This article describes the design details of storm fault tolerance.
1. What happens when a worker process dies?
When the worker dies, supervisor will restart it. If it tries to turn on multiple failures and cannot send a heartbeat with Nimbus, Nimbus will redesign the worker to another machine.
2. What happens when a node dies?
The tasks assigned to this machine will time out and Nimbus will reassign these tasks to another machine.
3. What happens if Nimbus or supervisor Daemons dies?
Nimbus and Supervisor Daemons must be running under surveillance, and if Nimbus or supervisor Daemons dead, they will reboot like nothing has happened.
4. What happens if a single-node Nimbus fails?
If you lose the Nimbus node, the workers process will continue to run. Additionally, supervisors will continue to restart the workers process when these workers processes are dead. Then, no nimbus,workers will not be reassigned to other machines when you lose a working machine.
5, Storm guarantee data reliability processing?
Storm series of three--fault tolerance