The start_time of execution does not follow the chronology of the activation time JOBS.
When the same JOBP (same Job Plan, same Tasks, same Priority on same Jobqueue) is started multiple times with an interval of n seconds, it can happen that the start time of these Runs does not follow a FIFO rule based on the activation time.
This can result in a chaotic situation where, jobs that were activated at a later timestamps were executed and ended before JOBS with an earlier activation time.
This can have an impact on the production under specific circumstance where the activation chronology is critical.
Component : AUTOMATION ENGINE 12.3.3
This is a bug : A problem has been fixed where the SEQUENCE SQ_FIFO is changed and NOCACHE is now used instead of CACHE 1000.
This is a bug corrected in AE 12.3.5
Possible Workaround for those who can now immediately update:
If an Oracle RAC is used - it is necessary to make sure that all AE processes are connecting to the same NODE.
If no RAC is in place and all is setup as it should - the oracle sequence cache for the FIFO generation must be disabled:
ALTER SEQUENCE SQ_FIFO NOCACHE;