mirror of
https://github.com/open-webui/open-webui.git
synced 2026-05-08 04:16:03 -05:00
[GH-ISSUE #14380] bug: redirects to localhost:3000 on 0.6.11 azure ad sso #32758
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 @fanuelsen on GitHub (May 27, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/14380
Check Existing Issues
Problem Description
After updating the docker container to the latest version login with azure got broken for me, i get this error in the logs on the client: "localhost:3000/auth#token=eyJhbGciOiJIUzl1NilsInR5cCI6IkpXVCJ9.eyJpZCI6ImFmZD
RhMzg5LTg0YjAtNDgxNi1iY2MyLWM4Y2RIZGNIOTEyNiJ9.7qaZ7WXrfd2LG8GbMcZLrxy
ZRUxZ1PwQ7tqwaynOtFQ" because Safari can't connect to the server "localhost".
the open-webui
2025-05-27 06:53:27.959 | INFO | uvicorn.protocols.http.httptools_impl:send:476 - - "GET /oauth/microsoft/callback?code=redacted &session_state=redacted HTTP/1.1" 307 - {}
Desired Solution you'd like
to logon as i used to do in version 0.6.10
Alternatives Considered
No response
Additional Context
This happend after pulling the new image. 0.6.11
@tsueri commented on GitHub (May 27, 2025):
I use it with zitadel and got the same problem.
@Breidenbach-Murach commented on GitHub (May 27, 2025):
We have the same issue after the update to 0.6.11.
A simple rollback of the container using the "0.6.10" tag did not fix it, as the login prompt seemingly gets stuck after the redirect.
@tsueri commented on GitHub (May 27, 2025):
I "fixed" it by manually rolling back the Commit
d4c3cfb@tjbck commented on GitHub (May 27, 2025):
PLEASE READ THE CHANGELOGS BEFORE UPDATING
@tsueri commented on GitHub (May 27, 2025):
I have read the changelog. My problem persists even with the environment variable.
This is my compose.yml:
@fanuelsen commented on GitHub (May 27, 2025):
I see you expose a host port, do you add the port in the webui_url? It worked for me when adding the env var, but i'm running it behind a reverse proxy.
here are the docs:
https://docs.openwebui.com/tutorials/integrations/browser-search-engine#step-1-set-the-webui_url-environment-variable
@Classic298 commented on GitHub (May 27, 2025):
@tsueri do not set the environment variable. It is a persistent config.
Instead set the WEB UI URL in the admin panel and try again.
@tsueri commented on GitHub (May 27, 2025):
That worked! Thank you!
@ArcCal commented on GitHub (May 28, 2025):
Encountered the exact same problem
@cwittenberg commented on GitHub (Jun 2, 2025):
This worked. Would suggest documentation or perhaps an additional envvar on this - that overrides the UI?
@aleprj commented on GitHub (Jun 2, 2025):
I think it should be an additional var for this redirect. It's a change intented for a very specific corner case and probably broke things for more people than the ones that will actually benefit from it.
Anyway, if we are going to keep things this way, some documentation needs updating:
Someone starting a clean install will be lost right now.
@Classic298 commented on GitHub (Jun 2, 2025):
Feel free to update the docs
@avmturo commented on GitHub (Jun 5, 2025):
Does this not mean, that if you are deploying a new instance, with a new database, the first user, if using oidc, will not be able to log in? If there is no user yet to access the admin panel, how can someone update the WEBUI_URL?
@Classic298 commented on GitHub (Jun 5, 2025):
@avmturo use DISABLE_PERSISTENT_CONFIG and use that to then set the WEBUI_URL (check docs for exact env var names, this is from the top of my head)
but to answer your question: there's a way for everything
@avmturo commented on GitHub (Jun 5, 2025):
Ahh okay, I was confused by this one. Checked the doc and it iis an environment variable. So if you are deploying a new isntance, even with persistant config, it should work with using the env var. If I'm not mistaken
@aleprj commented on GitHub (Jun 5, 2025):
Unless you already started it before and the database was generated. This env var has a default value, so if the system was already started I would just delete the database, since it's a new setup anyway. I usually start my aplications before configuring SSO, so it would probably happen.
@aleprj commented on GitHub (Jun 5, 2025):
The var is ENABLE_PERSISTENT_CONFIG (need to set to false). But how does it work? If I set it to false it will use the env var only for this specific run or will it save the new value as well (ie. it disables only loading of the variables, or does not touch them at all)?