Overload manager action to fail readiness check

[inssein]

The recommendation in https://www.envoyproxy.io/docs/envoy/latest/configuration/best_practices/edge#best-practices-edge for configuring an overload manager looks like so:

  - name: "envoy.overload_actions.shrink_heap"
    triggers:
    - name: "envoy.resource_monitors.fixed_heap"
      threshold:
        value: 0.95
  - name: "envoy.overload_actions.stop_accepting_requests"
    triggers:
    - name: "envoy.resource_monitors.fixed_heap"
      threshold:
        value: 0.98

For folks who are running envoy in a cloud native environment, would it not be better to provide an action that will start failing the healtcheck, so the /ready endpoint starts failing? In this scenario, traffic won't be sent to the pod until the service is healthy again.

[yanavlasov]

I've added help_wanted to find someone who wants to implement this feature.

[inssein]

thanks @yanavlasov - any thoughts on whether this makes sense? I was kind of feeling like it doesn't exist for a reason.

If someone uses the readiness check for a liveness check, it could be bad news as that instance would be removed from rotation, but as long as its readiness only, seems fine to me.

[yanavlasov]

Your request seemed reasonable to me. It needs someone motivated enough to get implemented.

[yash97]

Hi @yanavlasov can i work on this? I am motivated enough ;).

[vikaschoudhary16]

@inssein what you think could be triggers for this new action?