KEDA Config Changes Do Not Automatically Reflect in HPA

[shashiik]

Hi,

This issue is with KEDA 1.5. I've been noticing an issue when updating a KEDA config for a deployment. For example, if I have a KEDA scalar object for a deployment running with a maxReplica=4, and I change the configuration to maxReplica=10, the ScaledObject (with kubectl get ScaledObject <>) shows the correct configuration with maxReplica=10. But the HPA doesn't reflect this change. It stays stuck at the 4. It usually take several hours until the change is correctly reflected in the HPA.

I was wondering what the cause for this is? Or how to make that change instantaneous?

Another question that might be related to this: How many instances of the api-metric server do I need running in production? Is there a formula or guide available to say the # of KEDA-enabled deployments is X, the number of instances of the metric-server/operator I need running is Y?

Thanks!