FAQ: App works locally, but not when deployed

FAQs
1

Problem

Perhaps your app is working fine locally but when you’ve deployed the app it is no longer working. This is a common scenario encountered by users; let’s take a look at how we can resolve this.

Solution

Scenario 1: Discrepancy of library version of those used locally vs. cloud

In most cases, this issue is most likely caused by a discrepancy in the library version of those installed locally vs. installed on the cloud.

This can be solved by making sure that the library version that is used on the cloud is the same as that installed locally.

For example, if you’re using numpy version 1.26.0 locally and if you’re specifying only numpy in your requirements.txt file then this would most likely lead to the server automatically using the latest version instead of the older version used by your app when it was created.

Typically, library dependencies may change over time and therefore it is always a great idea to pin a specific version in your requirements.txt file. Thus, you can instead specify numpy==1.26.0 in your requirements.txt file.

Scenario 2: Missing library dependencies

Another plausible explanation is that your deployed app may be missing dependent libraries. Perhaps the library is installed locally but is missing in the cloud deployment.

To figure out which libraries are missing, you can look at the imported libraries in your streamlit_app.py file and make sure to include them in your requirements.txt file.

For example if your app has import statements such as:

import streamlit as st
import pandas as pd
import numpy as np

Please ensure that you have streamlit, pandas and numpy in your requirements.txt file like so:

streamlit==1.38.0
pandas==2.2.2
numpy==2.1.0
1 Like
2

Another resource to add here would be to understand Streamlit’s Client-Server architecture.

When running locally, your machine is both the client and the server. When deployed, the server is where the Python script runs, and the client is the user’s browser.
So, the hosted server might not have access to components in your local server (for example, audio drivers). It might also be running on a different OS, which can impact the app if you use OS specific functions.