Question

I have a CakePHP script that should hopefully be run by a cron job. It runs fine from the command line, but seemingly not from the cron. The cron line is something like:

*/2 * * * * cd /path/to/app;../cake/console/cake do_update

The script itself – and this is the bit that I think may possibly be too wacky, to the extent of throwing off the cron – loops through a subset of a Realtors table in the database, using the system time to decide which 50-record slice of the database to update:

$realtors = $this->Realtor->find('all',array(
'conditions'=>array('Realtor.zone_id'=>1),
'order'=>array('Realtor.num DESC'),
'limit'=>50,
'offset'=>date("i")*25
));

So my question(s) is/are – is there anything I’m doing here that would obviously throw the cron job for a loop? And, perhaps more importantly, is my method of splitting a database into manageable chunks over the course of an hour crazy? (I’m pretty much a programming newbie, so I try a lot of things without knowing whether they’re good practice or not.) Can anyone suggest a better way of looping through and updating large numbers of database records via a cron, that prevents the individual queries from being too huge for the system to handle?

EDIT: not only does it always work from the command line, it also works when run by cron script on a different server. I guess there’s just something messed up on the particular server, so it seems doubtful there’s a code-related solution! I’ll just accept an answer from among the useful cron-related insights below…

Answer 1

Generally, you’d want to put all your command line stuff into an shell file

/path/to/cake/console/cron.sh

#!/bin/sh
cd /path/to/app
php ../cake/console/cake do_update

*/2 * * * * sh /path/to/cake/cron.sh

Normally I don’t care about where the application actually starts (if i’m not using files) and just do

*/1 * * * * php /my/path/my_php.php