Skip to main content
This page covers the license errors the Admin Server reports, what each means, and how to recover. For how licenses work, see Licensing. Start by checking the current state:
The Licensing page of the Admin UI shows the same state, the limits next to current usage, and any violations.

Changes are refused with 423 Locked

The license is expired or invalid, so the management plane is read-only. Your databases keep serving traffic. Restart, start, stop, switchover, on-demand backups and deployment deletes still work. The text after the colon gives the reason: Activate the replacement from Licensing in the Admin UI, or with the CLI:
Sign-in and licensing stay available in read-only mode, so you can always activate a license.

A change fails with License limit exceeded

The change would take usage past one of the license’s limits: deployments, CPU, memory or users. The message names the resource. Nothing was applied.
  • Free capacity: delete or scale down a deployment, or lower its resources. CPU and memory count the requests of every node, including Sentinels and every shard.
  • Or raise the limit: resize the cluster’s license in the Enterprise Portal, within your organization’s quota, and activate the new license.
When usage is already over a limit, the state is locked. Actions that keep usage flat or reduce it still work, so you can bring it back within the license.

Activation errors

Activating a license, from the Admin UI, CLI, API or Helm, validates it first. These are the reasons it can fail:

The license is bound to another cluster

A license works only on the cluster whose kube-system namespace uid it was issued for. Compare the two:
Decode the license payload to read its k8sClusterId:
If they differ, register this cluster in the Enterprise Portal with its ID. A rebuilt cluster has a new ID, so it needs a new license. See Disaster recovery into a replacement cluster.

The cluster ID can’t be read

The Admin Server couldn’t read the kube-system namespace, because of an API server outage, a timeout or missing RBAC. The license stays valid for the grace period, 24 hours by default, and the Admin UI shows when it ends. Once it ends, the license becomes invalid with:
Check that the Admin Server can reach the API server and read the namespace:
The license becomes valid again with the next successful check. Change the grace period with adminServer.license.verificationGracePeriodHours.

A license supplied through Helm isn’t applied

The Admin Server reads a Helm-supplied license when it starts. If that license is invalid, expired or bound to another cluster while the installed license is still valid, it keeps the installed one and logs:
Fix the value in adminServer.license.value, or in the Secret named by adminServer.license.existingSecret.name, and upgrade the release. See Enterprise license.

The license is lost after a restart

A license activated in the Admin UI, CLI or API is stored in /data/license-state.json on the Admin Server’s volume. With adminServer.persistence.enabled: false, it is lost on every restart. Keep persistence enabled, or supply the license through Helm. To protect the file, see Admin Server backup.