When there is a failure in job itself rather than a non-zero exit code of the job execution, the alert doesn't contain any hostname so you don't know what rundeck instance it occurred on.

Could you include a field containing the rundeck hostname on all alerts?