> ## Documentation Index
> Fetch the complete documentation index at: https://notes.kodekloud.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Swarm Operations

> This article explains how to manage Swarm nodes, including promoting, demoting, draining, and removing nodes for maintenance and high availability.

In this lesson, you’ll learn how to:

* Promote and demote Swarm nodes
* Drain nodes for maintenance
* Remove nodes from your cluster

These operations help maintain high availability and streamline cluster management.

## Promote and Demote Nodes

Only Swarm manager nodes can orchestrate the cluster. You can elevate a worker node to manager or revert a manager back to a worker. Run these commands on any current manager.

### Promote a Worker to Manager

```bash theme={null}
docker node promote worker1
```

<Callout icon="lightbulb" color="#1CB2FE">
  Promoting a node requires you to be connected to an active manager. Check your current context with `docker info`.
</Callout>

Verify the updated node list:

| ID       | HOSTNAME | STATUS | AVAILABILITY | MANAGER STATUS | ENGINE VERSION |
| -------- | -------- | ------ | ------------ | -------------- | -------------- |
| 91uxgq6… | manager1 | Ready  | Active       | Leader         | 19.03.8        |
| 2lux7z6… | worker1  | Ready  | Active       | Reachable      | 19.03.8        |
| w0qr6k2… | worker2  | Ready  | Active       |                | 19.03.8        |

### Demote a Manager to Worker

```bash theme={null}
docker node demote worker1
```

Confirm the demotion:

| ID       | HOSTNAME | STATUS | AVAILABILITY | MANAGER STATUS | ENGINE VERSION |
| -------- | -------- | ------ | ------------ | -------------- | -------------- |
| 91uxgq6… | manager1 | Ready  | Active       | Leader         | 19.03.8        |
| 2lux7z6… | worker1  | Ready  | Active       |                | 19.03.8        |
| w0qr6k2… | worker2  | Ready  | Active       |                | 19.03.8        |

## Drain a Node for Maintenance

Before patching or upgrading, remove workloads from a node to avoid downtime. The following diagram shows a Swarm cluster with one manager and two workers running a “Web” service.

<Frame>
  ![The image illustrates a Docker Swarm setup with three nodes: one manager node and two worker nodes, each labeled with an IP address and a "Web" service.](https://kodekloud.com/kk-media/image/upload/v1752873939/notes-assets/images/Docker-Certified-Associate-Exam-Course-Swarm-Operations/docker-swarm-setup-nodes-web.jpg)
</Frame>

Drain `worker1` to migrate its tasks:

```bash theme={null}
docker node update --availability drain worker1
```

<Callout icon="triangle-alert" color="#FF6B6B">
  Draining a node stops new tasks from being scheduled and reschedules existing ones on other workers. Ensure your cluster has enough capacity before draining.
</Callout>

Check the drain status:

| ID       | HOSTNAME | STATUS | AVAILABILITY | MANAGER STATUS | ENGINE VERSION |
| -------- | -------- | ------ | ------------ | -------------- | -------------- |
| 91uxgq6… | manager1 | Ready  | Active       | Leader         | 19.03.8        |
| 2lux7z6… | worker1  | Ready  | Drain        |                | 19.03.8        |
| w0qr6k2… | worker2  | Ready  | Active       |                | 19.03.8        |

After maintenance, reactivate scheduling:

```bash theme={null}
docker node update --availability active worker1
```

Now `worker1` will accept new tasks:

| ID       | HOSTNAME | STATUS | AVAILABILITY | MANAGER STATUS | ENGINE VERSION |
| -------- | -------- | ------ | ------------ | -------------- | -------------- |
| 91uxgq6… | manager1 | Ready  | Active       | Leader         | 19.03.8        |
| 2lux7z6… | worker1  | Ready  | Active       |                | 19.03.8        |
| w0qr6k2… | worker2  | Ready  | Active       |                | 19.03.8        |

## Remove a Node from the Swarm

To permanently remove a worker node:

1. Drain the node:

   ```bash theme={null}
   docker node update --availability drain worker2
   ```

2. Log in to `worker2` and leave the swarm:

   ```bash theme={null}
   docker swarm leave
   ```

Example confirmation:

```bash theme={null}
Node left the swarm.
```

At this point, `worker2` is safely removed, and your cluster continues with the remaining nodes.

## References

* [Docker Swarm Overview](https://docs.docker.com/engine/swarm/)
* [Docker Node Management](https://docs.docker.com/engine/reference/commandline/node/)
* [Docker Maintenance Best Practices](https://docs.docker.com/config/containers/resource_constraints/)

<CardGroup>
  <Card title="Watch Video" icon="video" cta="Learn more" href="https://learn.kodekloud.com/user/courses/docker-certified-associate-exam-course/module/16b8b1e1-1e1f-4e11-976f-8d5c1223c53d/lesson/278411b6-0d56-49fa-a8a3-51f60262f6fd" />
</CardGroup>