When the batch subsystem is configured to use the JDBC store to store execution metadata, accumulating large number of records can lead to degraded performance when deploying an application that uses batch jobs.

For more details, see discussion in WFLY-14275.

Report is that the database contains 6 jobs, with 100 000 job instances each and single job execution per instance. Deployment times can increase to several minutes. Most time seems to be spent on data retrieval from the database, actual queries are quite fast. Database latency appears to be a factor.

Another issue, apart from the deployment time, is that the Web Console is unable to load job execution data, so when user tries to display deployment details, the operation fails with OOM exception. The Web Console UI is not designed to display that number of records anyway.

Loading the job execution data via CLI works fine.

Proposed solution is to add an attribute in the batch subsystem, that would allow to limit the number of records loaded by the JDBC store.