Reboot a persistent resource

You can reboot any persistent resource that's in the RUNNING or ERROR state. Rebooting a persistent resource lets you recover from errors that the persistent resource can't recover from on its own. You can also reboot a persistent resource to manually obtain more up-to-date clusters. This page shows you how to reboot a persistent resource by using the Google Cloud console and the REST API.

Required roles

To get the permission that you need to reboot a persistent resource, ask your administrator to grant you the Vertex AI Administrator (roles/aiplatform.admin) IAM role on your project. For more information about granting roles, see Manage access.

This predefined role contains the aiplatform.persistentResources.update permission, which is required to reboot a persistent resource.

You might also be able to get this permission with custom roles or other predefined roles.

Reboot a persistent resource

Select one of the following tabs for instructions on how to reboot a persistent resource. Make sure there's no training jobs running on the persistent resource.


To reboot a persistent resource in the Google Cloud console, do the following:

  1. In the Google Cloud console, go to the Persistent resources page.

    Go to Persistent resources

  2. Next to the name of the persistent resource that you want to reboot, click the vertical ellipses ().

  3. Click Reboot.

  4. Click Confirm.


Before using any of the request data, make the following replacements:

  • PROJECT_ID: The Project ID of the persistent resource that you want to reboot.
  • LOCATION: The region of the persistent resource that you want to reboot.
  • PERSISTENT_RESOURCE_ID: The ID of the persistent resource that you want to reboot.

HTTP method and URL:


To send your request, expand one of these options:

You should receive a JSON response similar to the following:

    "name": "projects/123456789012/locations/us-central1/persistentResources/test-persistent-resource/operations/1234567890123456789",
    "metadata": {
      "@type": "",
      "genericMetadata": {
        "createTime": "2024-03-18T17:31:54.955004Z",
        "updateTime": "2024-03-18T17:31:55.204817Z",
        "state": "RUNNING",
        "worksOn": [
      "progressMessage": "Waiting for persistent resource shut down."

Rebooting a persistent resource is a long running operation, during which the persistent resource can't be deleted. The operation contains a progressMessage field that populates with an error status if one occurs. After the operation indicates "done: true", check the status of the persistent resource. If the persistent resource is in the RUNNING state, the reboot is successful and it's ready to run training jobs.


The following are limitations for rebooting a persistent resource:

  • In some cases, it's possible to lose capacity of scarce resources when rebooting a persistent resource. Full resource retention is not guaranteed.
  • Reboot is not available on Ray on Vertex AI.
  • Persistent resources containing autoscaled worker pools reboot with the minimum replica count.

What's next