Skip to content

Latest commit

 

History

History
102 lines (69 loc) · 2.7 KB

uninstall.md

File metadata and controls

102 lines (69 loc) · 2.7 KB

Uninstalling the ngrok Operator

This guide covers safe uninstallation of the ngrok-operator, ensuring proper cleanup of ngrok API resources and preventing stuck Kubernetes resources.

Quick Start

Helm (Recommended)

helm uninstall ngrok-operator -n ngrok-operator

The pre-delete hook automatically triggers drain mode, waits for cleanup, then completes uninstall.

Manual / Non-Helm

# 1. Trigger drain
kubectl delete kubernetesoperator <name> -n <namespace>

# 2. Wait for completion
kubectl wait --for=delete kubernetesoperator/<name> -n <namespace> --timeout=300s

# 3. Delete operator resources
kubectl delete -f operator-manifests.yaml

Drain Policies

Configure via the drainPolicy Helm value:

Policy ngrok API Resources Best For
Retain (default) Preserved in your account Production - keep your configuration
Delete Removed from your account Dev/testing - clean slate

Both policies remove finalizers from all managed Kubernetes resources.

Monitoring Drain Progress

kubectl get kubernetesoperator <name> -n <namespace> -o yaml

Status fields:

  • drainStatus: pending → draining → completed/failed
  • drainProgress: X/Y (processed/total)
  • drainErrors: Error messages if any

Multi-Instance Installations

When multiple operator instances exist, drain only affects resources managed by that instance:

  • Ingress: Filtered by IngressClass
  • Gateway/Routes: Filtered by GatewayClass
  • Other resources: Filtered by namespace (if watchNamespace is set)

Troubleshooting

Resources Stuck in Terminating

  1. Check if operator is running: kubectl get pods -n ngrok-operator
  2. Manual finalizer removal (may orphan ngrok resources): 
    kubectl patch ingress <name> -n <namespace> \
  --type=json -p='[{"op":"remove","path":"/metadata/finalizers"}]'
  3. Or re-install temporarily to let it clean up properly

Drain Timeout

Increase the hook timeout:

cleanupHook:
  timeout: 600  # 10 minutes

Orphaned ngrok Resources

Delete manually from ngrok Dashboard or via the ngrok CLI.

Helm Configuration

drainPolicy: "Retain"  # or "Delete"

cleanupHook:
  enabled: true        # default
  timeout: 300         # seconds

Architecture

See internal/drain/ for implementation:

Component Role
Orchestrator Coordinates the drain workflow
StateChecker Detects drain mode (caches once triggered)
Drainer Processes resources based on policy

Controllers check drain.IsDraining() before adding finalizers or syncing new resources.