Add removal instructions.

README.md

@@ -277,4 +277,45 @@ If the cluster is in "Update failed" state, and the underlying issue (e.g., quot
 openstack coe cluster resize inf110 $(openstack coe cluster show inf110 -f json | jq .node_count)
 ```
 
-The above is a request to resize to the current (known) node count. If the current count is incorrect, you can specify a different count.
+The above is a request to resize to the current (known) node count. If the current count is incorrect, you can specify a different count.
\ No newline at end of file
+
+## Removal
+
+To remove the cluster, the VMs, and the resources created for Coder, start by deleting all Coder workspaces (using the notebook):
+
+Then, delete the Coder deployment:
+
+```bash
+helm uninstall coder -n coder
+helm uninstall coder-db -n coder
+# Alternatively:
+kubectl delete --all deployments --namespace=coder
+kubectl delete --all services --namespace=coder
+kubectl delete --all pvc --namespace=coder
+kubectl delete --all pods --namespace=coder
+# The PVCs are not deleted by the above command, so they need to be deleted manually
+kubectl delete pvc -n coder $(kubectl get pvc -n coder -o json | jq -r '.items[].metadata.name')
+```
+
+Delete the Kubernetes dashboard service and other kube-system resources:
+
+```bash
+kubectl delete --all deployments --namespace=kube-system
+kubectl delete --all services --namespace=kube-system
+kubectl delete --all pods --namespace=kube-system
+```
+
+Delete the VMs and the cluster:
+
+```bash
+openstack server delete inf110-registry
+openstack server delete inf110-reverse-proxy
+openstack coe cluster delete inf110
+```
+
+Check that no load balancers or volumes are left:
+
+```bash
+openstack loadbalancer list
+openstack volume list
+```
\ No newline at end of file