Recovering from a cluster shutdown
This recipe will cover the procedure to follow in case of a cluster shutdown. We discuss the split brain problem and also explain how to start a cluster without all storage nodes.
How to do it...
This section covers the procedure to follow—both for a partial failure (some nodes fail, but the cluster remains operational) and a complete failure (all nodes fail!).
- Partial cluster failure
In the event of a single node failing, you will notice the following:
- If the node that fails is a management node—no immediate problem occurs, but other nodes cannot restart and activities requiring a management node (online backups, centralized logging) will not take place
- If the node that fails is a storage node—assuming one node ...