Teardown

Hub › Kubernetes › Argo Stack › Teardown

Goal

Remove all Argo Stack resources and optionally delete the kind cluster. A clean teardown matches the clean setup — each piece you installed comes out in reverse order.

Step 1: Remove Argo Events

kubectl delete namespace argo-events

Deleting the namespace removes everything inside it in one shot:

• The EventBus (Redis StatefulSet + Service)
• The EventSource (webhook) and its Service
• The Sensor (webhook-sensor)
• The RBAC bindings (operate-workflow-sa, roles) created in page 02
• All the Workflows the Sensor created in page 04 (they live in this namespace)

Verify it is gone:

kubectl get namespace argo-events
# Error from server (NotFound): namespaces "argo-events" not found

The NotFound error is the success signal — the namespace no longer exists.

Step 2: Remove Argo Workflows

kubectl delete namespace argo

This removes the workflow controller Deployment and the argo-server, plus the build-pipeline WorkflowTemplate (it is cluster-scoped by default for templates, but the controller + server live in this namespace).

Optionally remove the Argo Workflows CRDs for a fully clean slate:

kubectl delete crd workflowtemplates.argoproj.io workflows.argoproj.io

Skip the CRD deletion if you plan to reinstall Argo Workflows soon — keeping them preserves any global template state and saves a reinstall cycle.

Step 3: Remove ArgoCD (optional)

If you want to keep ArgoCD for GitOps on other projects, skip this step — ArgoCD is independent of the Events/Workflows stack and will keep running fine.

Otherwise:

kubectl delete namespace argocd

For the full ArgoCD teardown walkthrough (including CRDs and the argocd CLI config), see GitOps with ArgoCD.

Step 4: Delete the kind cluster

If you're done with Kubernetes entirely for now:

kind delete cluster

This removes everything — the control-plane container, the containerd runtime, the Docker network, and every resource across all namespaces. After this, kubectl get nodes returns a connection error, which is the expected tidy end state.

Confirm it is gone

kubectl get nodes   # error / no cluster — expected

You should see an error like Unable to connect to the server — that means the cluster is fully removed.

What you learned

Over the full Argo Stack section you built and tore down:

• Argo Events: webhook EventSource → EventBus (Redis) → Sensor with a trigger
• Argo Workflows: a WorkflowTemplate with sequential steps and parameter passing between them
• Pipeline: a Sensor k8s trigger creates Workflow resources automatically from the template on every webhook event
• Event-driven GitOps: a webhook triggers the build pipeline; in production ArgoCD handles the deploy side

The shape you built locally — push → event → workflow → deploy — is the same shape production CI/CD systems run, just with GitHub webhooks instead of curl, real image builds instead of echo, and ArgoCD syncs instead of simulated deploys.

Where to go next

• Argo Events docs — official documentation, including the full list of EventSource types (S3, Kafka, calendar, GitHub, …)
• Argo Workflows docs — official documentation, covering DAGs, artifacts, retries, and conditional steps
• GitOps with ArgoCD — revisit the ArgoCD tutorial that the pipeline's deploy step simulated
• Production patterns to explore next:
  • GitHub webhooks instead of curl (a real EventSource)
  • kaniko for building container images inside a workflow step
  • ArgoCD Image Updater for automatic rollouts when a new image tag is pushed

This is the final page of the Argo Stack section — you're done.