Still not that familiar with Kubernetes and GKE, so I was curious whether some of you might know more about this.
I have a GKE cluster with auto-pilot enabled. This seems to work great.
I don’t have any real traffic yet, everything is in development. I therefore do not expect auto-scaling to scale up to more than 1 node at the moment. I have a few workloads/services and they request 2.25vCPU and 3.75 memory in total (so not much).
As I want to keep costs low during development, I was wondering why it looks like 5 nodes have been created?
kubectl get nodes shows 5 different nodes, 4 in the ‘default pool’, 1 in a pool with a seemingly-random name.
Does anybody know:
- why 2 node-pools have been created? Is this something the auto-pilot does? Are both node-pools required?
- why there are 5 nodes in total? Is this the minimum and would resizing it to 1 (in total, so removing a node-pool and resizing) work? Or are all 5 nodes required?
Thanks a lot in advance. If I missed documentation online, sorry for that. I have been looking around, but most of the documentation around auto-pilot still feels a little vague compared to the ‘standard’-version. I do prefer to keep using auto-pilot to reduce developer time.