STONITH
also: fencing, STONITH device, fence agent
STONITH (Shoot The Other Node In The Head) is a fencing mechanism in high-availability clusters that forcibly isolates or reboots a failed node to prevent data corruption and split-brain scenarios.
STONITH is a critical safety feature in clustering software like Pacemaker that ensures only one node can access shared resources at any time. When a node becomes unresponsive or fails, STONITH forcibly removes it from the cluster by power-cycling, resetting, or isolating it—preventing the failed node from continuing to write data or hold locks.
Without STONITH, a network partition could create a "split-brain" scenario where multiple nodes think they're the active cluster leader, causing data corruption in shared storage. For example, in a database failover, if node-A fails but node-B cannot verify it's truly dead, both might try to write to the database simultaneously—corrupting it.
STONITH is implemented via fence agents—specialized tools that interact with hardware like IPMI, PDUs (power distribution units), or cloud APIs to power-cycle or isolate the offending node. Most production high-availability clusters require STONITH to be enabled for data safety.