Erlangelist
This repository contains the complete source code of The Erlangelist site:
Contributing
If you want to contribute a minor correction to some article (e.g. language and grammar fixes) just make a pull request to the master. If you'd like to make a bigger change, please open a GitHub issue first.
Running the site locally
Prerequisites:
Erlang, Elixir, and node.js. See ./.tool-versions for exact versions. You can also use asdf version manager to install these prerequisites.
Starting:
cd site
mix deps.get
pushd assets && npm install && popd
iex -S mix phx.server
If all went well, the server will listen on ports 20080 (http) and 20443 (https).
Deploying to the production server
Note: Don't deploy to a machine where you run some other system, or which is of some other importance to you. Setup script installs some packages and changes the system configuration (e.g. creates a user, modifies iptables rules, ...).
Server prerequisites
The target host must run Debian 8 and needs to have sudo
installed. It also needs the Internet access, because various packages, docker images, and other dependencies are fetched during the initial setup and subsequent deploys.
There is a support available for starting a local Vagrant box that can act as a "staging" server. You need to first run cd provisioning/vagrant && VM_IP=192.168.54.54 vagrant up
to start the server with a given IP address (feel free to use another IP address).
Client prerequisites
On your machine you need bash
, ansible
, and a git
client. You also need to configure automatic passwordless ssh login to the server machine (under a login that has sudo privilege).
If you started a local Vagrant box, as explained above, you can run cd provisioning/vagrant && vagrant ssh-config --host ip_of_the_vm
to get the ssh configuration for your server.
Filling in the blanks
You need to create provisioning/git_keys
folder that must contain public keys of all allowed deployers.
Initial setup
This step installs required packages and sets up the system. You can run provisioning/setup-server.sh server_ip network_interface
, where network interface is the name of the interface which accepts the requests from the outer world (for example eth0
).
For example, to set up a local VM with an IP address 192.168.54.54, you can run provisioning/setup-server.sh 192.168.54.54 eth1
.
Setting up the git remote
The deploy is performed by pushing to the git repository on the remote server. During the setup, the git
user is created on the server. You can login as a git
with any of the keys supplied in provisioning/git_keys
folder, as explained above.
In your local git repository, you need to add another remote which will be the deploy target. For example:
git remote add local_deploy [email protected]:erlangelist
Deploying
If all things are properly configured, you can simply deploy the site via
git push local_deploy master
The first push will run for some time, because it builds all the docker images from scratch. Once everything is done, the site should be listening at the port 80.