Starting Up and Cloning Repository on Infinite Loop

Community Cloud
,
1

Hello everyone.

I would like to respectfully seek some expert advice in debugging a problem I’ve encountered when deploying my Streamlit app. I noticed that it can’t get past the “Starting Up Repository” and “Cloning Repository” steps. It just goes on an infinite loop.

I’ve already tried the following debugging actions to no avail:

  • Rebooted the app deployment process
  • Refreshed the app deployment process
  • Deleted the existing app and initiated a new app deployment (same app name)
  • Deleted the existing app and initiated a new deployment (different app name)
  • Revised my GitHub .gitignore file to exclude tracking folders with huge file sizes
  • Updated my source GitHub repository to shorten folder paths for files that my app uses
  • Initiated the app deployment with or without VPN
  • Allowed the deployment process to run overnight

In case it helps, these are some additional info:

  • The app has been tested successfully when implemented locally
  • This is my 3rd deployed app. I did not have issues with my previous two (Scikit-Learn-based models). But this is the first time I’m deploying a CNN model (Keras-based model).
  • Total file size must be around 115 MB.
  • While in the process of finalizing the app, I’ve intermittently performed app deployment and there was one instance where it was deployed successfully. But I’ve pushed changes to my GitHub repository since then that prompted the app to refresh.
  • I can’t typically connect to the web without a VPN due to access restrictions in the country where I’m currently residing.

Any help would be appreciated. Thank you!

Attached below are details about my application:

App Description:
CNN model for image classification using Keras and post-prediction visualization using gradient class activation mapping (Grad-CAM) for results interpretation.

Package Versions:

  • Python = 3.12
  • Streamlit = 1.38.0
  • Keras = 3.6.0
  • Tensorflow = 2.18.0

Link to Deployed App:
Public App (Deployed on Community Cloud)

Link to Source GitHub Repository:
App’s Public GitHub Repository

Streamlit Log:

StreamlitLog
StreamlitLog1645×592 95.8 KB

1 Like
2

I forked your repo and had the same issue when deploying, then I made changes until I had a minimal streamlit application and the issue persisted:

Of course if I create the same application from scratch I can deploy without issues. So it seems to be something about your repo, that replicates across forks but is unrelated to the actual source tree. Misterious.

As a workaround, try recreating your app in a new repository.

2 Likes
3

Thank you very much for your analysis and recommendation. :pray:
I will definitely try this out.
I have also tried recreating a smaller-sized model, but the issue persisted.
This is very much appreciated.

4

Hi there @Goyo!

This is to confirm that I was able to successfully deploy my Streamlit app after -

  • Creating a new GitHub repository from scratch
  • Using a different Streamlit app deployment link name
  • Only copying all Streamlit-related apps on my new GitHub repository
  • Executing model deployment after a fresh PC bootup

As my original GitHub repository contains Python notebooks; images; markdown / HTML files; and raw model files (.keras), I’ll try to individually add them to my new repository to try tracking what might have caused the earlier issue. But for now, I’m glad my Streamlit app is already working fine!

Thank you very much for helping me resolve the issue. I truly appreciate it. :pray:

1 Like
5

This topic was automatically closed 2 days after the last reply. New replies are no longer allowed.