Deploy pattern

[wsantos]

Sorry create this thing as an issue, but I'm wondering what pattern use for a production deploy, what you use Jesse? You blog is a git clone of master branch ? Do you suggest any pattern ?

[ajdavis]

Yeah, I guess I could document this better.

emptysquare.net is a Rackspace server. I made a Python 2.7 virtualenv on it, cloned Motor-Blog's master branch there, and installed its requirements.txt.

I put a copy of motor_blog.conf.example in ~/etc/motor_blog.conf and customized it a bit.

I put a copy of the example supervisord.conf in ~/etc as well.

Same for nginx.conf.

NGinx and Supervisord are started from /etc/init.d scripts.

Supervisor runs mongod and four instances of Motor-Blog's server.py.

Supervisor, MongoDB, and Motor-Blog are all configured to log to ~/log/.

That's about it. The scripts in /etc/ are a good starting point. I can add my init.d scripts there too.

[ajdavis]

Oh, supervisord is installed into its own virtualenv with pip, so the init.d script that runs it from there.

[ajdavis]

Deployment is a hack: I ssh into emptysquare.net, git pull the latest Motor-Blog, then restart server.py with supervisorctl.

[wsantos]

Thanks ajdavis.

[ajdavis]

I'm actually going to keep this open until I've updated the documentation. I could be useful to people. I also want to review @bdarnell's tornado-production-skeleton for ideas.