Shutdown Clusters Cleanly

In cases where you need to physically move hardware, or you have no other choice you will need to shut down your cluster in a safe manner.

Before you begin

  • Confirm no applications are running that are using the cluster as backend storage.

  • Confirm that you know which nodes are the seed nodes. Seed nodes are specified in the scylla.yaml file.

Procedure

On each node, in parallel:

  1. Run the command nodetool drain.

  2. Validate that the drain procedure has completed by running nodetool status. If the node’s status is listed as DN, then the drain command has been executed successfully.

  3. Stop the node after drain completed successfully.

    sudo systemctl stop scylla-server
    
    sudo service scylla-server stop
    
    docker exec -it some-scylla supervisorctl stop scylla
    

    (without stopping some-scylla container)

  4. To start the nodes again safely, proceed to the Start Clusters Cleanly procedure.