AllBugsFeaturesChangelogSearchNew post
Sign in with Fluxer
Bugs

[Self-Hosted]Admin Interface Is Not Accessible When Running on a Non-Standard Port (non-443)

(#618) Bug Fixed self-hosting
by RexSystem edited originally by @Sinterdial on GitHub

Summary

When I change the externally exposed port (FLUXER_PUBLIC_PORT) from the default one in the environment variables (for example, to 5366), I can no longer access the admin interface correctly. Visiting: https://domain.com:5366/admin/ automatically redirects me to: https://domain.com/oauth2/authorize?... During this redirect, the port information (:5366) is lost. If I manually add the port back after the redirect, for example: https://domain.com:5366/oauth2/authorize?... the page then reports that the redirect URL is not registered (see Screenshot 2).

Steps to reproduce

  1. Change the exposed port (FLUXER_PUBLIC_PORT) from the default value to a non-standard port (e.g. 5366) via environment variables.
  2. Access https://domain.com:5366/admin/.
  3. Observe that the request is redirected to https://domain.com/oauth2/authorize?... without the port.
  4. Manually add the port back to the redirected URL.
  5. Observe the "redirect URL not registered" error.

Environment

Version: BUILD_VERSION 2026.616.45607 OS: Docker 27.5.1 on unRAID 7.2.4, with Pangolin 1.19.2 as reverse proxy before Fluxer's Caddy Device: iPad, iPhone and Windows PC Browser: FireFox 151.3.2

Logs or screenshots

image image
  • 609051840-8c6bd494-0f54-44a5-ae9b-a8f501d68321.jpeg

    609051840-8c6bd494-0f54-44a5-ae9b-a8f501d68321.jpeg

    1206×2534 | 105 kB

  • 609051841-5ef68d39-9b33-4200-a1d4-5c82fe10884b.jpeg

    609051841-5ef68d39-9b33-4200-a1d4-5c82fe10884b.jpeg

    1206×2622 | 147 kB

Merged posts

These posts were merged into this one. Their comments are now part of the conversation below, marked with where they came from.

Merged from #685 Admin page is inaccessible on non-default port with docker compose instructions

RexSystemoriginally by @naelstrof on GitHub

Report details

Summary

Setting FLUXER_PUBLIC_PORT in .env to something other than 80 prevents being able to access the admin page. For example, if FLUXER_PUBLIC_PORT is set to 8080, Accessing https://chat.example.com:8080/admin, attempts to load https://chat.example.com/oauth2/ which fails to load. Stuck showing "Starting Authentication..." I've found a fix, I've changed the following in the docker-compose.yml, changing 
admin:
    <<: *fluxer-service
    image: ${FLUXER_REGISTRY:-ghcr.io/${FLUXER_REGISTRY_OWNER:-fluxerapp}}/fluxer-admin:${FLUXER_IMAGE_TAG:-v1}
    environment:
      <<: *fluxer-env
      FLUXER_ADMIN_HOST: 0.0.0.0
      FLUXER_ADMIN_PORT: "8080"
      FLUXER_ADMIN_BASE_PATH: /admin
      FLUXER_API_ENDPOINT: http://api:8080
      FLUXER_ADMIN_ENDPOINT: ${FLUXER_PUBLIC_SCHEME:-https}://${FLUXER_DOMAIN}/admin
      FLUXER_APP_ENDPOINT: ${FLUXER_PUBLIC_SCHEME:-https}://${FLUXER_DOMAIN}
      FLUXER_MEDIA_ENDPOINT: ${FLUXER_PUBLIC_SCHEME:-https}://${FLUXER_DOMAIN}/media
      FLUXER_STATIC_CDN_ENDPOINT: ${FLUXER_PUBLIC_SCHEME:-https}://${FLUXER_DOMAIN}
      FLUXER_ADMIN_OAUTH_REDIRECT_URI: ${FLUXER_PUBLIC_SCHEME:-https}://${FLUXER_DOMAIN}/admin/oauth2_callback
    depends_on:
      api: {condition: service_healthy}
to 
admin:
    <<: *fluxer-service
    image: ${FLUXER_REGISTRY:-ghcr.io/${FLUXER_REGISTRY_OWNER:-fluxerapp}}/fluxer-admin:${FLUXER_IMAGE_TAG:-v1}
    environment:
      <<: *fluxer-env
      FLUXER_ADMIN_HOST: 0.0.0.0
      FLUXER_ADMIN_PORT: "8080"
      FLUXER_ADMIN_BASE_PATH: /admin
      FLUXER_API_ENDPOINT: http://api:8080
      FLUXER_ADMIN_ENDPOINT: ${FLUXER_PUBLIC_SCHEME:-https}://${FLUXER_DOMAIN}/admin
      FLUXER_APP_ENDPOINT: ${FLUXER_PUBLIC_SCHEME:-https}://${FLUXER_DOMAIN}:${FLUXER_PUBLIC_PORT}
      FLUXER_MEDIA_ENDPOINT: ${FLUXER_PUBLIC_SCHEME:-https}://${FLUXER_DOMAIN}/media
      FLUXER_STATIC_CDN_ENDPOINT: ${FLUXER_PUBLIC_SCHEME:-https}://${FLUXER_DOMAIN}
      FLUXER_ADMIN_OAUTH_REDIRECT_URI: ${FLUXER_PUBLIC_SCHEME:-https}://${FLUXER_DOMAIN}:${FLUXER_PUBLIC_PORT}/admin/oauth2_callback
    depends_on:
      api: {condition: service_healthy}
By adding ${FLUXER_PUBLIC_PORT} to botht he app endpoint and oauth redirect, the admin page appears to work normally.

Steps to reproduce

Follow instructions at https://docs.fluxer.app/operator/get-started/, but when prompted to customize .env, Set FLUXER_PUBLIC_PORT to something non-default like 8080. After setup, attempt to access the admin page at https://chat.example.com:8080/admin, and see it fails.

Environment

Fluxer on Arch Container running Docker.

3 comments

Sign in with Fluxer to comment and vote.
Comment by @Sinterdial
RexSystem 1 vote originally by @Sinterdial on GitHub OP
solved by adding port in admin container's env
Permalink
Deleted comment
Removed by moderator Rex: Removed a general status note that was posted on many GitHub threads. It no longer applies here.
Comment by @ajsnyde
RexSystem 1 vote Merged from #685 originally by @ajsnyde on GitHub
I'm also having issues that may be related to this when using flux on a non-default port. I feels related, but when initially creating an admin account on an instance, the networking tab shows a 200: image But immediately shows a 401({"code":"UNAUTHORIZED","message":"Unauthorized."}) for any related login etc and refreshes the page back to the 'create an admin' prompt: image I can't use the new admin user, but the email shows as already used in subsequent account creation forms. Relevant logs: 
api-1               | {"level":"warn","time":"2026-09-08T15:12:23.254Z","service":"fluxer-api","env":"production","domain":"asdf.asdf","code":"ESERVFAIL","error":{"type":"Error","message":"queryMx ESERVFAIL asdf.asdf","stack":"Error: queryMx ESERVFAIL asdf.asdf\n    at QueryReqWrap.onresolve [as oncomplete] (node:internal/dns/promises:297:17)","code":"ESERVFAIL","syscall":"queryMx","hostname":"asdf.asdf"},"msg":"Email DNS lookup failed with a transient error, allowing request"}
api-1               | {"level":"info","time":"2026-09-08T15:12:23.422Z","service":"fluxer-api","env":"production","method":"POST","path":"/v1/auth/register","status":200,"durationMs":529,"msg":"Request completed"}
seaweedfs-1         | I0908 15:10:18.199722 master_server.go:348 snapshot taken to persist TopologyId 61d770ff-257d-4471-b397-796e6fd399df
api-1               | {"level":"info","time":"2026-09-08T15:12:23.462Z","service":"fluxer-api","env":"production","method":"GET","path":"/v1/admin/instance/config","status":401,"durationMs":6,"msg":"Request completed"}
seaweedfs-1         | I0908 15:10:18.277268 volume_grpc_client_to_master.go:71 Volume server start with seed master nodes: [192.168.240.6:9333]
I have FLUXER_EMAIL_ENABLED=false in the .env, so I'm surprised to see the server presumably checking the DNS of the supplied email.
  • 647973850-291e4b07-e89a-4cf5-a7a5-5b7b7554dafb.png

    647973850-291e4b07-e89a-4cf5-a7a5-5b7b7554dafb.png

    1265×1022 | 257 kB

  • 647974376-76f4628e-7637-40e7-a6f9-2bc833d817f6.png

    647974376-76f4628e-7637-40e7-a6f9-2bc833d817f6.png

    2559×1014 | 639 kB

Permalink
Comment by @naelstrof
RexSystem 1 vote Merged from #685 originally by @naelstrof on GitHub
This has since been fixed, there's a new article on the setup instructions here: https://docs.fluxer.app/operator/reverse-proxy/ Or perhaps, I didn't understand what reverse proxy meant when I initially followed the instructions. Anyway, I took my completely stock installation and added the following to the .env: 
COMPOSE_FILE=docker-compose.yml:docker-compose.proxy.yml
FLUXER_EDGE_BIND=0.0.0.0:8080
Then I just pointed my reverse proxy to that port, and it works as expected.
Permalink