Challenges with Deploying Streamlit Apps on AWS Elastic Beanstalk

[segidas]

Hello

I am facing some issues while trying to deploy my Streamlit app on AWS Elastic Beanstalk. The app works perfectly in my local environment but once deployed; I face challenges such as connection errors, environment variable misconfigurations & difficulty in setting up the requirements.txt file correctly for the environment.

I have tried tweaking configurations based on several guides but the app either crashes on start / shows a generic error page.

One specific issue seems to revolve around the correct version of Python & compatibility with the streamlit library. Even after specifying the Python version and updating my packages; I still face compatibility warnings. Additionally; I am unsure how to correctly set up the Elastic Beanstalk environment to handle Streamlit’s web server properly; especially with custom ports.

If anyone has successfully deployed a Streamlit app on AWS Elastic Beanstalk, I would love to know your approach, including how you handled environment variables, server port configurations & ensuring dependencies are resolved without errors.
I have checked HOWTO: Deploying Streamlit app to AWS Elastic Beanstalk (no containers) Java guide for reference .

Any examples, tutorials and step-by-step instructions would be immensely helpful!

Thank you !

[Augusto2]

To deploy your Streamlit app on AWS Elastic Beanstalk (EB), you need to consider how Elastic Beanstalk interacts with Python applications and how Streamlit, as a non-WSGI framework, fits into this environment.

Understanding Elastic Beanstalk and Streamlit

AWS Elastic Beanstalk is a platform that simplifies deploying and managing applications. For Python environments, EB typically expects a WSGI-compatible application (e.g., Flask or Django) , which it runs using a WSGI server such as Gunicorn.

However, Streamlit operates differently. It’s not a WSGI application; instead, it runs its own web server, typically started with the command streamlit run app.py. This mismatch means that EB’s default behavior—looking for a WSGI entry point—won’t work directly with Streamlit. Fortunately, EB allows you to override this default behavior using a Procfile, which lets you specify a custom command to run your application.

Deploying Streamlit on Elastic Beanstalk

Since Streamlit isn’t WSGI-compatible, the simplest and most recommended approach is to use a Procfile to tell Elastic Beanstalk how to run your Streamlit app. Here’s how you can do it:

1. Keep Your Streamlit App in app.py

   • The Streamlit code you provided should be saved as app.py. This file contains the core logic of your dashboard and is the main application you want to deploy.

2. Create a Procfile

   • Add a file named Procfile (no extension) in the root directory of your project with the following content:

     web: streamlit run app.py --server.port=$PORT --server.headless=true
     

   • Explanation:
     • web: tells EB that this is the command for the web process.
     • streamlit run app.py is the standard command to start a Streamlit app.
     • --server.port=$PORT ensures Streamlit listens on the port assigned by EB, which is stored in the PORT environment variable (EB dynamically assigns this, often 5000, but should be 8501).
     • --server.headless=true is necessary because EB runs in a headless environment (no GUI), and Streamlit needs this flag to function properly.

3. Install Dependencies

   • Create a requirements.txt file listing your dependencies:

     streamlit
     numpy
     matplotlib
     

   • EB will install these when deploying your application.

4. Directory Structure
   Your project should look like this:

   your-app/
   ├── app.py          # Your Streamlit app
   ├── Procfile        # Custom command for EB
   ├── requirements.txt # Dependencies
   

5. Create a options.config

   • Create a .ebextensions folder inside your project with the file options.config:

   option_settings:
    aws:elasticbeanstalk:container:python:
      WSGIPath: app:main
    aws:elasticbeanstalk:application:environment:
      PORT: 8501
   

6. Deploy to Elastic Beanstalk

   • Zip your project directory and upload it to EB via the AWS console or CLI (eb deploy).
   • EB will use the Procfile to start your Streamlit app, and the app will be accessible via the EB-provided URL.

Final Recommendation

For your use case—deploying a Streamlit dashboard—stick with the Procfile approach:

• Use app.py as your Streamlit app.
• Add a Procfile with web: streamlit run app.py --server.port=$PORT --server.headless=true.
• Include requirements.txt for dependencies.
• Create options.config file for environment variables.

This setup ensures your Streamlit app runs correctly and is accessible via the EB URL.