Remove a Node from a ScyllaDB Cluster (Down Scale)¶

Removing a Running Node¶

1. Run the nodetool status command to check the status of the nodes in your cluster.

   Datacenter: DC1
      Status=Up/Down
      State=Normal/Leaving/Joining/Moving
      --  Address        Load       Tokens  Owns (effective)                         Host ID         Rack
      UN  192.168.1.201  112.82 KB  256     32.7%             8d5ed9f4-7764-4dbd-bad8-43fddce94b7c   B1
      UN  192.168.1.202  91.11 KB   256     32.9%             125ed9f4-7777-1dbn-mac8-43fddce9123e   B1
      UN  192.168.1.203  124.42 KB  256     32.6%             675ed9f4-6564-6dbd-can8-43fddce952gy   B1
   

2. If the node status is Up Normal (UN), run the nodetool decommission command to remove the node you are connected to. Using nodetool decommission is the recommended method for cluster scale-down operations. It prevents data loss by ensuring that the node you’re removing streams its data to the remaining nodes in the cluster.

   If the node is Joining, see Safely Remove a Joining Node.

   If the node status is Down, see Removing an Unavailable Node.

   Warning

   Review current disk space utilization on existing nodes and make sure the amount of data streamed from the node being removed can fit into the disk space available on the remaining nodes. If there is not enough disk space on the remaining nodes, the removal of a node will fail. Add more storage to remaining nodes before starting the removal procedure.

3. Run the nodetool netstats command to monitor the progress of the token reallocation.

4. Run the nodetool status command to verify that the node has been removed.

   Datacenter: DC1
   Status=Up/Down
   State=Normal/Leaving/Joining/Moving
   --  Address        Load       Tokens  Owns (effective)                         Host ID         Rack
   UN  192.168.1.201  112.82 KB  256     32.7%             8d5ed9f4-7764-4dbd-bad8-43fddce94b7c   B1
   UN  192.168.1.202  91.11 KB   256     32.9%             125ed9f4-7777-1dbn-mac8-43fddce9123e   B1
   

5. Manually remove the data and commit log stored on that node.

   When a node is removed from the cluster, its data is not automatically removed. You need to manually remove the data to ensure it is no longer counted against the load on that node. Delete the data with the following commands:

   sudo rm -rf /var/lib/scylla/data
   sudo find /var/lib/scylla/commitlog -type f -delete
   sudo find /var/lib/scylla/hints -type f -delete
   sudo find /var/lib/scylla/view_hints -type f -delete
   

Removing an Unavailable Node¶

If the node status is Down Normal (DN), you should try to restore it. Once the node is up, use the nodetool decommission command (see Removing a Running Node) to remove it.

If all attempts to restore the node have failed and the node is permanently down, you can remove the node by running the nodetool removenode command providing the Host ID of the node you are removing. See nodetool removenode for details.

Example:

nodetool removenode 675ed9f4-6564-6dbd-can8-43fddce952gy

The nodetool removenode command notifies other nodes that the token range it owns needs to be moved and the nodes should redistribute the data using streaming. Using the command does not guarantee the consistency of the rebalanced data if stream sources do not have the most recent data. In addition, if some nodes are unavailable or another error occurs, the nodetool removenode operation will fail. To ensure successful operation and preserve consistency among replicas, you should:

• Make sure the status of all other nodes in the cluster is Up Normal (UN). If one or more nodes are unavailable, see nodetool removenode for instructions.

• Run a full cluster repair before nodetool removenode, so all existing replicas have the most up-to-date data.

• In the case of node failures during the removenode operation, re-run repair before running nodetool removenode (not required when Repair Based Node Operations (RBNO) for removenode is enabled).

After Upgrading from 2024.1¶

The procedure described above applies to clusters where consistent topology updates are enabled. The feature is automatically enabled in new clusters.

If you’ve upgraded an existing cluster from version 2024.1, ensure that you manually enabled consistent topology updates. Without consistent topology updates enabled, you must consider the following limitations while applying the procedure:

• It’s essential to ensure the removed node will never come back to the cluster, which might adversely affect your data (data resurrection/loss). To prevent the removed node from rejoining the cluster, remove that node from the cluster network or VPC.

• You can only remove one node at a time. You need to verify that the node has been removed before removing another one.

• If nodetool decommission starts executing but fails in the middle, for example, due to a power loss, consult the Handling Membership Change Failures document.