Description
GKE node auto-repair periodically checks node health and repairs persistently unhealthy nodes to help maintain cluster availability. Disabling it can increase manual intervention and delay recovery.
It is enabled by default for new Standard node pools and cannot be disabled in Autopilot. An omitted setting alone does not establish that it is disabled; check the actual node pool.
Potential impact
- Unhealthy nodes can remain longer and affect workload availability.
- Recovery can depend more heavily on manual operations.
Remediation
Set management.autoRepair to true and verify the effective node-pool setting. Prepare workload replicas and persistent data for nodes to be drained and re-created during repair. Auto-repair does not replace application fixes or backups.
Examples
These excerpts show node-pool management settings. Add the resource name, target cluster and other required configuration for your environment.
Before
apiVersion: container.gcp.crossplane.io/v1beta1
kind: NodePool
spec:
forProvider:
management:
autoRepair: false
initialNodeCount: 3
This explicitly disables node auto-repair.
After
apiVersion: container.gcp.crossplane.io/v1beta1
kind: NodePool
spec:
forProvider:
management:
autoRepair: true
initialNodeCount: 3
This enables node auto-repair. Check node health and repair events after deployment.