depl
is a pre-alpha prototype. We want to start a discussion about how
developers deploy. Please tell us why you think depl is not the right tool
for you!
Deploying stuff is hard, managing nginx and postgres painful. Why don't we
solve that problem? Simple deployments should not involve tweaking
configurations that you have never dealt with before. depl
solves this.
Let's deploy Django to a live
server! Create a file called depl.yml
:
deploy: - django: settings: settings_prod ssl: key: /path/to/key certificate: /path/to/cert hosts: - myhost.com
And run depl deploy
in that same directory. After this two step process
your website will be running on https://myhost.com
and
http://myhost.com
. It only works on Ubuntu servers for now (maybe also
Debian), but will work on other systems in the future as well.
depl
would connect to myhost.com
by SSH, install all necessary software
packages and initialize your project. In our Django example this would involve
installing nginx
, uwsgi
, postgresql
(if required by Django's
settings.py
). It would install all Python dependencies and run on your
production settings.
Currently this is also working for Meteor and WSGI (which includes flask).
Have you ever tried those tools? ansible, puppet and chef are really complicated to understand. They are made for big deployments, for large scale operation engineers. You can read the blog posts below, they explain a bit more. Generally depl solves the use case for all small deployments on your own server, those tools do something different.
Docker and Vagrant are completely different tools. They make controlling VMs easy. They do not install your software. However, depl could be used in combination with either of them.
If you want to know more in general, just read the documentation at http://depl.rtfd.org!
Currently depl supports Ubuntu/Debian (apt-get), the others:
- Arch Linux (pacman) could be implemented
- Fedora/CentOS/RHEL (rpm) could be implemented
- Windows/Apple: Discussions needed (but probably not?!).
Depl is written in Python. Python 3 support will be ready once fabric supports Python 3.
- DNS configuration
- Central logging
- High Availability tools like HAProxy to make it easy not to go offline for a few seconds while deploying.
- Could generate e.g. an ansible recipe with depl. Possible?
- Write a "depl server" with different backends (e.g. amazon clouds) to automate the process of deploying and increasing servers.
- Add support for automatic travis testing with depl files. Is that even possible?
Read CONTRIBUTING.md to check how you can contribute! There's also a small developer documentation available.
Use a virtual machine for testing, please. Why?