On Wed, 30 May 2007 22:00:27 +1200
Nick Rout <[EMAIL PROTECTED]> wrote:

> I want to have a script updating a web page. Basically it downloads a 
> podcast file, amends some tags and saves it to a directory in apache's 
> htdocs, (and from there I have other podcast reading software read it.) 
> It needs to be run via cron.
> 
> I don't want to run the script as root, but the htdocs area is all owned 
> by root.
> 
> I seem to be completely clueless about how to get the script to be able 
> to update in the htdocs hierarchy, it could be the headache i have, or 
> the headache could be caused by thinking about it. I can only come up 
> with either changing ownership of directories inside 
> /var/www/localhost/htdocs (yuk) or running the script as root (yuk).
> 
> So someone make a suggestion and save my brain.


Your http process will either be running as nobody or www-data. With the 
exception of any upload directories, all it requires is read access to the 
relevent apache directory tree. Now some prople contend that these files should 
also be owned by nobody/www-data, but I contend that they *DONT*, as it's then 
an extra hurdle that any hacker needs to jump through before screwing with your 
website. 

So, make the htdocs diretriry(ies) owned by yourself, and ensure that 
directories have read/execute permission set for other, and read for normal 
files. And no write permissions anywhere unless a log file or an upload 
location.

That way you can run anything from your own cron.

Steve

Reply via email to