You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@flink.apache.org by Ritesh Nadhani <ri...@gmail.com> on 2023/02/23 16:32:06 UTC

[Flink kubernetes operator] Usage of HorizontalPodAutoscaler

Hello

I was reviewing the HPA example at:
https://github.com/apache/flink-kubernetes-operator/blob/main/examples/hpa/basic-hpa.yaml
and I am a bit confused as to which spec this applies to.

Practically, I would assume that it's for scaling TaskManager worker pods.
Is that correct?

Then how does it relate to an example and the parallelism part:
https://github.com/apache/flink-kubernetes-operator/blob/main/examples/flink-beam-example/beam-example.yaml#L41
. It seems on start, the operator will start a number of task manager pods
as defined by the parallelism value.

So does this mean parallelism is the starting count of the number of pods
and then HPA kicks in? What if I set parallelism value to be greater than
the minReplicas (
https://github.com/apache/flink-kubernetes-operator/blob/main/examples/hpa/basic-hpa.yaml#L25)
then which one takes precedence?

Ritesh



-- 
Ritesh