Skip to Content

Remove an App

A Context administrator or the person who created a native App can remove it. Removal is available from the App detail page, not the creation setup page. Legacy CECG Apps do not support this removal flow. Labels may differ according to your Context’s naming.

  1. Select the App’s Context and open its detail page.
  2. On the detail page, open the three-dot actions menu in the App name card and choose Remove. Read the confirmation dialog and type the App’s exact name.
  3. Confirm removal. The App remains visible as Pending deletion until all remaining registered targets finish deleting its deployment boundary.
  4. Remove blocking resources from the namespaces shown for each cluster, including environment subnamespaces, using your existing cluster access. Stop pipelines that recreate those resources. Ask someone with cluster access if needed.

Removal cannot be cancelled. Namespace data may be lost. The GitHub repository is kept, and workflows already started in GitHub are not cancelled automatically. The App’s name remains reserved after removal.

Pending deletion

The App detail page keeps its full Path to Production graph and tabs visible. A compact Pending deletion notice appears above the App name card. Expand Deletion details and cleanup guidance for cluster status, namespace names, and the requester’s email.

The Control Agent requests deletion of the App’s Argo Application. Argo’s existing finalizer controls cleanup. The agent does not directly delete namespaces, workloads, storage, environment anchors or access rules, and does not remove finalizers to force completion.

Remaining resources and HNC environment subnamespaces can prevent completion. Check each namespace shown in Portal and remove its blocking resources using the normal cluster procedures. Do not strip finalizers to bypass cleanup. Portal continues checking automatically; there is no need to submit removal again.

Waiting for the cluster agent means current evidence is unavailable. Restore the assigned cluster connection if it is still registered. A registration still being removed also remains pending. Once its registration is removed from Portal, that target no longer blocks application deletion, including Apps already waiting for the agent. The receipt says Cluster removed — cleanup not confirmed. Resources may remain on that cluster and require manual cleanup. Deletion blocked requires an operator to check resource ownership, access, and the Argo Application’s finalizer.

Once every remaining registered target confirms that the Argo Application is absent, Portal removes the App from its catalogue and keeps a completion receipt at the removal view. Each target distinguishes confirmed Argo Application absence from a removed cluster registration. Completion does not assert cleanup on removed clusters, of unrelated resources or of external cloud services.

Advanced subnamespace cleanup

Expand Advanced: clean up subnamespaces in the pending-deletion notice to see one command box and one Copy button per cluster. When connection details are available for the assigned cluster registration, the block starts with corectl connect CLUSTER --context CONTEXT_NAME to configure and select its Kubernetes context. The following commands explicitly use that context. Otherwise, replace the quoted <KUBE_CONTEXT> placeholder with your existing local context for the named cluster. Portal does not execute these commands. Review the namespace names before running the block. Commands run in order and stop if a preceding command fails.

The block deletes each observed subnamespace using its parent namespace:

kubectl --context '<KUBE_CONTEXT>' -n '<PARENT_NAMESPACE>' delete subnamespaceanchors.hnc.x-k8s.io '<SUBNAMESPACE>'

Deleting a subnamespace deletes its contents, including workloads, Secrets and persistent-volume claims. Stop pipelines that recreate resources. Remove nested subnamespaces deepest first and wait for each deletion to finish; inspect anchors in each parent namespace. Do not enable cascading deletion or remove finalizers to bypass blockers. See the HNC deletion guide .

When Runtime has fresh namespace hierarchy and subnamespace annotations, Portal fills in observed names and orders deletion commands deepest first, showing up to 100 subnamespaces per cluster and the last-observed time. This is not a complete inventory of blockers: Runtime does not collect anchors themselves. Observations older than five minutes, missing hierarchy data or an unavailable Runtime service hide cleanup commands. Refresh observations fetches current information. Removed or completed cluster targets have no commands in this section. None of this grants additional cluster access or changes normal deletion or Force delete behavior.

Force delete from Portal

While an App is pending deletion, its creator or a Context administrator can select Force delete in the notice. Read the warning and type the App’s exact name, then choose Confirm force delete.

This removes the App from Portal immediately without waiting for cluster cleanup. Resources may remain in the cluster, continue incurring costs, and require manual cleanup. Deletion already in progress may still continue. The GitHub repository is kept, the App name remains reserved, and this action cannot be undone.

The receipt says Removed from Portal and explicitly states that cluster cleanup has not been confirmed. It retains the cluster and namespace names for manual cleanup, and shows the original requester and force-deletion actor by email. If a user record is unavailable, Portal displays Unknown user.

Find deleted Apps

Open Advanced at the bottom of the sidebar, then Deleted Apps. Labels follow your Context’s naming configuration. This native-Context history is available to all members of the selected Context and shows completed removals, including Apps removed during creation. Pending deletions remain in the catalogue.

The list shows the name, creator, deletion actor, completion time in UTC and deletion type. For normal deletion, Deleted by identifies the person who requested removal. Removed from Portal identifies force deletion: it shows the force actor and the original removal requester separately, and warns that cluster cleanup is unconfirmed. An identity that can no longer be resolved is shown as Unknown. Normal completion also does not prove cleanup on removed clusters or of external resources.

Use Older records for the next page and Latest records to return to the newest entries. History is read-only; it does not restore Apps or release their reserved names. If history cannot be loaded, use Retry.