- We assume we have a JobScheduler cluster configuration.
- If the database server is very slow, e.g. takes more than 1 minute to respond, then a warning is raised that a thread lock is blocked for more than 15s:
and then the heartbeat-watchdog-thread which should abort the JobScheduler if the last heartbeat is too old doesn't work.
- The heartbeat-watchdog-thread should abort the JobScheduler if it is necessary even if a thread lock occurred.
- To activate this behavior the default setting for automated restart has to be disabled by use of the following setting with ./config/scheduler.xml (see
- If looping through pending objects takes more than 2 minutes in a cluster configuration then JobScheduler will perfom the following actions:
- Kill all child processes
- Remove its PID file
- Abort operation and terminate the JobScheduler instance
- A backup instance will become active (if configured)