Asking for help? Comment out what you need so we can get more information to help you!
Cluster information:
Kubernetes version: 1.26
Cloud being used: AWS
Installation method: kOps
Host OS: Linux
Question
I’m running a cluster with a lot of instances of cloud native application that consists of ~20 microservices. Most of these microservices are tiny, but they may overload node if too many of the same type microservice scheduled on the same node. So I’m looking for a way to define PodTopologySpread for most of them w/o explicit changes to pod specs and helm charts. Helm charts / application spec have no topology constraints in it and it’s a bit complicated to deliver these changes to the upstream.
I’ve checked Cluster-wide PodTopologySpread constraints, but it seems it’s impossible to define multiple rules with labelSelector, yet topologySpread occurs only for pods that belong to the same Deployment/Statefulset/Service/etc, which is not my case - different instances of this microservice monster have separate deployments and services.
Before starting very long/painful/clunky process of delivering topologySpreadConstraints to the upstream, I decided to check here, maybe I’m missing some way to solve my puzzle.
Thank you in advance!