[Backgroundrb-devel] timeout resp. handling of crash situations

[Tobias Overkamp]

Am 12.03.2009 um 18:57 schrieb hemant:

> On Thu, Mar 12, 2009 at 8:13 PM, Tobias Overkamp
> <tov-backgroundrb at overkamp.org> wrote:
>> [...]
>> I want to make use of the "enq_method" call to make sure that no  
>> request
>> will get lost but I can not see any mechanism for a housekeeping  
>> after a
>> crash. Do I miss something or do I have to query the database  
>> directly?
>>
>
> Well if worker crashed for any reason, your unfinished tasks will
> remain marked as unfinished  in bdrb_job_queue  table and hence on
> restart you can decide what do do with them.

Hi,
but do you have any suggestion how to do this via backgroundrb?
If I understand it correctly, it seems that a "taken" job will remain  
into this state until I finish it (using the finished method) or I  
perform an SQL update. Of course the SQL wouldn't be that hard but I  
was just searching for an already implemented feature in backgroundrb.
Jonathan Wallace has just recommended me to use "god" as a supervising  
daemon for backgroundrb (thanks for this, btw. :-) ). This will make  
sure that the workers keep running after crashes, but will it clean up  
the taken jobs?

Don't get me wrong: I think I'll find a solution for tidying up (and  
of course I hope that my server will not crash that often) but somehow  
I have the feeling that this functionality should be part of  
backgroundrb.


Regards,
Tobias.