How to Schedule Jobs to only run during a time window

Question

I have a long running task that needs to schedule jobs to process data.
I only want these scheduled jobs to start during a specific window of time each day, probably 10:00 PM to 6:00 AM.
If the scheduled jobs do not begin during the specified time frame, they must wait until the next day to start running.
Each scheduled job will only be executed once and then auto dropped.
How should I go about creating these scheduled jobs?

Answer 1

Hi Jeff,
I agree that the documentation isn't clear enough about the purpose of windows.
You can indeed use windows for changing the resource plan, but you can also use them for scheduling your jobs.
I did a simple test in real-time to illustrate the latter.
At around 10.30 am today I created a table that will populated by a job:
CREATE TABLE TEST_WINDOW_TABLE(EVENT_DATE DATE);
Then, I created a window whose start_date is today at 10.40 am :
dbms_scheduler.create_window(
                             window_name     =>'TEST_WINDOW',
                             resource_plan   => NULL,
                             start_date      => to_date('10/04/2014 10:40:00', 'dd/mm/yyyy hh24:mi:ss'),
                             repeat_interval => NULL,
                             duration        =>interval '5' minute
You can see that this window doesn't have a resource plan, and its repeat interval is NULL (so it will be opened only once).
The window will stay open for 5 minutes.
Finally, I created a one-off job whose schedule is the previously created window:
DBMS_SCHEDULER.create_job (
                           job_name      => 'TEST_WINDOW_JOB',
                           job_type      => 'PLSQL_BLOCK',
                           job_action    => 'BEGIN insert into test_window_table values (sysdate); COMMIT; END;',
                           schedule_name => 'SYS.TEST_WINDOW',
                           enabled       => true,
                           auto_drop     => true
Checking the user_scheduler_job_log before 10.40 would return no rows, which mean the job hasn't started yet since the window was not open.
Now, from 10.40, it shows one entry:
SQL> select log_date, status from user_scheduler_job_log where job_name = 'TEST_WINDOW_JOB';
LOG_DATE                                                                         STATUS
10/04/14 10:40:02,106000 +02:00                                                  SUCCEEDED
The TEST_WINDOW_TABLE has also got the row:
SQL> select * from TEST_WINDOW_TABLE;
EVENT_DATE
10/04/2014 10:40:02
Voilà.
In your case, since you want to run the jobs daily between 10 pm and 6 am (duration of 8 hours), the window would look like this:
dbms_scheduler.create_window(
                             window_name     =>'YOUR_WINDOW',
                             resource_plan   => NULL,
                             repeat_interval => 'freq=daily;byhour=22;byminute=0;bysecond=0',
                             duration        =>interval '8' hour
For your jobs, you may need to specify an end_date if you want to make sure the job gets dropped if it couldn't run in its window.