The Primary Node Fails
Read this chapter for more information about single node failures.
Procedure
If the primary node in a three node cluster fails, the other nodes continue to operate as usual and one will take over responsibility as the new primary node.
When the failed node comes back online, it automatically synchronizes as secondary node with the other nodes before it is allowed back into the cluster.
Note:
Before restarting the failed node, please check the
CLUSTER_URL setting and unset
CLUSTER_BOOTSTRAP. No data is lost in single node failures.