Backgroundrb devel - Jul 2007 - Scheduler skipping schedules

There is a line in the backgroundrb codebase:

sleep 0.1  # TODO this is dangerous; could skip over some jobs

Could this explain why my scheduler stops triggering?

It only seems to work for about 12 - 24 hours before schedules are
being skipped...

Once I changed the times to using cron trigger in
backgroundrb_schedules.yml and set the times to every 37th and 59th
minute of the hour, instead of the previous repeat_interval of
60.minutes it seemed to improve but it still stops being triggered
within a day and doesn''t seem to start again unless I stop and start
background rb...

Sometimes this is both my schedules, sometimes just one of them...
Other tasks that are directly triggered still work...

This is really driving us a bit nuts as we can''t get to the bottom of
it and nobody else seems to have any ideas?

I did see a post on an Exception not being caught and took care of
that but it didn''t seem to make any difference... The code is
thoroughly tested and I really doubt it''s the application code unless
the sheer volume of data could affect it?

On 7/19/07, Levent Ali <levent at leventali.com>
wrote:
> There is a line in the backgroundrb codebase:
>
> sleep 0.1  # TODO this is dangerous; could skip over some jobs
>
> Could this explain why my scheduler stops triggering?
>
> It only seems to work for about 12 - 24 hours before schedules are
> being skipped...
>
> Once I changed the times to using cron trigger in
> backgroundrb_schedules.yml and set the times to every 37th and 59th
> minute of the hour, instead of the previous repeat_interval of
> 60.minutes it seemed to improve but it still stops being triggered
> within a day and doesn''t seem to start again unless I stop and
start
> background rb...
>
> Sometimes this is both my schedules, sometimes just one of them...
> Other tasks that are directly triggered still work...
>
> This is really driving us a bit nuts as we can''t get to the bottom
of
> it and nobody else seems to have any ideas?
>
> I did see a post on an Exception not being caught and took care of
> that but it didn''t seem to make any difference... The code is
> thoroughly tested and I really doubt it''s the application code
unless
> the sheer volume of data could affect it?
Hi Levent,

We are working on a new core that hopefully fixes these issues with
scheduling(I have myself observed problems with scheduling and
difficult to reproduce cases), in meanwhile I would see if this
problem can be fixed with a workaround.