When a Job has a Chain, the next run date can be skipped

Question

I have read more than once in this forum that if a job is still running when the next run date arrives, the job is internally queued and runs as soon as the previous run is finished. The following code was posted to demonstrate :
begin
dbms_scheduler.create_job('j1',
  job_type=>'plsql_block', job_action=>'dbms_Lock.sleep(70);',
  repeat_interval=>'freq=minutely;bysecond=0',enabled=>true);
end;
/On both 10.2.0.4 and 11.2.0.1, the job starts every 70 seconds. Fine.
Now, if I create a job using a PROGRAM and a CHAIN, then the job skips every other run date, so it runs every two minutes. Here is my test case:begin
dbms_scheduler.create_program (
   program_name => 'P1',
   program_type => 'PLSQL_BLOCK',
   program_action => 'DBMS_LOCK.SLEEP(70);',
   enabled=>true);
dbms_scheduler.create_chain (chain_name => 'C1');
dbms_scheduler.define_chain_step('C1', 'STEP1', 'P1');
dbms_scheduler.define_chain_rule ('C1', 'TRUE', 'START STEP1');
dbms_scheduler.define_chain_rule ('C1', 'STEP1 COMPLETED', 'END STEP1.ERROR_CODE');
dbms_scheduler.enable('C1');
dbms_scheduler.create_job (
   job_name => 'J1',
   job_type => 'CHAIN',
   job_action => 'C1',
   repeat_interval => 'freq=minutely;bysecond=0',enabled=>true);
end;
/Is this a bug? I need the chain because my real job has two steps, so is there anything I can do to get the "non-skipping" behavior with a chain? Thanks in advance,
Stew Ashton

Answer 1

Replying to myself :)
While walking to my car, I realized that my "chain" logic could be replaced by an anonymous block with two lines in it!
I guess I outsmarted myself by trying to use all the appropriate features of the Scheduler instead of doing things more simply, but I still think the behavior I observed is incoherent.
I am leaving this question "unanswered" since I won't be able to test my simpler approach until Monday.
Stew Ashton