mirror of
https://github.com/open-webui/open-webui.git
synced 2026-03-11 00:04:08 -05:00
issue: CloudFlare / Reverse proxy tunnel - Failed to fetch dynamically imported module #4367
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @matbeedotcom on GitHub (Mar 10, 2025).
Check Existing Issues
Installation Method
Docker
Open WebUI Version
Latest docker
Ollama Version (if applicable)
No response
Operating System
Ubuntu 22.04
Browser (if applicable)
No response
Confirmation
README.md.Expected Behavior
Loads
Actual Behavior
Doesnt load
Steps to Reproduce
Run a reverse proxy
Forward request
Logs & Screenshots
Additional Information
No response
@francesco-carrella commented on GitHub (Mar 12, 2025):
Hello! I have this exact problem. Is there some other open issue about this or some info about the solution?
Thank you!
@xellos866 commented on GitHub (May 20, 2025):
We have experienced a similar error and maybe our findings will also help someone:
After debugging, we have probably found a way to solve the error. It is likely a problem with the number of simultaneous connections from the reverse proxy (nginx) that are sometimes triggered when the login window is opened by the client.
In some cases, up to 150 parallel HTTP connections were being established, which exceeded the limits for parallel connections in our nginx reverse proxy (default was 100). As a result, some of the connections were cut off by nginx. This in turn caused the application not to initialize correctly and triggered the error message.
When reloading, only some of the connections are re-established, and it also seems that there is some short-term caching in Open WebUI itself, so that significantly fewer JS files are sometimes requested in parallel. Furthermore, the error only appeared on fast internet connections, meaning that when we throttled our own connection speed, the error reliably did not occur anymore.
We have now significantly increased the number of allowed parallel connections, and in our tests, we could no longer reproduce the error.