Document settings for scaling/resources in k8s for runners
I have a dedicated cluster in google cloud k8s for runners. 5 VMs.
In the hours of active developing, there are a lot of pipelines in the state "waiting for pods".
We are trying to optimize our pipelines, but anyway we are looking for a solution or a cheat sheet of how to do our cluster right.
- How to do scaling right?
- Setting for runners?
- How to measure if we claim too many resources for each pod?
- Any other pieces of advice?
Edited by Steve Xuereb