In this post we are going to see what is split brain syndrome in clusters?
Split brain syndrome is a state on a cluster where each cluster nodes are divided into small clusters and each one will believe themselves as itself only an active cluster.
Each nodes believes that all other nodes are dead and simultaneously will try to access the same data/ disks, which can be lead to data corruption. This situation will occur during the cluster reformation.
When one or more nodes fails, the cluster itself will reform the cluster with the available nodes.
Note: High Availability clusters will use some mechanism like CMAN, Pacemaker, HP ServiceGaurd, and Linux HA to avoid split brain syndrome.
Common methods to address split brain syndrome:
- I/O Fencing
- Quorum/ Local Disk
- Quorum Server
- Tie – Breakers
- STONITH(Shoot The Other Node In The Head)