Decommissioning StorageOS Nodes

StorageOS nodes can be decommissioned and removed from the cluster using the StorageOS CLI.

This functionality is only available when StorageOS is deployed with KV_BACKEND=etcd, so the KV store is external to StorageOS. Please see the deploy etc-as-svc example for more information for more information.

There are safeguards to make sure data is not lost unintentionally. Only nodes in state Offline can be removed from the StorageOS cluster. Note that once removed from the cluster, nodes may not partake in StorageOS operations, and may not run container applications that require StorageOS backed persistent storage.

The recommended procedure is as follows.

  1. Cordon the node

     $ storageos node cordon node03
     node03
    
  2. Drain the node

     $ storageos node drain node03
     node03
    

    Wait until the node drain is finished. Check the volumes located on that node with storageos node ls and wait until there are no Masters or Replicas on the drained node. If there are no eligible nodes for replicas to be created on, the drained node will keep hosting them.

  3. Stop the node

  4. Delete the node from the cluster

     $ storageos node delete node03
     node03