A simple template for a app with users

If you're looking to kickstart your Django application and have it up and running in a production environment within a few hours, consider forking the code from this repository. For comprehensive step-by-step guidance and additional alternatives and options, check out the accompanying blog post. It provides in-depth insights into each of the steps.

Local installation

First clone or fork the repository. Your own repository might be private.

$ git clone https://github.com/nhatcher/users-template.git

After that create a virtual environment and install requirements

$ cd users-template/
$ python -m venv venv
$ source venv/bin/activate

Then you need to apply the migrations. This will create an sqlite3 file in server/.

(venv) $ cd server
(venv) server $ pip install -r requirements.txt
(venv) server $ python manage.py migrate

Create a superuser

(venv) $ python manage.py createsuperuser

Install caddy, only the binary in necessary. Note in particular that caddy shouldn't be running as a daemon in your local machine. I normally download the binary, copy it to /opt/caddy/ and add the path to PATH so it can be run in a terminal.

Running in a development environment

You need to have three servers running:

A proxy server:

$ caddy run

The front end:

frontend_test$ python -m http.server 5173

The django server:

(venv) server$ python manage.py runserver

That's it, you can visit now http://localhost:2080/ and test your application locally!

Running tests:

That will run the linter, formatter, type checker and tests

(venv) $ ./run_tests.sh

Create necessary accounts

1. Get your own domain name with Namecheap (will cost you ~10€ per year)
2. Create a Zoho account and link it to your domain name (free). In your zoho account create two email accounts. One for you the superuser and one for 'support@example.com'
3. Create an account in Sentry (free).
4. Create an account on DigitalOcean. Free until we create a droplet.

Server provision and initial setup

Get the cheapest droplet in DigitalOcean. At the time of writing this is 4$/month. Choose the Ubuntu OS with the latest version (at the time of writing 23.04 x64, but the LTS version will work as well).

Make sure the Authentication method is 'SSH Key' and not 'Password'. You should setup the VPS in a way that you can ssh with your username only if ssh keys are correctly setup. Feel free to change the Hostname.

Once the droplet is created make sure the DNS is configured properly and points to your computer. You should be able to ssh root@example.com. This usually involves changing DNS settings in you domain registrar provider. Note that it might take some time for the DNS config to propagate. This is a good moment to make yourself some coffee or go for a stroll. If you are lucky it's just a few minutes.

Don't install anything on your VPS or modify config files.

Create the file server_config.ini filling each entry carefully.

Create an deployments_scripts directory in the remote machine and copy the contents of everything inside deployment_scripts/ into that folder:

jsmith@local:~/$ rsync -a deployment_scripts root@example.com:~/
jsmith@local:~/$ scp ~/secrets/server_config.ini root@example.com:~/deployment_scripts/

In the remote computer, as a superuser run:

root@remote:~/deployment_scripts# ./install.sh

After that reboot your system. Just (from any directory):

# reboot

That's it! You are now ready for deployment!

As a final configuration we should create a user in the machine that you will use to mange the machine. Add the user to the sudoers list. Choose a good password.

# adduser jsmith
# usermod -aG sudo jsmith

Copy the ssh keys from root to jsmith and setup the right permissions

# mkdir /home/jsmith/.ssh/
# cp ~/.ssh/authorized_keys /home/jsmith/.ssh/authorized_keys
# chown jsmith:jsmith /home/jsmith/.ssh/ -R
# chmod 700 /home/jsmith/.ssh/ -R

Make sure you can ssh as jsmith from your local machine and run sudo commands.

In the remote machine, as root, remove your authorized_keys:

# rm ~/.ssh/authorized_keys

Now root@example.com should not work but jsmith@example.com should.

Production logs

1. Issues with Systemd

   Please refer to our Systemd services crash course

2. Caddy and Gunicorn logs

If configured correctly Caddy access logs should be at /var/log/caddy/access.log and ``. A nice way to inspect those logs is jq:

# tail -f /var/log/caddy/access.log | jq .request.uri

Similarly logs created by Gunicorn will be at /var/log/django/access.log and /var/log/django/error.log.

3. Application logs

This are the logs generated by your application and they will be sent to Sentry. You can also find the at:

# tail -f /var/log/django/django.log

Troubleshooting

Please have a look at the blog post.

Deployment

If everything has been correctly set up, deployment will be a matter of running one script.

First push the code you want to deploy to the main branch on GitHub

As root in the remote computer just run

root@remote:~# deploy.sh

That will pull the latest code, create the virtualevn, install the dependencies, make migrations if needed and copy all the files needed.

Your app should be up and running at https://app.example.com

License

Licensed under either of

• Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
• MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)

at your option.