"sudo service uwsgi restart" returns [fail] however uwsgi logs show processes have been created and python app is running

Question

I'm running a local vagrant VM, Ubuntu 13.10 with nginx reverse proxying to uwsgi.

Running

sudo /etc/init.d/nginx status 

returns

* nginx is running

However running

sudo /etc/init.d/uwsgi status 

returns

* which one?

If I take a look at the log file for the wsgi app I can see that uwsgi is running, worker processes have been created etc... so is there a hidden instance of uwsgi running somewhere that's confusing the service restart command?

I installed uwsgi using:

apt-get install uwsgi

Here's the app.ini file:

[uwsgi]
plugin = python
socket  = /run/uwsgi/app/myapp/socket
chdir = /var/www/myapp
module = project.wsgi:application
pidfile = /tmp/myapp-master.pid
master = True
daemonize = /var/log/uwsgi/myapp.log

Any ideas?

Answer 1

Changing the pidfile option to pidfile2 seems to fix this issue.

pidfile2 = /tmp/myapp-master.pid

Interestingly the service uwsgi stop returns [OK] but the service uwsgi start returns [fail]

So I'm assuming the error happens when a non privileged user (i.e. www-data) is trying to write to the pidfile which has been created by a privileged user (e.g. root).

pidfile2 will create the pidfile after privileges drop - so www-data can happily write to it.

If someone else can shed light on whether this is the case that would be great.