How do I remove a leaf node from the cluster?
Singlestore's distributed architecture allows you to add, remove or replace nodes in case of cluster issues or if you need to expand or shrink your cluster size to add or remove capacity.
You can review our full guide on how to Resize your Cluster.
In general, the most common reason to remove a leaf node is due to some kind of hardware or host level issue where there has been a failure or a need to do maintenance or replace hardware.
If a node fails in a cluster with High Availability active, the node fails over to its pair node to ensure that the cluster remains available and can continue to process the workload. Failover doesn't move any partitions to other nodes in the cluster, it only runs a series of partition promotions on the paired node in order to change their role to master so these partitions can process the workload.
In case the failed node can't be recovered (for example, due to failed hardware), you might want to consider removing this leaf node from the cluster, so it can be replaced and re-deployed.
For a full guide on how to remove and replace failed nodes, please review our Dealing with Cluster Failures guide.