'EKS Managed Nodes with Cluster-Autoscaler scaling too much nodes
I'm sorry if such an issue was raised before; however, I couldn't find a solution so far.
I have an EKS cluster with a managed nodegroup and cluster-autoscaler running, and Terraform does everything.
The nodegroup is set to desired_size of 4 nodes.
When I upgraded the cluster and nodegroup versions, I saw that my nodegroup was scaled x3 times more than the actual size of the group. The cluster-autoscaler logs revealed that it kept adding nodes until it met the requirements for two pods; basically, those pods needed to run on a different AZ than other pods.
The managed nodegroup is set to run on all 3 AZ, so I don't understand why it failed to find a node for a specific time in one AZ while it scaled the group to run 12 nodes on 2 AZ.
Any help to optimize this behavior will be more than welcome :)
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|