Restart Hybrid Manager components

Suggest edits

This how-to guide explains how to safely restart Hybrid Manager (HM) platform components running on Kubernetes.

This is useful when:

  • Applying configuration changes
  • Forcing component reinitialization
  • Recovering from transient errors or unexpected behavior
  • Performing troubleshooting steps

Important: This procedure applies to Hybrid Manager platform components only — not to EDB-managed Postgres clusters. See Manage Kubernetes storage and Manage Kubernetes networking for managing related infrastructure.

When to restart Hybrid Manager components

Common scenarios:

  • UI/API component appears unresponsive
  • Operator logs show errors that require reinitialization
  • Configuration change (ConfigMap or Secret) applied
  • Routine troubleshooting or operational checks

What can be safely restarted

The following components can be restarted independently:

  • Hybrid Manager UI (upm-ui namespace/component)
  • Hybrid Manager API services (upm-api-* namespaces/components)
  • Postgres operator (postgresql-operator-system namespace)
  • PGD operator (pgd-operator-system namespace, if used)
  • Transporter services (transporter-* namespaces/components)
  • Storage location operator (storage-location-operator namespace)
  • Beacon and telemetry components (upm-beacon, upm-beaco-ff-base)
  • Other UPM components as applicable (upm-* namespaces/components)

How to restart components

Preferred method: delete Pods → Kubernetes will recreate them based on Deployments.

Steps:

  1. Identify the namespace of the component you wish to restart.
  2. Use kubectl get pods -n <namespace> to list component Pods.
  3. Use kubectl delete pod <pod-name> -n <namespace> to delete Pod(s).

Kubernetes will automatically recreate the Pod(s) based on the Deployment spec.

Notes:

  • Deleting Pods does not delete underlying PersistentVolumes or ConfigMaps.
  • Stateless services (UI/API) will restart cleanly.
  • Operator restarts may trigger reconciliation of managed resources — monitor carefully.
  • Stagger restarts if troubleshooting multiple components to isolate impact.

What not to restart manually

Do not restart Kubernetes control plane components unless specifically guided by your Kubernetes administrator. Avoid manually restarting core Kubernetes system Pods unless required.

Do not forcibly delete StatefulSet Pods for Postgres unless following documented Postgres operator procedures.

Monitor after restart

After restarting components:

  • Monitor Pod readiness → Pods should return to Running and Ready state.
  • Check relevant logs for errors on startup.
  • Validate Hybrid Manager UI/API responsiveness.
  • Validate operator behavior → check reconciliation activity in operator logs.

Summary checklist

  • Identify Hybrid Manager component to restart.
  • Delete Pod(s) for component → let Deployment recreate clean Pods.
  • Monitor readiness and validate behavior.
  • Avoid restarting StatefulSet Pods for Postgres clusters manually.
  • Document any restarts performed as part of troubleshooting.

Related topics

Could this page be better? Report a problem or suggest an addition!