Shutting Down a Single Swarm Node

In the Army, we call this a “Tactical Withdrawal.” You don’t just pull the plug; you move your troops out first.

To gracefully shut down a node in a Docker Swarm, you should “drain” it. This tells the Swarm manager to move any tasks (containers) running on that node to other available nodes in the cluster.

Procedure

1. Drain the node:

   docker node update --availability drain <node-id>

2. Verify tasks are moved:

   docker node ps <node-id>

3. Shut down the OS: Once all tasks have been successfully moved, you can safely shut down or reboot the node.

Bringing it back online

When the node returns, you should set it back to “active”:

docker node update --availability active <node-id>