Forum: NGINX nginx using access.log.1 after using a logrotate script under ubuntu (from source)

Hi

I've built nginx from source on ubuntu in /usr/local/nginx according to 
the
tutorial:
http://articles.slicehost.com/2009/3/4/ubuntu-intrepid-installing-nginx-from-source

and used the init.d script:
http://articles.slicehost.com/2009/3/4/ubuntu-intrepid-adding-an-nginx-init-script

read this about nginx log rotation:
http://www.ruby-forum.com/topic/134115

when forcing logrotate with: sudo logrotate -f /etc/logrotate.conf
nginx starts using access.log.1
but when using a signal manually nginx uses access.log: sudo kill -USR1 
`cat
nginx.pid`

If I'm using 'nginx' user in the logrotate script I'm getting an error 
that
this use doesn't exists so I have to use root.
My logrotate script under /etc/logrotate.d/nginx
/usr/local/nginx/logs/*.log {
        weekly
        missingok
        rotate 52
        compress
        delaycompress
        notifempty
        create 640 nginx adm
        sharedscripts
        postrotate
                [ ! -f /usr/local/nginx/logs/nginx.pid ] || kill -USR1 
`cat
/usr/local/nginx/logs/nginx.pid`
        endscript
}

What am I doing wrong?
Thanks

Maybe you used a tooooooooooold version?
Try the current!

On Sun, Mar 7, 2010 at 10:13 AM, pablo platt <pablo.platt@gmail.com> 
wrote:
> Hi
>
> I've built nginx from source on ubuntu in /usr/local/nginx according to the
> tutorial:
> http://articles.slicehost.com/2009/3/4/ubuntu-intrepid-installing-nginx-from-source

--
Ren Xiaolei

Are you sure, that  /usr/local/nginx/logs/nginx.pid has actual pid of
the master process?

2010/3/7 pablo platt <pablo.platt@gmail.com>:

2010/3/7 Борис Долгов <boris@dolgov.name>

> Are you sure, that  /usr/local/nginx/logs/nginx.pid has actual pid of
> the master process?
>

Yes according to the tutorial at
http://articles.slicehost.com/2009/3/4/ubuntu-intrepid-installing-nginx-from-source

You may want to try the current version from Jeff Waugh's (development) 
PPA:
https://launchpad.net/~jdub/+archive/devel

He has a stable PPA, too, but nginx 0.8.x is often used in production 
already.


Cheers,
Daniel

On Sun, Mar 7, 2010 at 3:13 AM, pablo platt <pablo.platt@gmail.com> 
wrote:
> http://www.ruby-forum.com/topic/134115
>         weekly
>         endscript
> http://nginx.org/mailman/listinfo/nginx
>
>



--
http://daniel.hahler.de/

>> Are you sure, that  /usr/local/nginx/logs/nginx.pid has actual pid of
>> the master process?
>
> Yes according to the tutorial at
> http://articles.slicehost.com/2009/3/4/ubuntu-intrepid-installing-nginx-from-source

Have you verified it, e.g. by catting it?

>> but when using a signal manually nginx uses access.log: sudo kill -USR1 `cat nginx.pid`

Here you aren't using the full path. Where have you executed this from?


Cheers,
Daniel

--
http://daniel.hahler.de/

I'll use the PPA but want to understand why my logrotate doesn't work.
Is it possible to install the PPA and use it reload nginx without 
loosing
existing connections?
Do I need to use 'make uninstall' on the previous nginx installation 
first?

On Mon, Mar 8, 2010 at 3:33 PM, Daniel Hahler <genml@thequod.de> wrote:

> >> Are you sure, that  /usr/local/nginx/logs/nginx.pid has actual pid of
> >> the master process?
> >
> > Yes according to the tutorial at
> >
> http://articles.slicehost.com/2009/3/4/ubuntu-intrepid-installing-nginx-from-source
>
> Have you verified it, e.g. by catting it?
>

What do you mean by verifying by catting it?


> >> but when using a signal manually nginx uses access.log: sudo kill -USR1
> `cat nginx.pid`
>
> Here you aren't using the full path. Where have you executed this from?
>

I've executed it from the /usr/local/nginx/logs folder