org.jberet.operations.JobOperatorImpl#getRunningExecutions currently retrieves all JobExecutions from the underlying job repository and then filter them by jobName and batch status. It may load too much unneeded data into memory that will just be filtered out later. When there are large number of job executions in the database, this loading may cause OOME.

For JdbcRepository, it means a query to retrieve all job executions. We should be able to pass more specific criteria to job repository to allow for repository-specific optimization. For example, a more specific query to select by jobName and batch status.