster. We didn't want to make rolling-upgrade and want to find another solution.
After analysing Etstur requirements, we decided to make stack level blue-green.

Client: ETSTUR
Project type: DevOps as a Service



Before requests coming to Kubernetes Cluster there was a HAProxy. In HAProxy, all kubernetes machines IP's
were defined. So any request comes to HAProxy then it was redirected to one of the Kubernetes machine.

Whenever we need to upgrade Kubernetes version, we setup second cluster using Ansible and made session draining using HAProxy. So we basically simulated stack level blue-green and we minimized the risk.
We switch traffic to new stack slowly and controlled manner.



Etstur can able to upgrade their stack without any risk. If something goes wrong, they can easily switch back to
old system.
Also they don't redirect whole traffic to new stack immediately. It happens in a controlled way.
This gives better results and more confidence