JBERET-655: Handle the job execution status during a server crash gra…

[theashiot]

…cefully

https://redhat.atlassian.net/browse/JBERET-655

[theashiot]

Please note that this is only an initial step towards a fix. What this update provides:

• If you restart WildFly that crashed while a job was still in progress, the updated code detects unfinished jobs and puts them in a "stopped" state. In the HAL console, the job no longer displays an "unkown" state.
• However, at this point, if you press the "Restart" button, which is not greyed-out anymore, the job does not restart and you get a " JBERET000601: Failed to get job xml file for job bst1083-restartable.xml" exception.

The following work is still required to be done:

• Make the job restartable; ie when a user presses the "Restart" button in the HAL console, a new job execution should start.
• Replicate the updates for other supported repositories like mongo, in-memory etc where applicable.
• Refine the SQL queries used: filter in SQL for unfinished jobs rather than fetching all and filtering in code.
• Update the code to work in the case multiple nodes use the same job repository.
• Add tests.

best,
ashwin

[jamezp]

In the BatchEnvironment.getJobXmlResolver() we might want to filter the selection based on the jobs defined within the current environment and only changes those.

My thought is if this is a shared database with other batch jobs in different services, we don't want to update the state of those jobs.