[ic] "Excessive size of job queue, stopping"

[David Christensen]

On Mar 7, 2010, at 4:27 PM, Grant wrote:

> I'm on 5.6.2 and I get over 500 "Excessive size of job queue,
> stopping" messages in my global error.log every day.  They all appear
> within 20 minutes of each other.  I understand that I may be adding
> too many jobs to the queue, but I only run about 25 per day so I'm
> wondering why the huge number of error.log messages?
>
> Does anyone know what the jobsqueue limit is, and if it can be  
> increased?


It looks like it's currently hard-coded to 20.  This check occurs  
during the housekeeping() subroutine, so you're going to see the  
message every time you hit enough processes that the housekeeping()  
routine is called, which will be in part determined by your traffic  
setting.  I don't know if there's any impact to increasing that hard- 
coded limit, or if there is a benefit in making that tweakable by the  
user, but perhaps it's worth some experimentation on your part.

The other option is to divide your jobs up into multiple categories  
rather than just sticking them all in the same jobs/<dir>  
subdirectory.  This can be useful if the jobs can be grouped logically  
and if they share similar characteristics (such as likely to need to  
be run in the same frequency).  I must confess that most times I've  
just ended up using cron rather than IC's jobs feature, so YMMV.

Regards,

David
--
David Christensen
End Point Corporation
david at endpoint.com