Dedicated Multiplayer Server Crash with only Sabel, Create, and Create Aeronautics Installed

[Mikeatron-User]

Context

Issue

I'm attempting to run a multiplayer server and the server crashes.

Context Information

Server Host: Bisect Hosting
Minecraft version: 1.21.1
Java version: 21.0.10
NeoForge version: 21.1.226

How to Recreate

1. Install the previously mentioned mods.
2. Turn on server.
3. Server Crashes

Mod List:

create-1.21.1-6.0.9.jar
create-aeronautics-bundled-1.21.1-1.0.2.jar
sable-neoforge-1.21.1-1.0.4.jar

Crash Report

https://mclo.gs/GX2IPRJ

[BlueMoonJune]

Looks to me like the same issue im experiencing. You should post your normal logs, crash reports don't say much here

[11oddloe]

this is a internal issue with the host your using(almost all hosts do this) to save on ram usage by deleating or modifying how things are called from mods and sable uses quite a bit of mixins witch is the main part that is modified by these hosts, my recomendation is to just host a localhost server on your own computer via the official neoforge server jar and then bypass port forwarding by using playit.gg allowing others to join the server

[RenPT23R]

it crashes also on 1.0.6 and 1.0.5 also on a client installed with these doesnt work iam using Modrinth app its not a server its a client

[uusai]

Четко и без воды: Попробуй удалить мод Create Aeronautics и Sable, должно запуститься!

[RenPT23R]

I made a tiny modpack that works with aeronautics and sable you must have some mod thats causing it that you didnt include in the list.

[Mikeatron-User]

I made a tiny modpack that works with aeronautics and sable you must have some mod thats causing it that you didnt include in the list.

Incorrect

[Mikeatron-User]

Issue is resolved, the server was not binding to the port.

From the Bisect Hosting Staff:

I got the server moved over to a new node to swap the IP, but both dedicated IP's available on this node failed to bind as well.
I then swapped the server over to a non-dedicated IP and the server started right up. If you would like to keep a dedicated port, we can attempt another node move once the backups finish transferring over from the previous node.

When a server fails to bind to a port, it is either because a server is already occupying that port (shouldn't have been the case here), or there is an issue with that port.
There was absolutely nothing wrong done on your end! 🙂

[Kringle-BH]

this is a internal issue with the host your using(almost all hosts do this) to save on ram usage by deleating or modifying how things are called from mods and sable uses quite a bit of mixins witch is the main part that is modified by these hosts, my recomendation is to just host a localhost server on your own computer via the official neoforge server jar and then bypass port forwarding by using playit.gg allowing others to join the server

Heya!

One of the BisectHosting peoples 👋 (Specifically I work in Game Integration, so I'd be who is behind messing with the load system 👀 ) wanted to clarify on this since I got roped in after this was mentioned to me, us (and most other hosts) as far as I know don't touch how the server loads its data (be it mixins/mods/plugins/kubejs/etc.); thats entirely on the Modloader in use or other scripts in use the Modpack / User installs - we're here to host it. Not mess with it! 👀

With that, we also dont do anything to worsen your experience just to save on RAM. We throw extra RAM to each container, not less! We eat the cost so you don't have to.

I can't speak 100% on any host but our own, but if you set it we let you run it 😅 Wouldn't do anyone any good if we on the backend we're quietly modifying these things and causing unneeded issue reports

After looking into this a bit myself, it seems like this came down to the bind that this mod was trying to use compared to what the user actually had

Seen here:

@Inject(method = "startTcpServerListener", at = @At("HEAD"))
  private void sable$startTcpServerListener(final InetAddress inetAddress, final int port, final CallbackInfo ci) {
      // ...
      this.channels.add(new Bootstrap()
              .channel(channelClass)
              .option(ChannelOption.SO_BROADCAST, true)
              .handler(/* ... */)
              .group(eventLoopGroup)
              .bind(port)
              .syncUninterruptibly());
  }

The method receives inetAddress (the server's configured bind IP) but the UDP bootstrap calls .bind(port) - as far as I can tell, nettys single arg overload binds to the wildcard address (0.0.0.0:<port>). On a dedicated-IP node where multiple servers
share port 25565 across different IPs, the OS rejects the wildcard bind because another process already owns that port on a specific IP

@ryanhcode pinging for vis, happy to make a PR if wanted 👀

To note, as far as I can tell this is quite literally replacing .bind(port) with .bind(inetAddress, port)) 👍