Process Engine High Availability

[fmasci]

Hi All,

Does someone has experience with process engine HA?
My scenario is a process engine instantiated programmatically into web application which run on two nodes with session replication.

My big doubt is what happens whether a process engine instantiates a job timer.
In case of the first node down, how can the second node manage this job?

Is there any document which explains these aspects?

Thank you so much.
Regards

[frederikherema1]

When 2 job-executors are running and one of them goos down while a job/multiple jobs were being executed, the jobs are marked in the database as "locked". After the configurable 'lockTimeInMillis' is expired (search forum for this parameter), the first job-executor will unlock the acquired jobs and run them.

The default timeout is 5 minutes. Note this can potentially cause a max delay of 5 minutes on timers that were being fired on the job executor that was down, but it also allows for jobs that run for more than a minute, not to be ran again because they appear to be "stuck" while the're just still executing.

So to conclude, if multiple job-executors are operating on the same database, and one goes down, the second one will take over the jobs that were executing or where aquired/locked by the job-executor.