User_jobs: Job reappears after bouncing the db

Question

Hi All,
This week, I've come across a strange thing. One of the developer has submitted a job using dbms_job pkg. But for some unknown reason, he wanted to stop it from running. So, from TOAD-- Schema Browser, he dropped that job and queried user_jobs to find "no rows".
But after few minutes, the job reappeared in USER_JOBS with its Next run date/time already passed by. He removed the job again. But it reappeared after sometime.
On which we tried this:
1) Shutdown db
2) set job_queue_process=0 (Earlier it was 4)
3) Start db
4) Shutdown db
5) set job_queue_process=4 (again)
6) Start db.
This solution worked for us.
My question is:
Why did it reappeared in USER_JOBS even after dropping the job?
Where can we see the status of a job (other than DBA_JOBS, DBA_JOBS_RUNNING)?
Why did it take the above 6 steps to resolve the issue?

Answer 1

> Why did it reappeared in USER_JOBS even after
dropping the job?
As David indicated - no commit. Changing the jobs table requires a commit.
One may want to add a job as part of a business transaction. The job must only be scheduled when that business transaction is committed. One does not want the job to be auto added to the job queue without an explicit commit.
Thus the DBMS_JOB package does not perform any implicit commits or autonomous transactions. As documented.
> Where can we see the status of a job (other than
DBA_JOBS, DBA_JOBS_RUNNING)?
In V$SESSION. That will list the actual Job Queue Session that is executing the DBMS_JOB. The actual o/s process details for that session will in turn be listed in V$PROCESS.
> Why did it take the above 6 steps to resolve the
issue?
Ignorance, sadly.