Job does not repeat schedule if job ends outside a given time-range

Request: (see also http://sourceforge.net/projects/jobscheduler/forums/forum/486122/topic/3534239)

I have setup a job series. The first job in the series is setup as follows:

<?xml version="1.0" encoding="iso-8859-1"?>
<job>
<script language="shell"><![CDATA[d:\bin\runJob1.bat]]></script>
<run_time>
<period begin = "00:01" end = "23:15" repeat = "1800"/>
</run_time>
<commands on_exit_code="success">
<start_job job="Job-StandardJob2"/>
<start_job job="Job-StandardJob3"/>
</commands>
</job>

The other 2 jobs are setup as follows:

<?xml version="1.0" encoding="iso-8859-1"?>
<job>
<script language="shell"><![CDATA[d:\\bin\runJob2.bat]]></script>
</job>

<?xml version="1.0" encoding="iso-8859-1"?>
<job>
<script language="shell"><![CDATA[d:\\bin\runJob3.bat]]></script>
</job>

The job series runs fine most of the days, repeating every 30 mins. But some days around midnight the job series will not reschedule. As you can see below, the job series had to be manually restarted at 8:23 am. For some reason the last execution at 23:41 completed but the next run was never scheduled.

67385 period_repeat 2010-01-28 09:06:00 0 2010-01-28 09:22:33
67376 queue_at 2010-01-28 08:23:45 0 2010-01-28 08:36:00
67280 period_repeat 2010-01-27 23:41:15 0 2010-01-28 00:00:40
67267 period_repeat 2010-01-27 22:54:31 0 2010-01-27 23:11:14

I see in the logs for the run 67280 that the scheduler tried to stop the job as the period ended. Maybe this caused the problem? Some days the same scenario happens (the job is running at midnight and the scheduler tries to stop the job) but it doesn't cause the series to stop scheduling.

2010-01-27 23:59:00.045 [info] SCHEDULER-278 <period> ended, terminating task
2010-01-27 23:59:00.045 [info] SCHEDULER-914 end() called, task will terminate
2010-01-27 23:59:05.045 [info] SCHEDULER-278 <period> ended, terminating task
2010-01-27 23:59:05.045 [info] SCHEDULER-278 <period> ended, terminating task
2010-01-27 23:59:09.607 [info] SCHEDULER-278 <period> ended, terminating task
............
2010-01-28 00:00:40.558 [info] Job finished with error code 0
2010-01-28 00:00:40.558 [info] SCHEDULER-915 Process event
2010-01-28 00:00:40.558 [info] SCHEDULER-965 Executing command <?xml version="1.0" encoding="ISO-8859-1"?><start_job job="Job-StandardJob2"/>
2010-01-28 00:00:40.605 [info] SCHEDULER-965 Executing command <?xml version="1.0" encoding="ISO-8859-1"?><start_job job="Job-StandardJob3"/>

Please help.

Answer:

Hello srgadi,

Are there any holidays defined in your Job Scheduler installation which might cause this?

Can you please send a complete scheduler.log file to info(at)sos-berlin.com ?

Best Regards,

Andreas

Reply:

Thanks for your response. There is no holidays scheduled in the scheduler.

There was a typo in my original post. The job series timing was setup as follows when the problem of non rescheduling occurred:
<period begin = "00:01" end = "23:59" repeat = "1800"/>

Last Thursday I changed it to the following to check if not running the job so close to midnight will help.
<period begin = "00:01" end = "23:15" repeat = "1800"/>

I haven't had the same problem since, although even with original setting the problem used to occur only one in 2 weeks or so. However, the job series is now skipping a day for some reason.
88433 period_repeat 2010-02-01 00:38:21 0 2010-02-01 00:44:19
88420 period_repeat 2010-02-01 00:01:00 0 2010-02-01 00:08:21
88167 period_repeat 2010-01-30 23:14:52 0 2010-01-30 23:20:05
88157 period_repeat 2010-01-30 22:38:34 0 2010-01-30 22:44:52

I will e-mail the scheduler.log shortly. Thanks for your help.

preliminary result of investigation:

If the end of the task is outside the defines time-range for the job it seems to be that the job is ended but never automatically scheduled again. If the job is rescheduled manually the expected behaviour is resumed.

Workaround for 0-24-time-range: set the time-range to 0 to 24 instead of 0:01 to 23:59 ) This workaround does
not help if the time-range is e.g. 1:00 to 22:00.

Check, wether the "let_run" attribute of the job-tag has an impact on this bug.