EKS Insufficient capacity errors

Question

In EKS cluster with managed node group often ran into capacity insufficient errors that come up with EKS autoscaling. How to handle this error and how to make EKS retry until it gets capacity?

Answer

Worker nodes are provisioned via EC2 Auto Scaling, both for self-managed and Managed Worker Nodes. Additional instances are called for by increasing the Auto Scaling Group's (ASG) DesiredCount parameter. If EC2 Auto Scaling is unable to provision an instance due to insufficient capacity, it will retry until an instance becomes available. Meanwhile, the Kubernetes scheduler will keep trying to place the pods until the additional instances become available and It will not fail and give up due to insufficient capacity, either. 
In order to avoid insufficient capacity issue then follow below strategies that can be applied:
1. Add more AZs to the Auto Scaling Group if possible.
2. Use a MixedInstancesPolicy in the AutoScalingGroup to allow for "fallback" instance type if the preferred instance type is unavailable.
3. Buy On-Demand Capacity reservations (ODCR) instances.

Answer

Hi - Thanks for reaching out. Are you looking for some automation to be built? If you receive the error while attempting to create an Amazon EKS cluster, then one of the Availability Zones you specified doesn't have sufficient capacity to support a cluster. Retry creating your cluster with subnets in your cluster VPC that are hosted in the Availability Zones returned by this error message. Are you getting in error message which AZ to select from?

EKS Insufficient capacity errors

Relevant content