Streamlit is a trending Python library for creating web applications in the field of data science and machine learning. It provides many useful input widgets and display elements wrapped in a clean Python API, allowing users to define feature-rich web pages purely in Python, without having to deal with HTML, CSS, and JS.
For programmers like me who are not very familiar with those languages, Streamlit greatly simplifies the front end development difficulties. Recently, I have been rewriting some previous full-stack Flask projects with Streamlit. Through the journey, I also explored the deployment of Streamlit application with docker and Nginx, which I would like to share with you in this post.
Dockerize Streamlit App
To deploy a Streamlit application with docker, first we need to build a docker image of the application. A Dockerfile like below serves the purpose:
FROM python:3.12-slim
WORKDIR /app
RUN apt-get update && apt-get install -y \
build-essential \
curl \
software-properties-common \
git \
&& rm -rf /var/lib/apt/lists/*
COPY app/ .
RUN pip3 install -r requirements.txt
EXPOSE 8501
HEALTHCHECK CMD curl --fail http://localhost:8501/_stcore/health
ENTRYPOINT ["streamlit", "run", "Home.py", "--server.port=8501", "--server.address=0.0.0.0"]
This Dockerfile is simply adapted from the one from tutorial, simply replacing the git clone with coping from a local repo app/. Then docker build command is used to create the image, and docker run command is used to launch the container.
Deploy together with Nginx
For any production deployment, I always tend to put Nginx in front of the application. One important reason is Nginx seems more efficient to serve static files. This is particularly important for Streamlit, because the current version of Streamlit only supports static serving of small media files – any other generic files will be sent with header Content-Type:text/plain, which will cause browsers to render in plain text.
Below is an example of Nginx configuration file, combing static file serving with traffic proxy passing for the Streamlit app.
server {
listen 80;
server_name _;
location ~ \.(pdf|djvu)$ {
expires max;
root /srv/www;
try_files $uri =404;
}
location / {
proxy_pass http://app:8501/;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header Host $http_host;
proxy_redirect off;
proxy_http_version 1.1;
proxy_set_header Upgrade $http_upgrade;
proxy_set_header Connection "upgrade";
}
}
My preference for deployment is to use separate containers for Streamlit app and Nginx. The most convenient way is to use Docker Compose. It’s a tool for defining and running multi-container applications. Here is a minimal configuration file compose.yaml:
services:
app:
build: .
nginx:
image: "nginx:latest"
volumes:
- ./static:/srv/www/static
- ./default.conf:/etc/nginx/conf.d/default.conf
ports:
- "8080:80"
By default, containers can now look up the service name (e.g., app and nginx in the config above ) and get the appropriate container’s IP address. That’s why we can use proxy_pass http://app:8501/ in Nginx config file.
The command to launch the application is simply docker compose up -d --build. Here --build forces rebuild of the image during development stage.
Concluding Remarks
My first impression on streamlit was just a prototype web app builder and did not intend to extensively use it for my projects. But later I realized in many cases, performance is simply not the major concern, especially when developing tools targeted to a small audience base. Streamlit saves me a lot of development/maintenance efforts, and let me focus more on the core part of the projects rather than designing/styling UIs. Hope this blog post is helpful when you start deploying your own Streamlit app.
References
License
This work is licensed
under a Creative Commons Attribution 4.0
International License.