Scheduled Jobs can be lost track of when a database operation fails at the moment the job is claimed

Summary

Scheduled Jobs can be lost track when a database operation fails at the moment the job is claimed by the Scheduler.

This only affects a single-node setup in Server, as in JIRA Data Center the "recovery" job will bring back to memory the lost jobs.

Environment

• JIRA Server (Single Node - Non Data Center)
• JIRA Cloud.

Steps to Reproduce

1. Schedule a Job.
2. Introduce a breakpoint at CaesiumSchedulerService.executeClusteredJob
3. Interrupt database connectivity.

Actual Results

1. JIRA will lose track of the job until a restart is performed.

Expected Results.

1. JIRA will run the job under its schedule as soon database connectivity is resumed

Notes

This is the root caused behind the issue reported at JSD-3653 but could affect any scheduled job in JIRA.

Workaround

Restart JIRA.