Repl deploy fails/dies at: Pushing Repl (cache) layer

Problem description:
The repl (private):
consistently fails to deploy a tiny streamlit app.

This is a second repl that fails the moment I add a few python packages as dependencies.

Expected behavior:
Successful deploy or at the very least a helpful diagnostic.

Actual behavior:

info: Build: 32469a68-1d85-4fc5-b1f8-fb7853db622a - 2023-04-19T19:07:03.701Zinfo: Installing packages - 2023-04-19T19:07:11.263Zinfo: Skipping Build as no Build command set - 2023-04-19T19:07:11.792Zinfo: Pushing pid1 binary layer... - 2023-04-19T19:07:12Zinfo: Created pid1 binary layer - 2023-04-19T19:07:15Zinfo: Pushing Repl layer... - 2023-04-19T19:07:20Zinfo: Pushing hosting layer... - 2023-04-19T19:07:20Zinfo: Nix layers for this Repl are uncached. - 2023-04-19T19:07:20Zinfo: Pushing nix-0 layer... - 2023-04-19T19:07:20Zinfo: Created hosting layer - 2023-04-19T19:07:21Zinfo: Created Repl layer - 2023-04-19T19:07:43Zinfo: Pushing Repl (cache) layer... - 2023-04-19T19:07:43Z
fatal: failed to push layers

Steps to reproduce:

  1. Click deploy on the repl.
  2. Watch it fail

Bug appears at this link:


Hey @gkossakowski! I noticed you were able to deploy, but it still says “please wait”.

Try this as your run command:

echo ‘[](’ | streamlit run --server.address --server.enableCORS=false --server.enableWebsocketCompression=false

streamlit run --server.address --server.headless true --server.enableCORS=false --server.enableWebsocketCompression=false

should work too (notice the headless mode config)

Thanks for the reply soren. Yes, indeed I managed to deploy in the end, but deploys are extremely flakey, see:

Only one of these failures was my mistake. All other failures are infra failure before the container wakes up. Am I doing something particularly unusual or this is just dust settling from a feature released a week ago (still not a good look)?

Also, the streamlit problem is due to websockets. Are websockets not supported in Replit deployments?

Websockets are supported! We’re investigating why the redeployments might be flaky now.


Running into this same issue on my app as well.

Sorry about this! Can you share a link to your Repl and any relevant info about the failure?