Control Pod Density
Pod density is the number of pods per node.
Kubernetes has a default limit of 110 pods per node. If you are using the EKS Optimized AMI on AWS, the number of pods is limited by instance type in the default configuration.
Do not use the
max-pods argument to kubelet. Karpenter is not aware of this value. For example, Karpenter may provision an instance expecting it to accommodate more pods than this static limit.
Increase Pod Density
☁️ AWS Specific
The number of pods on a node is limited by the number of networking interfaces (ENIs) that may be attached to a node.
AWS VPC CNI v1.9 introduced prefix assignment. In short, a single ENI can provide IP addresses for multiple pods. Much higher pod densities are now supported.
Run the Karpenter controller with the enviornment variable
AWS_ENI_LIMITED_POD_DENSITY (or the argument
--aws-eni-limited-pod-density=true) to enable nodes with more than 110 pods.
Environment variables for the Karpenter controller may be specified as helm chart values.
Limit Pod Density
Generally, increasing pod density is more efficient. However, some use cases exist for limiting pod density.
You can use topology spread features to reduce blast radius. For example, spreading workloads across EC2 Availability Zones.
Restrict Instance Types
Exclude large instance sizes to reduce the blast radius of an EC2 instance failure.
Consider setting up upper or lower boundaries on target instance sizes with the node.kubernetes.io/instance-type key.
The following example shows how to avoid provisioning large Graviton instances in order to reduce the impact of individual instance failures:
-key: node.kubernetes.io/instance-type operator: NotIn values: 'm6g.16xlarge' 'm6gd.16xlarge' 'r6g.16xlarge' 'r6gd.16xlarge' 'c6g.16xlarge'