Cómo crear un servidor Django que ejecute uWSGI, NGINX y PostgreSQL en AWS EC2 con Python 3.6

Poner en marcha un servidor para un nuevo proyecto cada vez puede llevar mucho tiempo o resultar difícil para los nuevos desarrolladores. Así que pensé en escribir una guía paso a paso que facilitaría el proceso de implementación.

Si no está de humor para leer, puede copiar y pegar cada paso como se describe (reemplazar valores) y poner su servidor en funcionamiento.

Prerrequisitos:

  1. Instancia de Amazon Linux EC2 en funcionamiento con el par de claves asociado ( acceso ssh ).
  2. El puerto 22, 80 debe estar abierto para esta instancia.
  3. Aplicación de Django que desea implementar.
  4. La configuración de la base de datos está configurada para usar PostgreSQL.
  5. Requirements.txt está presente en su aplicación y tiene una lista de dependencias para instalar.
  6. Repositorio de Git para tu aplicación Django.

SSH y actualización de la instancia de ubuntu

Necesita ssh en su instancia EC2, así que asegúrese de tener el puerto 22 abierto para su instancia y luego realice una actualización.

ssh -i path-to-your-key.pem [email protected] sudo apt-get update && sudo apt-get upgrade -y

Instalación de Python3.6.x en AWS EC2 (ubuntu 16.04)

Descargaremos el archivo tar.xz del sitio oficial y luego lo instalaremos manualmente. Antes de eso, necesitamos instalar algunas dependencias requeridas.

Construyendo e instalando dependencias

sudo apt install build-essential checkinstall sudo apt install libreadline-gplv2-dev libncursesw5-dev libssl-dev libsqlite3-dev tk-dev libgdbm-dev libc6-dev libbz2-dev libffi-dev

Descarga e instalación manual de la versión de Python requerida

cd /opt && sudo wget //www.python.org/ftp/python/3.6.6/Python-3.6.6.tar.xz sudo tar -xvf Python-3.6.6.tar.xz cd Python-3.6.6/ sudo ./configure sudo make && sudo make install

Eliminando archivo descargado

sudo rm -rf Python-3.6.6.tar.xz

Verificar la versión de Python

python3 -V > Python 3.6.6

Configuración de usuario de Ubuntu para nuestra aplicación

Django en sí es un marco muy seguro, estoy de acuerdo. Pero las aplicaciones web siguen siendo vulnerables. Es una buena práctica ejecutar su aplicación como usuarios del sistema con privilegios limitados que tienen acceso limitado a los recursos de su servidor. Entonces, en esta sección, agregaremos un nuevo usuario y grupo de permisos a nuestra instancia EC2.

Añadiendo el grupo del sistema ubuntu 'nombre de grupo' [aplicaciones web en mi caso] y asignando un usuario 'nombre de usuario' [conejito en mi caso] a este grupo

sudo groupadd --system webapps sudo useradd --system --gid webapps --shell /bin/bash --home /webapps/project_name bunny

Nota: Supongo que " project_name " es el nombre que podría haber usado durante " django-admin startprojectyo> "

Crea un directorio para almacenar tu aplicación

Cree un directorio para almacenar su aplicación en / webapps / project_name /. Cambie el propietario de ese directorio por el usuario de su aplicación bunny:

sudo mkdir -p /webapps/project_name/ sudo chown bunny /webapps/project_name/

Permitir acceso limitado a otros usuarios del grupo al directorio de la aplicación

sudo chown -R bunny:users /webapps/project_name sudo chmod -R g+w /webapps/project_name

Ahora puede cambiar a su usuario

sudo su - bunny // your console will switch to something like this [email protected]:~$

Para volver al usuario sudo , simplemente hazlo ctrl+dy matará el terminal del usuario.

Instalación y configuración de PostgresSQL

Instalar PostgreSQL y crear una base de datos

sudo apt install postgresql postgresql-contrib sudo su - postgres [email protected]:~$ psql postgres=# CREATE DATABASE database_name;

Cambiar la contraseña predeterminada para postgres mientras está en la terminal psql

postgres=# \password

Implemente la aplicación Django en la instancia EC2 a través de Git en un entorno virtual

La implementación de su aplicación mediante un entorno virtual permite que su aplicación y sus requisitos se manejen por separado. Es una buena práctica mantener su aplicación aislada.

El uso del concepto de entorno es útil cuando está implementando más de una aplicación Django en una sola instancia para mantenerlas y sus dependencias aisladas entre sí.

Crearemos un entorno virtual en nuestro directorio de usuarios del sistema ( conejito ). Antes de eso, instalaremos git como usuario de sudo .

Instalar Git y extraer su código del repositorio de git

sudo apt-get install git sudo su - bunny // change to your repo https or ssh link [email protected]:~$ git remote add origin [email protected]:/.git [email protected]:~$ git pull origin 

Tenga en cuenta que no hemos clonado nuestro repositorio completo aquí. En su lugar, configuramos manualmente nuestro enlace git y solo sacamos la rama que queremos implementar en esta instancia. Puede tener una instancia diferente para su aplicación web lista para desarrollo, beta o producción correspondiente a cada rama en git.

Creando un entorno virtual usando Python3.6 en el directorio actual

[email protected]:~$ python3.6 -m venv . [email protected]:~$ source bin/activate (project_name)[email protected]:~$ pip install -r requirements.txt

En este punto, hemos configurado con éxito nuestro proyecto. Ahora necesitamos ejecutar algún comando manage.p y . Esto requerirá que estemos en el directorio donde está presente nuestro manage.py, o cada vez que necesitemos darle una ruta:

(project_name)[email protected]:~$ python manage.py migrate (project_name)[email protected]:~$ python manage.py createsuperuser (project_name)[email protected]:~$ python manage.py collectstatic

Note: collectstatic command requires that the STATIC configuration is setup properly. We are not discussing that here, though, as it is not in the scope of this tutorial.

(project_name)[email protected]:~$ python manage.py runserver 0.0.0.0:8000

This will start up the development server on port 8000. Assuming port 8000 is also open for your instance, you can visit your server's domain name or IP address followed by 8000 in your browser.

//your_server_domain_or_public_IP:8000
//your_server_domain_or_public_IP:8000/admin

Note: Don’t forget to add your domain or IP to ALLOWED_HOST in your settings.py

Setting up the uWSGI Application Server

Now that we’ve got our project set up and ready to go, we can configure uWSGI to serve our app to the web instead of the lightweight development server provided by Django.

If you’re thinking of running the runserver command on a screen, drop it. The dev server with Django is terribly lightweight, highly insecure, and not scalable.

You can install uWSGI either in virtualenv or globally and configure it accordingly.

In this tutorial, we’ll be installing uWSGI in virtualenv. Before we can install uWSGI, we need the Python development files that the software relies on.

Installing uWSGI along with its dependencies

sudo apt-get install python3-dev
sudo su - bunny
[email protected]:~$ source bin/activate
(project_name)[email protected]:~$ pip install uwsgi

Let’s run the server using uWSGI. This command does the same thing a manage.py runserver would do. You need to replace values accordingly to successfully test with this command.

(project_name)[email protected]:~$ uwsgi --http :8000 --home  --chdir  -w .wsgi

Creating uWSGI configuration file

Running uWSGI from the command line is only useful for testing. For actual deployment, we will create a .inifile somewhere in our system user directory. This file will contain all the configuration for handling a heavy request load, and can be tweaked accordingly.

Later in this tutorial, we will run uWSGI behind NGINX. NGINX is highly compatible with uWSGI and has built-in support for interacting with uWSGI.

Create a directory conf in your system user directory where you will store uwsgi.ini

(project_name)[email protected]:~$ mkdir conf
(project_name)[email protected]:~$ cd conf
(project_name)[email protected]:~$ nano uwsgi.ini

Copy the below code from the gist and save it I think the comments are explanatory enough for each option.

NOTE: updateMe is supposed to be you project name. It is the same name you gave above while creating the system user directory, so update accordingly.

[uwsgi] # telling user to execute file uid = bunny # telling group to execute file gid = webapps # name of project you during "django-admin startproject " project_name = updateMe # building base path to where project directory is present [In my case this dir is also where my virtual env is] base_dir = /webapps/%(project_name) # set PYTHONHOME/virtualenv or setting where my virtual enviroment is virtualenv = %(base_dir) # changig current directory to project directory where manage.py is present chdir = %(base_dir)/src/ # loading wsgi module module = %(project_name).wsgi:application # enabling master process with n numer of child process master = true processes = 4 # enabling multithreading and assigning threads per process # enable-threads = true # threads = 2 # Enable post buffering past N bytes. save to disk all HTTP bodies larger than the limit $ post-buffering = 204800 # Serialize accept() usage (if possibie). thunder-lock = True # Bind to the specified socket using default uwsgi protocol. uwsgi-socket = %(base_dir)/run/uwsgi.sock # set the UNIX sockets’ permissions to access chmod-socket = 666 # Set internal sockets timeout in seconds. socket-timeout = 300 # Set the maximum time (in seconds) a worker can take to reload/shutdown. reload-mercy = 8 # Reload a worker if its address space usage is higher than the specified value (in megabytes). reload-on-as = 512 # respawn processes taking more than 50 seconds harakiri = 50 # respawn processes after serving 5000 requests max-requests = 5000 # clear environment on exit vacuum = true # When enabled (set to True), only uWSGI internal messages and errors are logged. disable-logging = True # path to where uwsgi logs will be saved logto = %(base_dir)/log/uwsgi.log # maximum size of log file 20MB log-maxsize = 20971520 # Set logfile name after rotation. log-backupname = %(base_dir)/log/old-uwsgi.log # Reload uWSGI if the specified file or directory is modified/touched. touch-reload = %(base_dir)/src/ # Set the number of cores (CPUs) to allocate to each worker process. # cpu-affinity = 1 # Reload workers after this many seconds. Disabled by default. max-worker-lifetime = 300

I am trying to make everything easy with clear explanations. Cross check paths, directory name, and other inputs that you are required to replace.

We need to create the log file and run directory where our socket file will be created, that we just mentioned in our uwsgi.ini:

(project_name)[email protected]:~$ mkdir log
(project_name)[email protected]:~$ mkdir run
(project_name)[email protected]:~$ touch log/uwsgi.log

Make sure to change permissions for these two so that every group or user can write or execute files in these directories:

$ sudo chmod 777 /webapps/updateMe/run
$ sudo chmod 777 /webapps/updateMe/log

Now let’s try running the server using uwsgi.ini that we just created.

(project_name)[email protected]:~$ uwsgi --ini /webapps/updateMe/conf/uwsgi.ini

If everything up until now is setup correctly, then it should be running. If not, then you need to go back to check for anything you missed (like the path/project name, etc).

To check any uswgi log you can cat or tail uwsgi.log:

(project_name)[email protected]:~$ tail log/uwsgi.log

Create a systemd Unit File for uWSGI

At this point if everything is cool, you can even run this command in screen and detach it — but again, this is not a good practice at all. Instead we will create a system service and let systemd (Ubuntu’s service manager) take care of it.

Switch back to sudo user

$ sudo nano /etc/systemd/system/uwsgi.service

and copy paste code from the below gist. Don’t forget to update and crosscheck names/path that suit your app:

[Unit] Description=uWSGI instance to serve updateMe project After=network.target [Service] User=bunny Group=webapps WorkingDirectory=/webapps/project_name/src Environment="PATH=/webapps/project_name/bin" ExecStart=/webapps/project_name/bin/uwsgi --ini /webapps/project_name/conf/uwsgi.ini Restart=always KillSignal=SIGQUIT Type=notify NotifyAccess=all [Install] WantedBy=multi-user.target

After you save the above file and close it, you can run following commands:

Reload systemctl daemon to reload systemd manager configuration and recreate the entire dependency tree

$ sudo systemctl daemon-reload

Enable uwsgi service to start on system reboot

$ sudo systemctl enable uwsgi

Start uwsgi service

$ sudo service uwsgi start

Restart uwsgi service

$ sudo service uwsgi restart

Check uwsgi service status

$ sudo service uwsgi status

Take a deep breath here if everything ran smoothly. We just finished setting up most hectic part of this tutorial, so you should be proud.

Next we will setup NGINX, and then we’ll be done! I know this is taking a bit of time, but believe me — once done, you will be as happy as I will be after publishing this tutorial.

Setting Up NGINX on EC2 for uWSGI

NGINX is a lightweight server, and we’ll use it as a reverse proxy.

We could let uWSGI run directly on port 80, but NGINX has a lot more benefits which makes it desirable. Also NGINX natively includes support for uWSGI.

Enough talk, let’s install NGINX on our instance

$ sudo apt-get install nginx

Now when you go to //your-public-ip-or-address, you will see a Nginx welcome page. This is because NGINX is listening to port 80 according to its default configuration.

NGINX has two directories, sites-available and sites-enabled,that need our attention. sites-available stores all conf files for all available sites on that particular instance. sites-enabledstores the symbolic link for each enabled site to the sites-available directory.

By default, there is only one conf file named default that has basic setup for NGINX. You can either modify it or create a new one. In our case, I am going to delete it:

$ sudo rm -rf /etc/nginx/sites-available/default
$ sudo rm -rf /etc/nginx/sites-enabled/default

Let’s create our nginx-uwsgi.conf file to connect the browser request to the uwsgi server we are running in site-available:

$ sudo nano /etc/nginx/sites-available/nginx-uwsgi.conf

and copy the following code from the gist below:

upstream updateMe_dev { server unix:/webapps/updateMe/run/uwsgi.sock; } server { listen 80; server_name your-IP-or-address-here; charset utf-8; client_max_body_size 128M; location /static { # exact path to where your static files are located on server # [mostly you won't need this, as you will be using some storage service for same] alias /webapps/updateMe/static_local; } location /media { # exact path to where your media files are located on server # [mostly you won't need this, as you will be using some storage service for same] alias /webapps/updateMe/media_local; } location / { include uwsgi_params; uwsgi_pass updateMe_dev; uwsgi_read_timeout 300s; uwsgi_send_timeout 300s; } access_log /webapps/updateMe/log/dev-nginx-access.log; error_log /webapps/updateMe/log/dev-nginx-error.log; }

Create symbolic link into sites-enabled directory for same

$ sudo ln -s /etc/nginx/sites-available/nginx-uwsgi.conf /etc/nginx/sites-enabled/nginx-uwsgi.conf

That’s all, we’re almost there, about to finish up…

Reload systemctl daemon

$ sudo systemctl daemon-reload

Enable nginx service on system reboot

$ sudo systemctl enable nginx

Start Nginx service

$ sudo service nginx start

Test Nginx. It should return OK, Successful as a part of the result.

$ sudo nginx -t

If NGINX fails, you can check its last error-log or access-log on the path specified by us in its conf.

$ tail -f /webapps/updateMe/log/nginx-error.log
$ tail -f /webapps/updateMe/log/nginx-access.log

Restart Nginx Service

$ sudo service nginx restart

Check Nginx Service status

$ sudo service nginx status

You should now be able to reach your app at //your-public-ip-or-address

Well this is the end of this lengthy tutorial. I hope you got what you expected from it. Thanks for bearing with me.

PS: uWSGI + NGINX + Django is highly customizable to meet any large scale requirements. That being said, core optimization still lies at application level. How you code and make use of Django ORM or Raw SQL query, etc. will help you further.

Original text