I'm currently developing a bottle app within virtualenv. I intend to serve it using bjoern WSGI server (but that probably doesn't matter too much). I also intend to serve the app with a lighty or nginx reverse proxy. Anyhow, can the app be run from within its own virtualenv as a system service? And if so, how would one go about it?
问题:
回答1:
According to my experience, I suggest that you can use Supervisord to run your web server as daemon service. Although you can write some Linux service scripts in /etc/init.d, but they are really difficult to do it correctly. Here is an example init.d script for nginx to run it as a service in Ubuntu. You don't want to write one, do you?
To run a python server which depends on virtualenv as daemon service with supervisord, here is the configuration I am using in a production environment.
[program:web01]
command=/home/victorlin/tg2env/bin/paster serve production.ini ;
process_name=%(program_name)s ;
directory=/home/victorlin/ ;
user=victorlin ;
priority=999 ;
redirect_stderr=true ;
stdout_logfile=/home/victorlin/logs/web01_out.txt ;
stderr_logfile=/home/victorlin/logs/web01_err.txt ;
environment=PYTHON_EGG_CACHE=/home/victorlin/.python-eggs ;
You can use /path/to/virtualenv/bin/python to run your own python script in the command field. And, to run the supervisord on start-up, you can write crontab like this in your root account:
@reboot /usr/local/bin/supervisord -c /home/root/supervisord.conf
Of course, if you don't have port numbers lower than 1024 to open, you can write this start-up crontab in a non-privilege account.