Karpenter is the modern Kubernetes node autoscaler that provisions right-sized nodes just in time. Master NodePools, NodeClaims, consolidation, disruption budgets, and drift patterns for efficient cluster operations.
0 / 5 completed
1 / 5
Your Kubernetes cluster uses Karpenter. A pod remains Pending because no existing node has sufficient resources. What Karpenter object defines what kinds of nodes can be provisioned?
A NodePool is the primary Karpenter custom resource that defines constraints for node provisioning: instance types, architectures, capacity types (spot/on-demand), zones, and labels/taints. When a pod is unschedulable, Karpenter evaluates matching NodePools to determine what node to launch.
2 / 5
In Karpenter's model, what is a NodeClaim?
A NodeClaim is a Karpenter custom resource (formerly Machine) that represents a single node lifecycle — from the cloud API call to launch it, through its registration as a Kubernetes Node, to its eventual termination. NodeClaims are created by Karpenter automatically; users typically don't manage them directly.
3 / 5
Karpenter's consolidation feature terminates a node and reschedules its pods elsewhere. What is the primary goal of consolidation?
Consolidation is Karpenter's cost optimization loop. It identifies nodes where pods could be rescheduled onto existing capacity (or a cheaper instance type), cordons the node, evicts pods, and terminates the node. This keeps the cluster right-sized without manual intervention, directly reducing cloud spend.
4 / 5
A NodePool has disruption.budgets configured. What does a disruption budget protect against in Karpenter?
Disruption budgets in a NodePool limit how many nodes Karpenter can disrupt at one time during voluntary disruptions (consolidation, drift, expiry). For example, nodes: "10%" means at most 10% of nodes in the pool can be disrupted simultaneously, preventing availability impact during aggressive consolidation.
5 / 5
A node in a Karpenter-managed cluster has karpenter.sh/disruption: drift status. What does drift mean in this context?
Drift occurs when a node's current configuration (AMI, instance type, security groups, etc.) no longer matches the desired spec defined in its NodePool or EC2NodeClass. Karpenter will voluntarily replace drifted nodes to bring them back into compliance, subject to disruption budgets.