Node pools

What they are

Node pools (in the Karpenter sense) are Declarative sets of rules that tell Karpenter how to provision worker nodes: which instance sizes, architectures, capacity types (for example Spot versus on-demand), and which NodeClass supplies networking and AMI settings. Each pool is a Kubernetes NodePool object; the controller creates and destroys nodes to match pending pods within those constraints.

In this platform, optional Karpenter node pool manifests are delivered as a small Helm chart so defaults and tenant-specific overrides stay in Git, alongside cluster definitions.

---

Next steps

For feature flags, value file layout (config/karpenter_nodepools/…), merge order, and examples (Spot-only, ARM64, and so on), see Karpenter node pool configuration.