Hello,
I noticed the following behavior.
1) stop nginx (/etc/init.d/nginx stop on a debian box)
2) remove all logs in /var/log/nginx/
3) start nginx
now /var/log/nginx/access.log and /var/log/nginx/error.log are owned by
root:root.
in fact it's the same for all other vhosts log files.
4) issue kill -USR1 `cat /var/run/nginx.pid`
now /var/log/nginx/access.log and /var/log/nginx/error.log are owned by
www-data:root, where www-data is nginx's user as per /etc/nginx/nginx.conf
configuration file.
same for all vhosts log files.
Why is it so?
Gregory
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.nginx.org/pipermail/nginx/attachments/20120311/a83ad235/attachment.html>
Hello!
On Sun, Mar 11, 2012 at 05:18:39PM +0100, Gr?gory Pakosz wrote:
> Hello,
>
> I noticed the following behavior.
>
> 1) stop nginx (/etc/init.d/nginx stop on a debian box)
> 2) remove all logs in /var/log/nginx/
> 3) start nginx
>
> now /var/log/nginx/access.log and /var/log/nginx/error.log are owned by
> root:root.
>
> in fact it's the same for all other vhosts log files.
>
> 4) issue kill -USR1 `cat /var/run/nginx.pid`
>
> now /var/log/nginx/access.log and /var/log/nginx/error.log are owned by
> www-data:root, where www-data is nginx's user as per /etc/nginx/nginx.conf
> configuration file.
>
> same for all vhosts log files.
>
> Why is it so?
When reopening log files nginx has to do this from inside worker
processes, and to be able to do so it chown's them to a user under
which nginx worker processes are running.
Maxim Dounin