Skip to main content

Uninstall Container Observability

Container Observability (CO) is deployed across three separate Helm releases that work together to collect, process, and visualize container metrics, logs, and traces. The three components are:

  • Shared services (opscruise-shared-services): Central backend services including the Kafka consumer and data pipeline, deployed in the oc-shared-services namespace.

  • Frontend (opscruise-frontend): the CO web interface, deployed in a configurable frontend namespace.

  • Backend (opscruise-backend): the backend API and processing cluster, deployed in a configurable backend namespace.

The CO South collector (deployed per monitored cluster as opscruise-bundle) should be removed separately from each monitored cluster before running this procedure.

Removal order

Follow the order while uninstalling CO:

  1. Remove the frontend and backend before the shared services.

  2. Shared services provide the data pipeline that the frontend and backend depend on.

  3. Removing shared services first causes errors in the remaining components.

Uninstallation steps

Perform the steps:

  1. Uninstall the CO frontend.

    The frontend serves the Container Observability user interface. Remove it first to stop user-facing CO access before removing the backend it depends on. Replace <frontend-namespace> with the actual namespace name (commonly oc-frontend or opscruise).

    helm uninstall opscruise-frontend -n <frontend-namespace>

  2. Uninstall the CO backend cluster.

    The backend processes telemetry data received from the South collectors and serves it to the frontend. Remove it after the frontend. Replace <backend-cluster-namespace> with the actual namespace name (commonly oc-backend or opscruise).

    helm uninstall opscruise-backend -n <backend-cluster-namespace>

  3. Uninstall CO shared services.

    Shared services include the central Kafka consumer, the data processing pipeline, and other infrastructure components shared across CO deployments. Remove this last, after the frontend and backend have been uninstalled.

    helm uninstall opscruise-shared-services -n oc-shared-services

  4. Verify all CO pods have terminated.

    After running the uninstall commands, verify that no CO pods remain running. The following command lists all pods across the CO-related namespaces. All pods should be terminating or absent.

    kubectl get pods -n oc-shared-services
kubectl get pods -n <frontend-namespace>
kubectl get pods -n <backend-cluster-namespace>

    If pods remain in a stuck Terminating state, you can force-delete them with the following command:

    kubectl delete pod <pod-name> -n <namespace> --force --grace-period=0
In this section: