RHEL

Deploy your application with uWSGI and Nginx on Red Hat Enterprise Linux

Red Hat is an Open Source leader providing an GNU/Linux operating system named RHEL that can run the Web Application Server uWSGI and the Web Server Nginx.

Upgrading Python

Most RHEL are delivered with a Python version older than 3.8 which is the oldest Python version supported by Taipy. If you are in that case, please install Python 3.8 (or newer):

sudo dnf install -y gcc openssl-devel bzip2-devel libffi-devel make
wget https://www.python.org/ftp/python/3.8.12/Python-3.8.12.tgz
tar xzf Python-3.8.12.tgz
rm -rf Python-3.8.12.tgz
cd Python-3.8.12
sudo ./configure --enable-optimizations
sudo make altinstall
cd ..
sudo rm -r Python-3.8.12

Info

This tutorial specifies the Python version for each command. If your default version is different, you must replace python3.8 with python.

Prepare your machine

The following software should be installed on your target machine:

• pip: for installing Python3 packages.

• uwsgi and gevent: the Web application server and its workers that will run the Taipy application.

• nginx: the Web server for the Internet exposition.

You can install all of these packages by running the following command:

sudo dnf install -y nginx
python3.8 -m pip install uwsgi gevent
sudo mv `pwd`/.local/bin/uwsgi /usr/bin/uwsgi
sudo restorecon /usr/bin/uwsgi

Note

If you are using a SQL database based on Microsoft SQL Server, you need to install your corresponding Microsoft ODBC Driver for SQL Server.

Run the application locally

If you want to deploy the following application:

from taipy import Gui

Gui(page="# Getting started with *Taipy*").run()

This would be placed in a file called app.py.
You need to create a requirements.txt file that contains:

taipy

On your local machine, start the application by doing:

$ python3.8 -m pip install -r requirements.txt
Collecting taipy
...
Successfully installed taipy
$ python3.8 app.py
 * Server starting on http://127.0.0.1:5000
 * Serving Flask app 'Taipy' (lazy loading)
 * Environment: production
   WARNING: This is a development server. Do not use it in a production deployment.
   Use a production WSGI server instead.
 * Debug mode: on
...

The application is running locally, you can access it with the browser on the URL http://127.0.0.1:5000/.

Note

The message:

WARNING: This is a development server. Do not use it in a production deployment.
Use a production WSGI server instead.

Is provided by Flask because the way of exposing an application on the Internet and developping an application locally is not the same, mainly for security and reliability reasons. This message will disappear when using the Web server.

Prepare the application for deployment

Deploying your application on a remote environment needs a little bit of configuration.

By default, Taipy applications run in Debug mode. Before deploying your application to the Internet, you should turn off the Debug mode by setting the debug parameter or the run() to False.
You must also inform Taipy not to run the application server on its own but rather delegate the execution by setting the parameter run_server to False.
The name of the variable where the Web application is stored is used in the uWSGI configuration: this allows the Web server to load the Web application:

from taipy import Gui

gui_service = Gui(page="# Getting started with *Taipy*")
web_app = gui_service.run(debug=False, run_server=False)

In our example, we store this application in the variable web_app (see line 3)

Make sure you upload this code on your targeted machine and install your dependencies with pip.

Important

The entry point filename and the app variable name are important for the proper configuration of the uWSGI Web application server. Please, keep them as is or adapt the configuration.

uWSGI application server

To expose your application over the Internet, you must use uWSGI instead of Flask as the application server. You would then leverage Nginx to expose the application.

uWSGI can be started manually. But, generally, it's better to start the application automatically when the machine starts. To order to do that, you should use Systemd which is installed by default on RHEL.

From the directory where app.py is located, run the following command to generate an adapted file for Systemd:

echo """
[Unit]
Description=App
After=syslog.target

[Service]
ExecStart=uwsgi --http 127.0.0.1:5000 --gevent 1000 --http-websockets --module app:web_app
WorkingDirectory=`pwd`
Restart=always
KillSignal=SIGQUIT
Type=notify
StandardError=syslog
NotifyAccess=all
User=`whoami`

[Install]
WantedBy=multi-user.target
""" > app.uwsgi.service

Then transfer this file to the correct folder by doing:

sudo mv app.uwsgi.service /etc/systemd/system/app.uwsgi.service

Now, you can start your application automatically on startup time of your machine by doing:

sudo restorecon /etc/systemd/system/app.uwsgi.service
sudo systemctl daemon-reload
sudo systemctl start app.uwsgi.service
sudo systemctl enable app.uwsgi.service

The application is now running locally but is not accessible yet from the Internet.

Exposing to the Internet

To expose your application on the Internet, you should use Nginx. Replace the content of /etc/nginx/nginx.conf by the following or:

sudo wget https://docs.taipy.io/en/latest/manuals/deployment/linux/nginx.conf -O /etc/nginx/nginx.conf

Allow the communication between Nginx and uWSGI:

sudo setsebool -P httpd_can_network_connect 1

Then restart Nginx:

sudo systemctl restart nginx

Note

This configuration is only for HTTP. If you need an HTTPS connection, please read the Nginx documentation.

Open the firewall

Your application is ready to receive traffic from the Internet, but your firewall still blocks the communication. Open the http port that is (i.e. port 80):

sudo firewall-cmd --zone=public --add-port=80/tcp --permanent
sudo firewall-cmd --reload

Your application is now accessible over the Internet!