Problem:
- any node in a heterogeneous cluster try to execute a delete process instance batch job even if the node is not registered for the deployment
- as result, the batch job may fail because of missing classes
AT:
- in a heterogeneous cluster, the delete process instance batch job is only executed on a node that is registered for the related deployment
Hints:
- set the deployment id of the batch job - see org.camunda.bpm.engine.impl.batch.AbstractBatchJobHandler#createJobs
- the behavior is already implemented for process definition migration jobs
- check the behavior of the other batch jobs
What is the expected behavior when the historic process instance query returns instances from different deployments and the batch configuration is set so that one job executes two or more deletions?