What happens when a node or disk fails?

When a failure occurs in a Swarm cluster, other nodes in the cluster begin servicing the content found on the failed node or disk. Assuming all content on that node or disk has been replicated at least once, the cluster will continue to respond to requests for those UUIDs. In addition, the cluster will attempt to write additional replicas of the content from the failed node or disk so that the replication constraints continue to be satisfied. See the Swarm Guide Data Protection for more information.