Update tutorial to use GKE autopilot

[meteatamel]

No description provided.

[meteatamel]

Autopilot does not seem to work with Istio/Knative Serving in my testing

[aatarasoff]

@meteatamel Hi!

Do you deep dive into the problem? I faced the same behavior and thought it could be a VPA issue.

[meteatamel]

@aatarasoff No, I didn't deep dive. Any ideas what could be missing?

[aatarasoff]

@meteatamel I think that it could be related to the Vertical Pod Autoscaler because Autopilot changes CPU and Memory requests and limits according to its policies, and it could blow Knative mind up. It's like a race condition between two autoscalers (Knative, as I know, has one). I tried to turn of VPA (see code below) but it did not help.

---
apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: hello-world-turn-off-vpa
spec:
  targetRef:
    apiVersion: "apps/v1"
    kind:       Deployment
    name:       hello-world-deployment
  updatePolicy:
    updateMode: "Off"

[meteatamel]

I'll reopen this, in case I or someone else might look into this again.