Restart Fleet-managed Elastic Agents

Restart the Elastic Agent process on one or more hosts directly from Fleet, without logging in to each host. A restart can resolve common problems, for example when an agent stops sending data.

You can also restart agents programmatically, using the POST /api/fleet/agents/{agentId}/restart endpoint for a single agent or the POST /api/fleet/agents/bulk_restart endpoint for multiple agents. For the endpoint specifications, refer to the Kibana API docs. For Elastic Cloud Serverless projects, use the Kibana Serverless API docs.

To restart the Elastic Agent service locally on the host instead, refer to Start and stop Elastic Agents on edge hosts.

  • You need All access for Agents in Fleet. Refer to Fleet roles and privileges.
  • The agents must be on version 9.6.0 or later. For agents on earlier versions, the restart action is unavailable.
  • The agents must be enrolled in Fleet. You can't restart unenrolled or inactive agents.
  • You can't restart agents that are enrolled in a hosted agent policy, because these agents are managed by an external orchestration solution, such as Elastic Cloud or Kubernetes. To restart them, use your orchestration solution.

To restart the Elastic Agent running on a specific host:

  1. In Kibana, find Fleet in the navigation menu or use the global search field, then select Agents.
  2. From the Actions menu next to the agent, select Restart agent. The action is also available in the Actions menu on the agent details page.
  3. In the confirmation window, click Restart agent.

Fleet sends the restart action to the agent, and the Elastic Agent process on the host restarts. To verify the result, track the restart progress.

To restart a group of agents in a single action:

  1. In Kibana, find Fleet in the navigation menu or use the global search field, then select Agents.
  2. Select the agents you want to restart.
  3. Select Actions → Restart x agents.
  4. In the confirmation window, click Restart agents.

Fleet restarts all eligible agents in your selection. Agents that can't be restarted, for example because they're on a version earlier than 9.6.0, are excluded from the action and receive individual error results. To verify the results, track the restart progress.

To follow the progress of the restart, click Agent activity on the Agents tab. For more information, refer to View agent activity. The activity log shows the restart action, when it completed, and any agents that couldn't be restarted. If any agents failed to restart, select Show errors to view the error message for each agent.

If an agent is offline when you issue the restart, it receives the action the next time it checks in with Fleet.