-
|
Environmental Info: Node(s) CPU architecture, OS, and Version:
Cluster Configuration: 3 server, 3 agents Describe the bug: I have 6 nodes, rke2-coredns tries to deploy 9 instances, but the antiaffinity rule does not allow deploying 2 coredns pods next to each other, due to this: https://github.com/rancher/rke2-charts/blob/main-source/packages/rke2-coredns/generated-changes/patch/values.yaml.patch#L56-L65 The reason it tries to deploy 9 instances is because rke2-coredns-autoscaler is telling it to do so, because I have a lot of CPU cores on the machines. Why is this antiaffinity rule required and not preferred? |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
|
You could probably better tune the autoscaler for your use case, or disable it entirely if you prefer. 9 pods for 6 nodes is probably overkill. The anti-affinity rule exists because HA is not improved by having a bunch more pods all running on the same node. |
Beta Was this translation helpful? Give feedback.
You could probably better tune the autoscaler for your use case, or disable it entirely if you prefer. 9 pods for 6 nodes is probably overkill.
The anti-affinity rule exists because HA is not improved by having a bunch more pods all running on the same node.