Decryption of retry rules after Oracle Database JOB failure

Since no relevant instructions are found in the official documents, the following tests are conducted here to find the retry time rule for oracle Database jobs that fail.

Database Version: 11.2.0.3

Test Description: A log table and procedure with certain errors are created for running a job. Here, you only need to record the next_date in the figure user_jobs for each job execution to infer the retry rule after the job fails to be executed.
To test the retry rules of jobs, I have done the following work:
Log table and sequence:

Create table job_exec_logs (id number, current_date date, next_date date, failures number, broken varchar2 (2); create sequence seq_job_exec_logs_id; Test procedure

create or replace procedure pro_my_test isbegin  insert into job_exec_logs selectseq_job_exec_logs_id.nextval , sysdate , next_date , failures , broken from user_jobs ;  commit ;  execute immediate 'select * from ddddsfs' ;end ;

Ddddsfs indicates that the Stored Procedure pro_my_test does not exist. That is to say, errors will occur when the Stored Procedure pro_my_test is run (but the log table can still be inserted normally)
Create job:

Var job number; begin sys. dbms_job.submit (job =>: job, what => 'Pro _ my_test; ', next_date => sysdate, interval => 'sysdate + 5'); commit; end; /In fact, there is nothing to say about the entire test process. Let the job run automatically, but wait for the next retry time of the job. Test result:

Dexter @ REPO> select trunc (next_date-lag (next_date, 1) over (order by 4) * 24*60) from (2 select * from job_exec_logs 3 union all 4 select 11111, sysdate, next_date, failures, broken from user_jobs 5 order by 4 nulls first); TRUNC (NEXT_DATE-LAG (NEXT_DATE, 1) OVER (order by 4) * 24*60) ------------------------------------------------------- 2 4 8 15 32 64 128 256 512 1024 1440 already selected 13 rows.The above results and other tests of the author can be judged as follows: 1. The retry time increases progressively. The first push is 2 minutes, 4 minutes, 8 minutes, 16 minutes. 2. When the time exceeds 1440 minutes, that is, 24 hours, the fixed retry time is 1 day. 3. I also learned from other experiments that after more than 16 retries, the job will be marked as broken and next_date will be 4000-1-1, that is, no job retry will be performed. 4. After the oracle database retry time reaches the set next execution time, the set job execution time prevails.For more information, see metalink.Broken Jobs and the Job Queue (Document ID 103349.1)