studio in Docker only listens on one interface

[patzm]

Bug report

• [x] I confirm this is a bug with Supabase, not with my own application.
• [x] I confirm I have searched the Docs, GitHub Discussions, and Discord. relevant might be https://github.com/orgs/supabase/discussions/21558, but rather related to #28105

Describe the bug

I found that the server is being launched from server.js: https://github.com/supabase/supabase/blob/3e279dac7e0e08ed5b8bab66cafca68d6ad99bc1/apps/studio/Dockerfile#L60

In the server.js, I found

const hostname = process.env.HOSTNAME || '0.0.0.0'

I tried to search this codebase but couldn't find the source of the file.

This results in the server only listening to studio:3000 but not on other interfaces it seems.

To Reproduce

Steps to reproduce the behavior, please provide code snippets or a repository:

1. Follow https://supabase.com/docs/guides/self-hosting/docker

2. attach at least two networks (network interfaces) to the studio container

3. inside the container, install requirements for debugging

   root@ba6dc8b3b0f8:/app# apt update
   root@ba6dc8b3b0f8:/app# apt install curl net-tools iproute2
   

   I have those two interfaces with the respective IPs

   1. eth1@if2497 with 172.20.0.10/16: this is the interface through which I would like to reach the server, which doesn't work
   2. eth0@if2511 with 172.21.0.13/16

   root@ba6dc8b3b0f8:/app# ip a
   1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
       link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
       inet 127.0.0.1/8 scope host lo
          valid_lft forever preferred_lft forever
       inet6 ::1/128 scope host
          valid_lft forever preferred_lft forever
   2496: eth1@if2497: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default 
       link/ether 02:42:ac:14:00:0a brd ff:ff:ff:ff:ff:ff link-netnsid 0
       inet 172.20.0.10/16 brd 172.20.255.255 scope global eth1
          valid_lft forever preferred_lft forever
   2510: eth0@if2511: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default 
       link/ether 02:42:ac:15:00:0d brd ff:ff:ff:ff:ff:ff link-netnsid 0
       inet 172.21.0.13/16 brd 172.21.255.255 scope global eth0
          valid_lft forever preferred_lft forever
   

4. check host resolution / server reachability

   root@ba6dc8b3b0f8:/app# curl localhost:3000
   curl: (7) Failed to connect to localhost port 3000 after 0 ms: Couldn't connect to server
   

   bad 👎

   root@ba6dc8b3b0f8:/app# curl studio:3000
   /project/default
   

   good 👍, as expected

5. check running servers

   netstat -tulpn | grep 3000
   tcp        0      0 172.21.0.13:3000        0.0.0.0:*               LISTEN      1/next-server (v14. 
   

   The server is only listening on the eth0@if2511 interface! That's the problem IMHO ⚠️.

Expected behavior

The server is available on all interfaces, not only the first. I can then wrap it with any reverse proxy, traefik in my case.

Fix

If I launch my container with the environment variable HOSTNAME=0.0.0.0, repeating the same steps as in 2.) - 4.), I see

root@525a492d6ae2:/app# netstat -tulpn | grep 3000
tcp        0      0 0.0.0.0:3000            0.0.0.0:*               LISTEN      1/next-server (v14. 

which is expected and now the studio is reachable on all interfaces as expected.

Also curl shows the expected results:

/project/defaultroot@525a492d6ae2:/app# curl studio:3000
/project/default
root@525a492d6ae2:/app# curl localhost:3000
/project/defaultroot

Screenshots

If applicable, add screenshots to help explain your problem.

System information

• OS: Docker image on a Linux (TrueNas Scale) server
• Browser: Firefox and Chrome latest
• Version of supabase-js: docker image supabase/studio:20240923-2e3e90c
• Version of Node.js: docker image supabase/studio:20240923-2e3e90c

Additional context

Add any other context about the problem here.