Solved
ADM Recommendation in Kubernetes
Our ADM pod in Kubernetes keeps briefly hitting the HPA threshold and scaling up and then immediately back down.
The pod has 4 cpu allocated with limit 6. HPA threshold is set at 85%.
Is it recommended to run multiple ADM pods (e.g. 2) or to increase the size of the 1 ADM pod to say 6 cpu limit 8?
***Edited by Moderator Marije to add Capability tags***
To see attachments, please log in.
Accepted Solution
Posted: 2 years ago
Updated: 2 years ago
Updated: 29 Jan 2024 15:53 EST
PEGA
Pegasystems Inc.
NL
Hi @AlexM171,
This depends a little bit on the situation, but since the ADM service only services background processes it is generally fine to have only a single pod. If the cheapest option for you is to size up the single node, then it sounds like a good solution.
One thing to maybe keep in mind though if you target only a single (larger) node, if the load increases in the future and the system scales to two pods you might now be over provisioned again. When having multiple smaller pods, the step increase of adding an additional pod is smaller.
Lastly, if it is scaling back down immediately it can also help to set or increase the sabilization window for scaling down: https://kubernetes.io/docs/tasks/run-application/horizontal-pod-autoscale/
Hope this helps.
Best,
Tom