dedicated-servers

[deep turret]

yep, also as I tried to show : it may be several related events for each upgrade
which maybe would make it even more chaotic if 2 different upgrading chain of events where to get applied to the same belt support at the same time for instance

that's why in lot of game actions may get ignored when done too quickly : it is easier to handle safely than to process everything at the same time

a good solution is often buffering actions

so when it happens to be launched so close to each others : one just waits for the other in a fast seamless fashion

[clear jungle]

really really bad

did nothing for dedicated servers

[lapis gazelle]

What specifically makes it really really bad?

When we tried it... some of the hypertubes would teleport you around... train position kept rubber banding

[chrome pumice]

Biggest thing is that vehicles are a coin flip if your or the server will crash.

[lapis gazelle]

Seems like this stuff could be fixed by allowing the server to trust the clients

[clear jungle]

Crashes vehicles falling through the world. Vehicles not where they are showing on the map. Player falling through the world, the death crate showing up on the other side of the world but it isn't there or where you died. That's just for starters.

[barren spear]

Me and my friend are having uses running a ds, we are using linxusgsm but for some reason when we start it, it wont let us. There is no error message or log. It worked before this latest patch. So if anyone could help that would be appreciated.

[steel osprey]

Server binary Changed in Experimental and i Bet lgsm is Not updated yet

but ASK this in lgsm Discord 😉

[marble berry]

you just add a line to your sfserver.cfg

[marble berry]

add this to sfserver.cfg

executable="./UnrealServer-Linux-Shipping"

it's strange you get no error though. I got a message that said Executable does not exist

[lilac pine]

are radar towers broken right now... its not showing resource nodes on my map

[marble berry]

it has a strange distance limit. I couldn't figure out how to extend it

there's a smaller radius around your character, but the tower knows that it's there

might be a bug, might be a feature

[broken orbit]

oooo DS exp updated to U8 👀

[graceful flicker]

DS has been been on U8 for a few weeks now, no?

[broken orbit]

U8 was released on 13th
Only seen the DS update today when restarted the server box

[graceful flicker]

Ah gotcha, I'm 99% sure we've been on it on the current save since the U8 update that included Lumen.

My dedi warns me that the server will restart in x minutes, we only get it when an update has been pushed

[deep turret]

did you power it ?

[lilac pine]

Yea... I think it only marks nodes you walk near.. or it just took a long time.. but eventually nodes started to show up on the map

[hasty tangle]

how do you validate save files on dedicated server?

having some real problems since the last few patches; just crashes seconds after starting to play so just want to rule out my save files

[frosty gale]

There's a validate parameter. If you use the suggestions from the wiki they're already in there. I have it in my update script, so whenever I run it and there's no update, it validates.

(I assume it validates when updating as well, but the console output isn't entirely clear so I tend to just run it once in a while to make sure)

Oh, I didn't actually ask if you use steamcmd or not. But then I guess if you use the steam client then there's a validate in the gui there like for regular games.

[glad nest]

dont think that would validate the save files though.
You may want to try and load the save in singleplayer to potentially rule out dedi server fault i'd say

[frosty gale]

My bad, I totally overlooked the "save files" part of the question.

[simple flare]

Anyone with an auto-restart .bat file upon a definite server-crash? 👀

[lofty vortex]

Hey is dedicated servers working in update 8?

I tried with some friends but ther server is offline for everyone except the host, and he can play.

The port is well forwarded

[simple flare]

working decently well over here

[narrow raft]

There's three different ports that need to be forwarded

[simple flare]

Uhh could anyone else confirm that
FG.AutoSaveInterval
Is not working anymore?

[narrow raft]

The command is broken but you can still configure an autosave interval in the ini files

[simple flare]

yep! Was just a bit baffled that the console command wouldn't work! 😄

[deep turret]

#patch-notes
• Temporary fix for audio log spam on Dedicated Servers when using vehicles which would lead into clients being kicked out of the Server
yay

[warm oasis]

hello can anyone tell me how to connnect to my epic gmaes servre?

[zinc moon]

Fixed a crash on startup for Dedicated Server when the server query port was set to a value that exceeded the maximum value of an int16

there's the issue I was having, surprised it's a bug though, figured it was some new limit

[deep turret]

I dont get it though unsigned int16 is 65536 which is maximum port possible

unless they mean it was signed int16 in their code instead of unsigned

and that starting at 32768 it would fail ?

[crimson gorge]

I really dont understand how people can build huge megafactories with highways and stuff when the belt system is barely functional

[narrow raft]

That's what I'm assuming happened

What's wrong with the belt system?

[deep turret]

^

graphically ok, but mechanically it works fine

[crimson gorge]

i wouldnt mind if it was just graphically but its not. like for example when making a manifold of like 8 or so smelters, the belts either wont fill the machines or wont empty them, sometimes both

[deep turret]

did you check the belt maximum throughput ?

did you get how every machine have to fill for manifold to feed the next machines ?

[narrow raft]

belts sometimes don't connect properly when the tick rate is really low, i.e. when the server is performing a save. outside of belts placed during saves I haven't seen that happen myself

[deep turret]

for instance lets say you have 5 machines, it will take 250 items at the very least to feed the 6th machine, and that's only if every machines are downclocked to 0 as they will eat the items that is fed into them at the same time

it often happens belts are not really connect too when you place them with splitter they sometimes dont get items from the splitters

you also have to be carefull about belt directions and splitters directions

but that's mostly part of the gameplay

the only part that is not from gameplay is when belt fake being connected to splitters

and that can be anoying but it's often quick easy to just reconnect it

[crimson gorge]

Its sometimes the first machine in the system that claims it doesnt have input or sometimes output when it clearly does. Yes I find myself having to reconnect belts multiple times to get it to work, and even then it'll work for a short while and then break again

Also sometimes the belt will be connected between an output of a machine and a splitter the correct way and the belt will be running backwards

[deep turret]

if it work when you load the same and dont touch it again, it is very unlikely to break after that

[deep turret]

that is only graphical though

but yea it is annoying

[tulip bronze]

Hi, im not sure if this is the right thread to ask but im looking for someone to try and connect to my server to see if it works and whether the port forwarding is working.
Im not looking for a group so I dont think that thread is correct. Thanks

[deep turret]

it's probably the good channel/thread for that
as it is a kind of troubleshouting

you really have no other person/friend to test it though ?

[tulip bronze]

yeah i usually do, i have a mate whom i played with but he alr went to bed

[tulip bronze]

Can you help me out? I would just dm you my ip and try and connect?

[zinc moon]

@deep turret agreed, happens all the time.

[dapper bay]

Anyone has any issues with dedicated servers on Experimental the game type keeps chaning back to Private from Friends only? I have saved the game on my computer as Friends only and uploaded to the dedicated server and it's went to Private.

[narrow raft]

The dedicated server isn't connected to steam/epic so the concept of friends only doesn't exist. either people join with ip and a password or you don't set a password and anyone can join

[limber gazelle]

Oh, dedicated got a little love in this patch 🙂

[deep turret]

ok so we did a test with @tulip bronze about timeouts (connecting through ipv4) :

• no ipv6 AND no special arguments : works
• ipv6 AND no special arguments : doesnt work at all
• no ipv6 AND multihome=0.0.0.0 : works only locally
• no ipv6 AND multihome=localip : works only locally
• no ipv6 AND -DisablePacketRouting : works only locally
• ipv6 AND multihome=0.0.0.0 : works only locally
• ipv6 AND multihome=localip : works only locally
• ipv6 AND -DisablePacketRouting : works only locally

I let you correct what I forgot

[tulip bronze]

no ipv6 and multihome=0.0.0.0 works only locally

so basically whenever multihome/DisablePacketRouting is used, it would connect but only locally (using public ip would timed out)

if multihome/DisablePacketRouting is not used AND IPv6 is enabled, noone can connect (timed out)
only if ipv6 is disabled AND no multihome/disablepacketrouting arg used, then everyone can connect

[deep turret]

so my truth table it ok ?

no ipv6 AND multihome=0.0.0.0 : doesnt work ? ??

[tulip bronze]

just have to fix that yeah

[tulip bronze]

this

[deep turret]

ok so now it's ok ?

[tulip bronze]

whenever multihome or disablepacketrouting is used, it would work locally no matter if ipv6 is enabled or disabled

yep

[deep turret]

corrected full table

[narrow raft]

is that with a local ipv6 or with full ipv6 connectivity? in case of the latter, did you configure the ipv6 firewall in your router for the right ports?

[deep turret]

as stated it is with ipv4 connection

[narrow raft]

if you have ipv6 enabled and no -multihome you have ipv6 connection bound on the server as well which it will advertise

[deep turret]

I wonder if my connection would work from public network though, but I didnt have ability to test recently yet

[deep turret]

but its not supposed to advertise link local addresses, and/or catch all binds, and the client is not supposed to try it when the connection is started from another ip

the connection must be established with the currently used ip that hello the server

this would confirm the client is not correctly handling multiple ip testing, it must test multiple ips ONLY when trying to first connect/find the server FROM DNS records

as if you stated a fixed ip in your servermanager it should use that

and at the very moment the connection is starting to get established the ip should be kept the same for all the session, and all the ports of the session

and that's where it looks like it fail

also the server is not correctly handling multihome nor DisablePacketRouting in its way to wait for session establishment and advertising its ips

[quiet quarry]

Hello, is there a way to properly shutdown a SteamCMD server from the command line?

[deep turret]

yes

[deep turret]

are you on linux or windows though, I assume you are on linux ?

[quiet quarry]

No im on windows, have a .bat file that starts the server in a dos box

[deep turret]

you can confirm it is closed in the taskmanager

[quiet quarry]

ok, thats what I've been doing, but it felt wrong...always feared it might corrupt a save

thanks 🙂

[deep turret]

maybe ctrl+c works too

maybe less scray

[deep turret]

ok apparently you must do ctrl+c

not closing the windows

but it suprise me the server cant listen gracefully to windows close event when the windows is closed

[quiet quarry]

noted thanks 🙂

[deep turret]

apparently things running inside the windows have 10second to close but I dont know if the server is listenning to CTRL_CLOSE_EVENT

ctrl+c is CTRL_C_EVENT

[hasty tangle]

no worries, done all the steamcmd bit - seems to a new patch out this afternoon so we will apply that and see how she blows

[quiet quarry]

yeah just ran the update and got a version mismatch after restarting, ran update again and after that it was ok

[hasty tangle]

yeah just updating my end - done the server and i'll how we get on or not! if not i'll validate the save file locally

maybe more RAM will make it better 😄 went from 16GB to 32GB and it's been a lot more stable now the factory is stupid size

need to finish my plutonimum build 😐

[deep turret]

what do you have running outside of satisfactory dedicated server ? because 16GB used to be largely enough, and u8 decreased base RAM requirement

[hasty tangle]

nothing just running a light version of ubuntu with a i3 10th gen and a GTX1050 I

[quiet quarry]

I run my dedicated server on Win11 on a Small form factor Lenovo pc with an i5-6500T & 16Gb, its runs noticably better with U8. I had areas where my frames dropped into the single digits, and now its double digits everywhere. 😄

[deep turret]

I'm finally going to test the Dunes

as a benchmark

(one of the heaviest saves outthere from one of people going by this server)

[quiet quarry]

@wicked geyser
I used a .bat file to update mine;

cd c:\Steamcmd
start steamcmd.exe +login anonymous +app_update 1690800 -beta experimental validate +quit
pause

[wicked geyser]

the update run successfully, but it is not starting

[tiny cape]

welp - the temporary fix for tractors...

made it worse and better at the same time.

Game disconnects me, I can reconnect and it no longer kills me and places the tractor at 0,0 like it did to me last night...

but now the server kicks me a lot more readily 😠

[quiet quarry]

it didnt start after the first try for me either, I closed the window and restarted it, then it started but gave a version error ingame, ran update again, after that all was ok..

[tiny cape]

Scratch that... I now get immediately booted when I join ;^;

[quiet quarry]

Are invisible objects (mostly noticable on traintracks & pipes) a known issue in U8? If so, any way to avoid it besides selecting them with the remove tool? (makes them visible, but they keep going invisible all over the place)

[deep turret]

I'm going to hell guys 😎 already 18GB RAM used at server launch
now going to connect to the game and start it

[visual wasp]

hello I downloaded the "additional content dedicated server for experimental" but I am having trouble to setting it up, does anyone have like a nice tutoriol or just advice ? Thank you !

[wicked geyser]

What means trouble 🙂 ?

[clear lagoon]

https://satisfactory.wiki.gg/wiki/Dedicated_servers

[visual wasp]

I can send u the logs if it helps

[visual wasp]

ty gonna take a look

[tidal relic]

@clear lagoon after turning off the pausing for when no one is present, I have not had another error about timing out

[clear lagoon]

huh interesting. I can imagine it causing a processing spike when unpausing 🤔
I never noticed because I have pausing disabled

[tidal relic]

I remember reading somewhere that hyperloops were awfully buggy on dedicated servers - is that still the case?

[clear lagoon]

hyper_loops_?

[tidal relic]

Hyptertubes*

[clear lagoon]

if you mean cyclotron hypertube cannons, they need to be altered slightly because of the new phy-

oh

[tidal relic]

Sorry, for some reason I was thinking it was named differently

[clear lagoon]

if hypertubes send you backwards you need to rebuild the first tube segment by starting from the entrance, direction matters.
long hypertubes have a tendency of desyncing your movement which may cause disconnects

ymmv

[warped coral]

Between yesterday and today my steamcmd stopped allowing me to login, and I cannot download the latest update 8 patch as a result. Nothing was changed on the host system (linux 64bit) to my knowledge. Satisfactory is my first time using steamcmd, anyone have any debugging tips?

Redirecting stderr to './Steam/logs/stderr.txt'
Looks like steam didn't shutdown cleanly, scheduling immediate update check
[  0%] Checking for available updates...
[----] Verifying installation...
Steam Console Client (c) Valve Corporation - version 1686881296
-- type 'quit' to exit --
Loading Steam API...dlmopen steamservice.so failed: steamservice.so: cannot open shared object file: No such file or directory
dlmopen libSDL3.so.0 failed: libSDL3.so.0: cannot open shared object file: No such file or directory
OK

Steam>force_install_dir /speed/steam/satisfactory_server

Steam>login anonymous

Connecting anonymously to Steam Public...OK
Waiting for client config...OK
Waiting for user info...FAILED (Timed out)
OK

Steam>app_update 1690800 -beta experimental validate
ERROR! Failed to install app '1690800' (Missing file permissions)```

[tulip bronze]

@deep turret I decided to redownload the server in a debian lxc container and multihome/disablepacketrouting now works for the public ip

[clear lagoon]

Alador: the user you're running steamcmd with does not have write permission in /speed/steam/satisfactory_server

[tulip bronze]

so ig its an issue with the ubuntu container itself or maybe some existing settings that somehow interfere?

[rare summit]

I'm using Apex hosting for our groups dedicated server since update 8 hit and we've been having consistent timeout/failed to connect and laggy sessions. Its a new save and we're barely using 5MB of ram so far. I've tested my connection to the server for packet loss with no problems. Anyone else having issues like these? Is this update 8 until things get ironed out or should I be looking for other hosting? I've already migrated our server once with apex.

[clear lagoon]

Are you seeing these connection issues when you're not using vehicles or hypertubes? Any chance you're timing out during the auto-save? How large is your save?

For failed connections I always suggest as a first step to launch the game server with the parameter -multihome IPADDRESS added before -log -unattended, if you have the possibility to change launch parameters

[deep turret]

interesting to know, thank you, so reinstalling everything may make it work as intedended as u7 ?

[deep turret]

it's permission access issues as Lena said

[deep turret]

you mean your save file is 5MB ?

[rare summit]

Yes even when not using vehicles, its a new save we are in phase 3. I will say that none of us can go anywhere near northern forest. It seems to cause serious lag for everyone if someone goes there.

I mean our dedicated server is only using about 5mb ram with us all on

[deep turret]

its not using 5mb ram

[rare summit]

gb

sorry

[deep turret]

that makes much more sense

[rare summit]

I had a tech check on the multihome param but they said the server is running IPv4 so shouldnt be needed

[deep turret]

that makes no fucking sense, you especially want this setting if the server is intended to run on ipv4 only

it set how the server is listenning to network

[rare summit]

I'm just a guy that wants to play satisfactory with my friends

[clear lagoon]

multihome is just to configure which address to bind to, instead of all available network interfaces. chances are the server provided by a game hosting co. only has a single public ipv4 nic

[rare summit]

my man

[clear lagoon]

If you're having issues but don't have direct access to the machine running your server, I suggest you contact support for the company providing your service

[deep turret]

its how the server bind to network i.e internal network of the hosting container, not how it can talk to outside world, so unless they specifically disabled any ipv6 networking inside the dedicated environment, you want this setting

[rare summit]

we've had a server before and gotten to late game with none of these issues. So really im just trying to determine if i need to switch hosts or if things will improve as patches come out.

[clear lagoon]

all we can do is guess at the architecture at the hosting co.'s end, either of us could be right but it's not really meaningful to discuss this further 😛

[deep turret]

anyway dont worry about what I'm talking about, because that are specifics you probably dont have access to
so you are not responsible for these anyway, all you are legitimate to ask is a working server

[clear lagoon]

you can have working networking on the current dedicated server patch for U8. if your hosting provider can't provide the necessary support, you can draw your own conclusions I believe

[deep turret]

btw, do someone know the .ini for setting the autosave interval ?

[rare summit]

Gotch, thanks for your help ya'll! Sorry for being a pleb 🙂

[clear lagoon]

No worries at all 🙂 you're paying for a managed service so you don't really have the necessary access to manage or break things. your provider is entirely responsible and it's what they're being paid for too, imho.

[deep turret]

you dont have to be sorry, it's just lot of us are selfhosting, so we usually have all power on dedicated server settings, and as we are experimenting, we usually are full of specific questions

[warped coral]

@clear lagoon @deep turret I believe the "file" it is talking about is in relation to not being able to access the dedicated server app from Steam due to the login failure. I just ran chmod o+rw /speed/steam/satisfactory_server and still get the same result.

shoot I forgot -R

one sec lol

[deep turret]

make sure your current user is the owner of the /speed/steam/satisfactory_server too

[warped coral]

it is, I checked with stat

[deep turret]

otherwise o+rw have no effect (obviously)

ok nice

wait I'm stupid

did I confused o and u

let me check I'm stating the obvious of how I'm wrong

[clear lagoon]

can you post the output of whoami and ls -lhd /speed/steam/satisfactory_server ?

[warped coral]

wait shoot now I don't remember either

[clear lagoon]

o = others, u = user, g = group

[deep turret]

a is for all, g for group o u

damn

so I was wrong

[warped coral]

dangit

[deep turret]

I'm stupid

at least I reacted instantly on my mistake but still I'm letting this message live for documentation purpose

[warped coral]

pleb@pleb7573 /s/s/satisfactory_server> whoami
pleb
pleb@pleb7573 /s/s/satisfactory_server> ls -lhd /speed/steam/satisfactory_server
drwxr-x--x 6 pleb pleb 4.0K Jun 19 20:10 /speed/steam/satisfactory_server/
pleb@pleb7573 /s/s/satisfactory_server>```

[deep turret]

"(ObViOuSlY)"

smh

[deep turret]

so you done -R now ?

[warped coral]

Yes

[warped coral]

Is this part relevant in any way?
Waiting for user info...FAILED (Timed out)

[deep turret]

^ if someone know, let me know

[clear lagoon]

1 sec, I posted it earlier

[warped coral]

Oh it's on the wiki somewhere

[clear lagoon]

SharklienX: #dedicated-servers message

[fair osprey]

• Made the Experimental Label bit more transparent and less bright red to help prevent OLED Burn-in

tbh having an option to remove the label or change it to something other than red would be better, red is still bad for OLED displays overall.

[clear lagoon]

haven't seen that before. 1 sec

[deep turret]

it's not unfortunately, at least not in dedicated server, nor dedicated server config files

[clear lagoon]

That is in Engine.ini

[/Script/FactoryGame.FGSaveSession]
mNumRotatingAutosaves

and in GameUserSettings.ini

[/Script/FactoryGame.FGGameUserSettings]
mFloatValues=(("FG.AutosaveInterval", 120.000000))

if you have existing mFloatValues, you separate as follows:

mFloatValues=(("some other setting", 123.450000), ("FG.AutosaveInterval", 120.000000))

[warped coral]

I found a console command, nothing about the ini file though https://satisfactory.wiki.gg/wiki/Dedicated_servers#Console_commands

oh but looks like Lena's got you covered

[clear lagoon]

just copy pasted my previous post XD

@warped coral can you run ls -lhd $HOME/.steam and ls -lhd $HOME/.steam/steamcmd for me?

[frank laurel]

@clear lagoon you're my only hope, can I also get some assitance?

[warped coral]

pleb@pleb7573 ~/satisfactory_server> ls -lhd $HOME/.steam
drwxr-xr-x 7 pleb pleb 4.0K Jun 21 19:55 /speed/steam/.steam/
pleb@pleb7573 ~/satisfactory_server>
pleb@pleb7573 ~/satisfactory_server> ls -lhd $HOME/.steam/steamcmd
drwxr-xr-x 9 pleb pleb 4.0K Jun 21 20:50 /speed/steam/.steam/steamcmd/
pleb@pleb7573 ~/satisfactory_server>```

[frank laurel]

can anyone explain to me why is my train station getting placed to 0,0,0 coordinates?

you can see me on the map there, my train is in the station as well

[warped coral]

it's just a bug with the map right now

[clear lagoon]

is the map icon there or is the station effectively being moved?

[deep turret]

yep that's it, I'm searching .ini setting and it's not in wiki 👍

[frank laurel]

train station is not there, but also cannot enable autodrive

cause it tells me train station unreachable

like wtf

[deep turret]

Ooh so "mFloatValues" is for console start command kinf of

[clear lagoon]

I wouldn't know tbh. you can change the auto-save interval in the game client, so I did and diffed the ini files. that was the only difference, tried it on the dedicated server and it worked 🤷‍♀️

[deep turret]

I searched but did not find it, you did find it, thank you
so I had it already set

I can check now

I was sure the server would save it somewhere but did find it

I wanted to check if it changed from u7 to u8

I will reinstall my server that's why

[clear lagoon]

just to make sure, you're not trying to place the train station outside of the map bounds? i.e. very far in the sky or under the map?

[frank laurel]

let me show what I am trying to do

i have three stations, my home => oil factory => bauxite mine

[deep turret]

I had ```[/Script/FactoryGame.FGGameUserSettings]
mIntValues=(("FG.NetworkQuality", 3))
mFloatValues=(("FG.AutosaveInterval", 900.000000))

[clear lagoon]

any chance you accidentally ran steamcmd with sudo/doas at some point? you can try deleting the $HOME/.steam directory

[frank laurel]

i have a bi-directional train that so far went from A to B and back to A

I expanded that line now to go A to B to C

and back to A

but it goes A to B and says C is unreachable

it's not in the sky by any means, i placed it literally on the ground on a 4m platform

it goes "high" like it has to go up for the bauxite

[clear lagoon]

can you try dismantling and rebuilding the train station?

[warped coral]

I actually already did try deleting that directory. I don't think I've ever ran steamcmd as root, let me search through my command history

[frank laurel]

yes

i did that with every station in the chain

still unreachable

[clear lagoon]

if that doesn't work, try loading the save in single player and doing it there

short of that I don't really have any ideas

[frank laurel]

A to C works

wtf

but if I do A to B to C

it dies

ayo what

i made sure my stations were facing correctly

[clear lagoon]

I haven't played around with trains yet myself so I'll have to defer to the others here

[frank laurel]

oh okay

[clear lagoon]

do you have any dns-based network blocking going on or dns issues on that box? can you ping steampowered.com on there?

[warped coral]

Ok well I found my issue -ish

I had a fish script /speed/steam/steamcmd.fish that changed my home directory to /speed/steam (because that's the SSD)

#!/usr/bin/fish
set -x HOME (dirname (status filename))
echo $HOME
exec steamcmd $argv```

Evidently something about my stupidly simple script is wrong, because if I set my home manually in a live fish shell and then launch steamcmd directly it all works.

anger 💢

oh, $HOME was getting set to . for some reason.

guess I'll just hardcode the path in 😢

[frank laurel]

ok I got the train working, sorry for the false alarm. i have 0 idea what I did to make it start, but hey, it works

[clear lagoon]

keep in mind that's kinda hacky because you're also not updating your XDG environment. but it might work for steamcmd

[warped coral]

shoot, I always forget about XDG variables. I'll keep that in mind 👍

[clear lagoon]

you know you could just create a new user that has its actual home on the SSD 😛

[warped coral]

https://tenor.com/w6TQ.gif

aww no gifs here (understandably)

[warped coral]

dang, never thought of that

[clear lagoon]

obligatory clicked it ❤️

[deep turret]

full reinstall of the server, editing the .ini files back, and restarting the server : done

[deep turret]

my dedicated server have its own user if that is inspiring to you

its home is in /srv/Satisfactory/

[clear lagoon]

hopefully running it in systemd or other service manager specifically as that user as well and not as root? :3

[deep turret]

yep

with filters on

[clear lagoon]

ok you're doing better than me at that point then XD

[deep turret]

#dedicated-servers message @clear lagoon

[clear lagoon]

don't you risk updating the running dir with additional instances if an update is released between spinning them up?

[deep turret]

good question

never thought of that

my original file didn't have it damned

[clear lagoon]

I went through the trouble of adding a socket for stdin before realising that the server console appears to be noninteractive XD

[deep turret]

yep
useradd -M -d /srv/Satisfactory -s /usr/bin/nologin satisfactory

So no interactive shell possible

[deep turret]

I will work on a check now, and will update it

[lapis gazelle]

Does the server need to restart in order for newly downloaded blueprints to appear?

[clear lagoon]

I haven't tested it myself, but it sounds trivial to find out

[deep turret]

I guess it would only need the game to be relaunched

[deep turret]

I'm thinking of implementing it with a dependent update service

[hasty tangle]

so pretty much every time I connect to the server - play for 10-20 seconds it kicks me off with an unreal error

have validated all the files - done some verifying of the games , any more ideas before I return to 7?

[deep turret]

do you have enough ram ? apparently it can take more ram on huge saves

[hasty tangle]

32GB RAM

[deep turret]

32GB free ram ? for the server to use ?

[hasty tangle]

just going to try and find the logs to see why it's crashing time I think

[tiny cape]

That's a fair whack of RAM compared to what mine is at currently...

Only using about 3 of the 16GB i have allocated to my LXC

That said it's only tier 4 and we've only just started mining the Northern side of the map

[hasty tangle]

We are using a dynamic DNS on it so i'm going to take that out and just use the IP and see if that makes any difference

Tier 8, end of game here with way too much stuff and a complex train network 😄 I think maybe we broek it

[tiny cape]

Haha that would explain the RAM usage

But uh yea... doubt it's your DynDNS or mine would also go screwy

[deep turret]

having a unit stop only when a bunch of other units are all down is a headache

[clear lagoon]

maybe just don't have systemd try to do automated package management for you XD

if you're running a multi-instance server, manually bringing it down and updating isn't that painful

[clear lagoon]

if you see the UE crash reporter, can you share the stacktrace?

[deep turret]

but I should at the very least make a update service that have conflict with server running

I may hack the way systemd handle putting down services and putting up services

[hasty tangle]

@deep turret one of these?

[deep turret]

you didnt ping the right person @clear lagoon ^

[deep turret]

and do both conflict and and requires statemetns

[hasty tangle]

sorry, half asleep 😄

[deep turret]

I wonder how systemd look at me if I put in the same file :

Requires=satisfactory-update.service```

[clear lagoon]

pretty sure the stack will also be in the crashcontext runtime xml

[hasty tangle]

oh yeah I can see lots of stack stuff in there; want me to share in here @clear lagoon

[clear lagoon]

yes please

[clear lagoon]

you crashed during memory allocation. how much RAM do you have? could you run memtest86 to validate your RAM?

[hasty tangle]

on the client or on the server?

64GB on my PC and 32GB on the Dedicated Server

[clear lagoon]

whichever one generated that crash report 🙂

[hasty tangle]

that was my PC

[clear lagoon]

you have plenty of RAM in your pc but it appears to be running an unstable overclock, might not be seated properly, or might otherwise be faulty

run a full memtest86

<CallStack>FactoryGame_Core_Win64_Shipping!FMallocBinned2::ReallocExternal() [C:\BuildAgent\work\78a794e88763017d\UE4\Engine\Source\Runtime\Core\Private\HAL\MallocBinned2.cpp:996]
FactoryGame_Core_Win64_Shipping!FMallocBinned2::Realloc() [C:\BuildAgent\work\78a794e88763017d\UE4\Engine\Source\Runtime\Core\Public\HAL\MallocBinned2.h:607]
FactoryGame_FactoryGame_Win64_Shipping!TArray&lt;FReplicationDetailData *,TSizedDefaultAllocator&lt;32&gt; &gt;::ResizeGrow() [C:\BuildAgent\work\78a794e88763017d\UE4\Engine\Source\Runtime\Core\Public\Containers\Array.h:3024]
FactoryGame_FactoryGame_Win64_Shipping!`AFGConveyorItemSubsystem::GatherTransformData_ISPC&apos;::`10&apos;::&lt;lambda_3&gt;::operator()() [C:\BuildAgent\work\78a794e88763017d\UE4\FactoryGame\Source\FactoryGame\Private\FGConveyorItemSubSystem.cpp:766]
FactoryGame_FactoryGame_Win64_Shipping!`ParallelForImpl::ParallelForInternal&lt;TFunctionRef&lt;void __cdecl(int)&gt;,`ParallelFor&apos;::`2&apos;::&lt;lambda_1&gt;,std::nullptr_t&gt;&apos;::`2&apos;::FParallelExecutor::operator()() [C:\BuildAgent\work\78a794e88763017d\UE4\Engine\Source\Runtime\Core\Public\Async\ParallelFor.h:116]
FactoryGame_FactoryGame_Win64_Shipping!LowLevelTasks::TTaskDelegate&lt;LowLevelTasks::FTask * __cdecl(bool),48&gt;::TTaskDelegateImpl&lt;`LowLevelTasks::FTask::Init&lt;`ParallelForImpl::ParallelForInternal&lt;TFunctionRef&lt;void __cdecl(int)&gt;,`ParallelFor&apos;::`2&apos;::&lt;lambda_1&gt;,s() [C:\BuildAgent\work\78a794e88763017d\UE4\Engine\Source\Runtime\Core\Public\Async\Fundamental\TaskDelegate.h:171]
FactoryGame_Core_Win64_Shipping!LowLevelTasks::FTask::ExecuteTask() [C:\BuildAgent\work\78a794e88763017d\UE4\Engine\Source\Runtime\Core\Public\Async\Fundamental\Task.h:656]
FactoryGame_Core_Win64_Shipping!LowLevelTasks::FScheduler::TryExecuteTaskFrom&lt;&amp;LowLevelTasks::TLocalQueueRegistry&lt;1024&gt;::TLocalQueue::DequeueGlobal,0&gt;() [C:\BuildAgent\work\78a794e88763017d\UE4\Engine\Source\Runtime\Core\Private\Async\Fundamental\Scheduler.cpp:360]
FactoryGame_Core_Win64_Shipping!LowLevelTasks::FScheduler::WorkerMain() [C:\BuildAgent\work\78a794e88763017d\UE4\Engine\Source\Runtime\Core\Private\Async\Fundamental\Scheduler.cpp:402]
FactoryGame_Core_Win64_Shipping!UE::Core::Private::Function::TFunctionRefCaller&lt;&lt;lambda_2d21a7280907265e6aab37ddf50d42b1&gt;,void __cdecl(void)&gt;::Call() [C:\BuildAgent\work\78a794e88763017d\UE4\Engine\Source\Runtime\Core\Public\Templates\Function.h:475]
FactoryGame_Core_Win64_Shipping!FThreadImpl::Run() [C:\BuildAgent\work\78a794e88763017d\UE4\Engine\Source\Runtime\Core\Private\HAL\Thread.cpp:70]
FactoryGame_Core_Win64_Shipping!FRunnableThreadWin::Run() [C:\BuildAgent\work\78a794e88763017d\UE4\Engine\Source\Runtime\Core\Private\Windows\WindowsRunnableThread.cpp:149]</CallStack>

crash during malloc

memory allocation

[hasty tangle]

i'll give it a test and see what it comes up with; I was starting to think it was my PC

[clear lagoon]

if you fail memtest86, try reseating your RAM. if you have EXPO/XMP enabled you can try disabling it and seeing if it then passes memtest86. if it does, you can try running with XMP/EXPO enabled with only a single RAM stick slotted in. try both. if one of them fails you know it's probably defective

just giving you the whole rundown since I'll be asleep soon 😛

[hasty tangle]

Thanks; yeah me too ill take a look on the morning 🙂

[fervent marsh]

can someone help me turn on the settings where you keep items on death on my dedicated server? do i have to edit a config file?

[deep turret]

I will try :

After=satisfactory-update.service
Wants=satisfactory-update.service```

it's a bit hacky but may work

[clear lagoon]

it's part of AGS

turn on advanced game settings on the save

if you're on U8 that is

[deep turret]

this is the moment for testing 😣

deep breath

[clear lagoon]

everyone get ready. he's doing witchcraft

xD

[fervent marsh]

i see ty

[deep turret]

ok, it correctly shutdown satisfactory server before updating, but it refuses starting update before starting server

I may try to use an intermediate satisfactory-update.target and see how it works

[clear lagoon]

I kinda feel responsible for having sent you on this tangent xD

[deep turret]

this is hell, if I do like I wanted to do, starting any instance would lead to stopping other instance for start update

this is a headache

I would need a Requires=satisfactory-update.service OR satisfactory-update.instance

or alike

I think I'll go to non automatic update

as automatic update dont make much sense for multiple instances

[deep turret]

but it's important it doesnt try to update while instances are running, you were completly right

ok I'll update like that, systemd is not flexible enough to implement this cleanly, manual updates will be needed

[clear lagoon]

if you really wanted to you could see if you can have steamcmd run an update check without actually updating on a timer maybe. if there's an update you can have it interact with systemctl to bring down all instances, run the update and spin them back up after. but that's just an idea xD

[deep turret]

systemd doesnt have any kind of 1:m relationship

it only have #x 1:1

so I cant ever make systemd aware of instances as a group

and it sucks

so I'll do manual update

¯_(ツ)_/¯

systemd satisfactoryd-update.service file :

Description=Satisfactory Dedicated Server Update
Wants=network-online.target
After=syslog.target network.target nss-lookup.target network-online.target

[Service]
Type=oneshot
User=satisfactory
Group=satisfactory
StandardOutput=journal
Restart=on-failure
Environment="LD_LIBRARY_PATH=./linux64:$LD_LIBRARY_PATH"
ExecCondition=/usr/bin/bash -c '! compgen -G "/srv/Satisfactory/instance-*.active" > /dev/null'
ExecStart=/usr/bin/steamcmd +force_install_dir /srv/Satisfactory +login anonymous +app_update 1690800 +quit
WorkingDirectory=/srv/Satisfactory
NoNewPrivileges=true
ProtectSystem=full
ProtectHome=true
PrivateTmp=true
PrivateDevices=true
ProtectKernelTunables=true
ProtectKernelModules=true
ProtectKernelLogs=true
ProtectControlGroups=true
LockPersonality=true
MemoryDenyWriteExecute=false
RestrictRealtime=true
RestrictSUIDSGID=true
RemoveIPC=true
SystemCallFilter=@system-service
SystemCallArchitectures=native x86

[Install]
WantedBy=multi-user.target```

systemd satisfactoryd.service file :

Description=Satisfactory Dedicated Server
Wants=network-online.target
After=syslog.target network.target nss-lookup.target network-online.target
After=satisfactoryd-update.service
Wants=satisfactoryd-update.service

[Service]
Type=simple
User=satisfactory
Group=satisfactory
StandardOutput=journal
Restart=on-failure
Environment="LD_LIBRARY_PATH=./linux64:$LD_LIBRARY_PATH"
EnvironmentFile=/srv/Satisfactory/server-Default.conf
ExecStartPre=/usr/bin/touch /srv/Satisfactory/instance-Default.active
ExecStopPost=/usr/bin/rm /srv/Satisfactory/instance-Default.active
ExecStartPre=/usr/bin/bash -c '! /usr/bin/systemctl is-active --quiet satisfactoryd-update.service'
ExecStart=/srv/Satisfactory/FactoryServer.sh ${OTHERSETTINGS} -ServerQueryPort=${QUERYPORT} -BeaconPort=${BEACONPORT} -Port=${PORT}
WorkingDirectory=/srv/Satisfactory
KillSignal=SIGTERM
NoNewPrivileges=true
ProtectSystem=full
ProtectHome=true
PrivateTmp=true
PrivateDevices=true
ProtectKernelTunables=true
ProtectKernelModules=true
ProtectKernelLogs=true
ProtectControlGroups=true
LockPersonality=true
MemoryDenyWriteExecute=false
RestrictRealtime=true
RestrictSUIDSGID=true
RemoveIPC=true
SystemCallFilter=@system-service
SystemCallArchitectures=native

[Install]
WantedBy=multi-user.target```

systemd satisfactoryd@.service file :

Description=Satisfactory %i Dedicated Server
Wants=network-online.target
After=syslog.target network.target nss-lookup.target network-online.target
After=satisfactoryd-update.service
Wants=satisfactoryd-update.service

[Service]
Type=simple
User=satisfactory
Group=satisfactory
StandardOutput=journal
Restart=on-failure
Environment="LD_LIBRARY_PATH=./linux64:$LD_LIBRARY_PATH"
EnvironmentFile=/srv/Satisfactory/server-%i.conf
ExecStartPre=/usr/bin/touch /srv/Satisfactory/instance-%i.active
ExecStopPost=/usr/bin/rm /srv/Satisfactory/instance-%i.active
ExecStartPre=/usr/bin/bash -c '! /usr/bin/systemctl is-active --quiet satisfactoryd-update.service'
ExecStart=/srv/Satisfactory/FactoryServer.sh ${OTHERSETTINGS} -ServerQueryPort=${QUERYPORT} -BeaconPort=${BEACONPORT} -Port=${PORT}
WorkingDirectory=/srv/Satisfactory
KillSignal=SIGTERM
NoNewPrivileges=true
ProtectSystem=full
ProtectHome=true
PrivateTmp=true
PrivateDevices=true
ProtectKernelTunables=true
ProtectKernelModules=true
ProtectKernelLogs=true
ProtectControlGroups=true
LockPersonality=true
MemoryDenyWriteExecute=false
RestrictRealtime=true
RestrictSUIDSGID=true
RemoveIPC=true
SystemCallFilter=@system-service
SystemCallArchitectures=native

[Install]
WantedBy=multi-user.target```

and voila

when update is started every instances are putted down

at least its clean

maybe I should do it as oneshot

f

I must add a check for instances to not kill this service

so I must bypass systemd anyway

I will go by my first mechanism idea then

touch files

[clear lagoon]

use PID files?

[deep turret]

I will use a more secure way, I dont want to update if instances somehow crashs

or orther things

ExecStopPost=/usr/bin/rm /srv/Satisfactory/instance-%i.active
ExecStartPre=/usr/bin/bash -c '! /usr/bin/systemctl is-active --quiet satisfactory-update.service'```

[deep turret]

now I'm considering this

man that rabbit hole is deep

[clear lagoon]

it's only as deep as you want to make it xD

[deep turret]

wonder if check is needed on instances if After is put in them on a oneshot type of update.service and without conlficting enabled

I will test that

I want them to automatically wait for update to finish successfully

would be the cleanest way possible and I would be able to make it automatic at any first instance started again

current satisfactory-update.service condition :
ExecStartPre=/usr/bin/bash -c '! compgen -G "/srv/Satisfactory/instance-*.active" > /dev/null'

return false if any instance active file is found

and no instance can be running without one being there

next step will be a update service that cancel and restart instances if launched manually

but maybe tomorrow

@clear lagoon IT WORKS !!!!!!!!!!!!!!!

🎉

[deep turret]

the 3 files

• when any instance started : update.service will simply fail
• when update.service is running, instance will wait for it to finish before starting
• when instance start and no other instance is started : it start update.service and wait for it
• when instance start and other instance are started : it simply start directly and no update is performed

update.service can be standalone launched when no instance is running, and without launching any instance

if instance is running as stated : it simply successfully fails

😎

now I could try to make the things a bit more clean and not polute systemd state with degraded server-update.service

but at least it works

@neon rapids improved multi-instance / standalone update | still automatic at single instance start -handling ^

[deep turret]

• making clean degradation of autocanceled update.service

maybe satisfactoryd-force-update.service
and rename satisfactory-update.service to satisfactoryd-update.service
to not conflict with the game, on linux gamers' pc

Note: these update services do not select experimental/public branches
these are completly branch agnostic

[neon rapids]

Hmm, it's there a TLDR? Not really sure I'm tracking what y'all are talking about, sorry.

[deep turret]

the 3 files @neon rapids I'm talking about, you can skip the conv

[deep turret]

the result behavior ^

[neon rapids]

Ah, seems cool but not sure it applies to how I have it set up.

I just have the one instance and we shut it down a lot to save on cloud costs.

Already updates automatically on boot.

[deep turret]

it's for multiple instance support

with these files you can launch any instance it will not update while other instances are running

yet it will update if it's the first starting instance

[neon rapids]

It's pretty cool though

[deep turret]

and you can update standalone

[neon rapids]

How does the game perform running multiple instances on one server?

[deep turret]

I would say it completly depends on the server

technically they dont interact with each others

[neon rapids]

I have the highest single-core performance server that Google Cloud offers and it's still maxed out at 100% all the time.

Just one instance too

[deep turret]

I think satisfactory mainloop always aim for 100% anyway

or you mean your whole cpu package ?

for 2 intances I would recommend starting at 6 threads at least Idk

[neon rapids]

Ah, interesting. I get glitches too though, or at least I did for U7. I get bad rubber-banding in U8 instead.

[neon rapids]

It doesn't utilize all the threads well, so just the one is constant 100%

[deep turret]

so they would share cpu time pretty well

[neon rapids]

I actually haven't played as much U8 because of the rubber-banding.

[deep turret]

it's on the RAM that it's a bit more expensive I think

[neon rapids]

16GB server

[deep turret]

you have to double literally

for 2 instances

[neon rapids]

That's not maxed either, I was at about 12GB last I looked

[deep turret]

anyway my units templates are ready for reliable multiple instances operations

with still the automatic update at startup on (which simply fallback if other instances are running)

I'm proud ngl

made my own qol feature

[neon rapids]

Very well done.

[deep turret]

fixed degraded state, now the fallback doesnt show up as failed

[deep turret]

note it supposed to trigger update even if all the instances are scheduled for automatic start, launching at the same time, with the server machine for instance, as they are all After=satisfactoryd-update.service

so it still do automatic update at the machine startup even when multiple instances automatically starts as well

(satisfactoryd.service not template is still set up as an instance, so everything applies to it as well)
it's exactly the same as [e-mail]

[calm stone]

I am sure this has been asked multiple times, how can I get the version of my dedicated server up to the Experimental Version, I have a version missmatch,
Game Ver: 239901
Server Ver: 211839

[deep turret]

depends if you are installing it from epic or steam (or steamcmd)

[calm stone]

steamcmd

[deep turret]

in your installation/update commandline, put -beta experimental instead of -beta public

[calm stone]

so: steamcmd +login anonymous +app_update 1690800 -beta experimental +quit

[deep turret]

wit the +force_install_dir if you want it but yes

just checked before telling you it's ok, sorry for delay

[calm stone]

sweet...thanks, I set the install dir already

another question any recomendations for specs, I have a virtual machine running the game and allocated 2 x 2.2GHz processors to the server with 16GB ram...do you suggest more cores? I usually only have 1 or 2 people on with me?

[deep turret]

satisfactory dedicated server is not very sensitive to raw power over multiple cores
tldr : more core doenst improve a lot

[calm stone]

Yea that was my understanding

[deep turret]

4 is good, up to 8 may help at very end game
but what matters is really single core performance

as long as you at least have 2 free one

[calm stone]

ok I think I will bump it to 8 then, do you suggest more then 16GB ram?

[deep turret]

so you better find the 2 best core you can have, even if you only have 4 total, because it is going to be way better for your game

16GB should be fine as long as you are not on enormous saves

for instance I'm testing someonelese save as a benchmark on my server :
it's a 2400hours save, it takes 25GB to run on u8 dedicated server

29MB save file

I would bet it's one of if not the biggest save out there

[calm stone]

nice, we are not that crazy, just me and my step son and little son

[deep turret]

the save takes so long to load it's scary

[calm stone]

we like starting fresh each big update

[deep turret]

anyway 16GB free you are fine for a long time

my 200h save is only taking up 7GB

[calm stone]

cool...I bumped the processor count to 4 with 2 threads each

Couldn't imageine a 29MB save file

that would be insane

[deep turret]

it's a save with a power storage of literally 72000 electric towers

[calm stone]

what specs are the server?

[deep turret]

it's all running undergournd

or was it 7200

[calm stone]

sorry what are the specs of your server?

[deep turret]

I dont remember but it was insane amount

[deep turret]

I didnt run the 29MB save on it yet, but it's a beaffy one :
selfhosted very lite vm with 4 threads, on a Ryzen 5900X host system

I recently increased the vm ram from 16GB to 32GB just to be able to run this save file

the game isnt even started, it's scary

mine is very just 5-7GB

take the fastest single core cpu you can afford even at low thread count

[calm stone]

dang

[deep turret]

dont need to go up 5900X though, even if .. yes you would notice the difference probably

I just so happen to have a big machine for my works

the game is 3600x recommended if I remember right

it likes fast cpus that's all you need

[calm stone]

I got an old Dell poweredge R730, 2xXeon E5-2630 v4 (20cpus), 64GB ram as the base unit. I have VMware runing on it with a virtual windows 10 box with 4 CPU's 2 threads each 16GB ram.

I have a couple other VM's on there one for Minecraft server and the other a DC.

it's not a 5900x speeds but my good machine is what I use to play on...haha

I think mine would die on your file

[deep turret]

it's half the speed of my cpu for that kind of workflow, so I dont know what to tell you, if you used to run the game on it, you will probaly notice better performance in u8 (experimental) dedicated server
but if you are new on it, don't expect very smooth experience

[loud minnow]

Why do you guys run it on vm instead of using something like docker… in vm you get performance hit.. I guess the main reason is just to run windows.. I mean there is other uses for vm but if you want all the performance is not really a good solution for game server hosting imo

[deep turret]

just know that it will not make a difference to put more core on it

it can even decrease performance in vm

[calm stone]

Yea...I run in a ESXi host not just a basic VM

[deep turret]

if you are comfortable enough running linux, I highly recommend running the dedicated server on linux

[calm stone]

I am TERRIBLE with linux

takes me like 3 hours to get my Minecraft server updated...haha

[deep turret]

also as you could see above in this channel, dedicated server on linux with steamcmd is very tested

[calm stone]

Personally, and I know this is going to generate alot of hate but I don't like Linux...it's just not for me.

[deep turret]

it's because mainly : linux runs way more efficiently than windows in a virtual machine

[calm stone]

power to you if you like it and can use it.

cd\

[deep turret]

windows like to run multiple things around, and that hits single thread performance in a vm

linux just dont run anything else until needed

[calm stone]

Yea I know the benifits of Linux and such

[loud minnow]

just wondering... as with docker you can do the same thing as a VM without the performace hit.. you can limit core/thread/cpu usage, ram, space, etc.. the only thing is you just can run windows... but if you like running windows then why not just install windows server or something like that to no get a performace hit by running linux then a vm to just run windos....

[calm stone]

but for me just takes me so much to do 1 thing

[deep turret]

if they want vms anyway, linux host may be a bit thinner than windows host

so on par for windows guests

[calm stone]

Mostly the performance hit you get with ESXi is negligible. Windows VM server sucks resources unless your good in no gui version, which I am not.

[loud minnow]

i run mine on 3950X and 64Gb of 3466mhz ram kit no performace issues.. using docker to limit the thread and ram usage for the server...

it was just a simple question why people like running VM

out of curiosity

[deep turret]

I think you can use docker on windows too ?

[calm stone]

Honest answer, I have never heard of Docker number 1, number 2 I am comfortable in ESXi

[deep turret]

wasnt there a thing about docker on windwos

[loud minnow]

yup you can use it on windows

[calm stone]

just like I am not comfortable in linux

also I run Multiple VM's on the server including 1 DC.

[deep turret]

if you are not comfortable and dont want to use something, it will not be more efficient for you for sure

it can take lot of time getting linux things a baseline for you, it really depends your needs

[calm stone]

well efficient is directly related to understanding/comfortability, the more uncomfortable I am with it the less efficient it is to get it setup imo

[loud minnow]

yeah same thing you can run lots of dokcer in one system but no performance hit... but yeah use what you want i dont judge lol

[calm stone]

all good

always great conversations

good to know what is out there

I will look into docker though.

[deep turret]

I dont run docker because I have to setup vm for my games anyway, and it's a lite one, also I can freely setup things as I want

[deep turret]

like all these things I wrote for linux multiple instance support 2h ago

[calm stone]

To be honest I wish I had the time to learn linux better, but I find it just very not user friendly...once it is setup it just works I know that.

It is not a resource hog

and is very secure

it has come a LONG way though

[deep turret]

it is not intermediate user friendly

but it is admin friendly
and end user friendly when the admin is

[loud minnow]

nope is VERY light weight... i used to use windows for hosting but i ended up learning linux this year actually and once you get the hang of it is pretty easy.. the main reason for me to move to linux was getting more UPS in factorio compare to windows 20 percent more just by OS change lol

[deep turret]

you can't really half know things to do complex things on linux, or you always run into weird complicated problems

[calm stone]

Yea and power to you for learning Ace. I agree SharklienX, and to me something simple is so time consuming

also like I said, if I take the time to figure out how to get it running for something like my minecraft server, then I don't touch it for months, then when it needs an update...Oh crap I have to relearn what I did.

if I had to use Linux more in my job then I probably would learn it faster.

daily use is a big thing with that.

[deep turret]

and you have to start very small, like just learning doing that systemctl start satisfactoryd-update with my setup I posted above, periodically

in a ssh

it's so many things to learn just to do a single thing

but it's all modular

once you got it, it's freedom

and it's always the same pattern or almost same

you can always fall into holes though because no one is god and there are lot of things

but that applies even more to windows

and at least on linux it's somewhat logical and not scam driven

[calm stone]

I just checked and my company has a plan with Udemy, maybe I will take the Linux for Beginners course.

[deep turret]

I think the other thing is : on linux you can actually try to learn it all, when on windows it is layered behind closed things, pro version things, special books things
so if you dont know how to limit what you want to learn NOW, you can get lost in a loop of learning other things but still being lost

and too many docs are for experienced users

so to get experienced first .. it can be a nightmare

[calm stone]

Yea I agree

[deep turret]

"if you get error with systemctl in your ssh, then just checks the respecting logs"
- BUT WHAT / WHERE ARE THESE F* LOGS ?????

it can look like that ^

HOW

and then you have to learn a new unrelated set of logs checking operations

etc..

[calm stone]

Yea Rabbit holes SUCK

[deep turret]

finally that's what is nice and worse with linux : it's a pretty f* big rabbit hole

but it also means it have almost anything you want in there

just have to figure out how to get it actually

[calm stone]

Great now my server isn't coming up...

[deep turret]

it can fail the first times after update

[calm stone]

I guess it did come up but still getting the same error

[deep turret]

steamcmd finished cleanly ?

you can add validate before +quit to be sure every thing is alright

I must go off though, I wish you getting it up nicely

I already pushed too late

[calm stone]

Ran it twice to be safe

and it said app 1690800 already up to date

ok thanks for the knowledge bud

[bold atlas]

I am having a strange issue where it says the server is offline yet is able to detect that the server doesn't have a name haha

I am guessing I just gotta reboot it and let it marinade a bit

hmmm still not working, strange, server was working fine pre update

[bleak flare]

anyone know how to fix this?

[spare orbit]

update the server?

[bleak flare]

I did online

using indifferent broccoli

and clicked update server and restart

[spare orbit]

its obviously not updated

[bleak flare]

i can see that

wondering if there another way to update it

since that way isnt working

[spare orbit]

i just use the intended way of steamcmd and it works

[bleak flare]

can u explain how to do that

or do yk video/link that shows how

[spare orbit]

https://satisfactory.wiki.gg/wiki/Dedicated_servers

[graceful flicker]

The game version is older than your server version.

The server has updated, you haven't.

[spare orbit]

client 239901 > server 239192

idk how to misread that xD

the server is older

[graceful flicker]

Oh sorry that is my mistake then

[spare orbit]

just the command line of steamcmd there

-beta public is u7, -beta experimental is u8

[bleak flare]

what folder is it in

in steamcmd

nvm fixed it gracias

[forest anchor]

anyone have an idea of why my server wont port forward in my pfsense firewall ? i went to nat > port forward and added port 7777 and the right ip for my server and its not working

[spare orbit]

also added the other two ports?

[forest anchor]

do i need thos now ?

[spare orbit]

yeah, these are to find the server

[forest anchor]

so i have port 7777 and 15777 and 15000 and its still not showing up

and when i do a port checker it shows its not open

[spare orbit]

well, then its closed

[forest anchor]

so idk what im doing wrong on my firewall cuz normally it worked but its just not letting me port forward

and here is my nat for port forwarding

[spare orbit]

:yuusha_shrug:

[graceful flicker]

Often times ports will only appear as open if something is actively using them.

[forest anchor]

hmm seems like it works now ty for the help

[spare orbit]

interesting case here.
except some rubber banding the server seems to be fine. except one thing. every player is allowed to join a world exactly once. a second time, no matter if clients closed or not always time out.
only fix is to load a save on the server.
so you can join without a problem, but when you quit or crash, you are gone.
can only join again after loading a save kekw

but thats mostly fine and not worse than before.
because before after crashing it was impossible to get the old character again, so it was needed to load a svae anyway and we are used to it.
now its just also when we quit.
but that just means to load the save that gets dome after quit before we join again.
still interesting problem

[hollow atlas]

Anyone know of some ways to make servers less "pop-in"?
There are some large builds on my server, but I don't see them until I get close, whereas I almost expect them to be on the horizon, and not only gradually load in as I get closer.

I don't mind if I have to load longer initially to load the entire map, but right now it's just jarring.

[simple flare]

Anyone else running into a crash upon mass deconstructing in a blueprint setup?

[clear lagoon]

you mean in the designer?

if you mean the dismantle blueprint feature, personally have not tried it yet

[simple flare]

I am talking about my server crashing whenever I deconstruct multiple pieces in short succession 😄

The console gets' spammed with missing audio entries prior to the crash.

[clear lagoon]

Are you running U7? it's a known issue

[simple flare]

Nah I'm running u8 😄

But yeah, if it's known, it's all good! 👍

[clear lagoon]

if you give us the crash stack we can take a look

[clear lagoon]

I'm only aware of a dismantle crash in U7 that is fixed in U8 personally

[simple flare]

which files you'd need?

The entire crash report folder zipped?

[clear lagoon]

the CrashContext.runtime-xml should contain the most important data

[simple flare]

Aight, will get you the file asap! :>

[full dew]

interesting

[simple flare]

Oh while we're on it @clear lagoon would you happen to know if NSSM allows for the console to be viewed?
I'm fairly sure I added -log and -unattended to the service I was creating, however, it never actually opens up a cmd window.

[broken orbit]

This is uncool

Both game and server updated. Both Epic clients restated. Both server and a client PCs restarted

[white elk]

hey where can i learn how to set up a dedicated server?

[broken orbit]

Have you read the wiki? the link is in the channel description at the top

[frosty gale]

aw man I give up lol

[simple flare]

looks like your dedicated server is running u8, while your client is using u7?

[simple flare]

https://satisfactory.wiki.gg/wiki/Dedicated_servers (The Pinned wiki is the old wiki, to my knowledge)

[white elk]

ty

[clear lagoon]

I'm sorry, I'm not familiar with windows tooling

[simple flare]

Aight, figured it'd be worth a shot! 😄

Who needs a console spamming log entries anyways, when your server just restarts automatically!

[white elk]

one more Q does that same method work for the epic version?

or is that different

[simple flare]

It's an entry in the wiki! :>
https://i.imgur.com/qVU1oO2.png

Steams' way is just propagating a hidden 'game', while Epic makes it easier accessible.

[white elk]

ah ok but is it a click and done thing or is there some tweaking the page is not detailed per say

i'll figure it out lol thanks for the help

[simple flare]

Aight

[potent terrace]

You wouldn't be able to read the output of NSSM as it'll run as a service under a Windows account

Although if you really want console output you could probably open Powershell and do something like this 'Get-Content %SatisfactoryServerDir%\FactoryGame\Saved\Logs\FactoryGame.log -Tail 25 -wait' considering you've enabled the console logging anyway

[deep turret]

you can also try u8 if you want, lot of these bug are fixed
we just are waiting a network bug fix so we can better connect to games

[spare orbit]

thats normal, because a server cant send you the whole map at the same time. only your near surrounding.
similar with stuff that was there before. like slugs will be there and disappear when getting close.
because your client will show the default map until the server sends the difference to that

[hollow atlas]

I'm aware of why it happens. I'm mostly curious if there's a way around it. Force the server to send the whole map at once. Or at least much earlier.

[spare orbit]

24/7 syncing a map of this size would take a ton of network and performance and is not that realistically. for a single person its dosble, because you know sending it once is enough because it wont change. when there are more players, stuff get complicated, because far away parts could change, so you resync everything.
except you send the map once to everyone, then only resync the surrounding of all players with everyone to cover changable area.
still adds a ton of potential complexity and probably isnt a good idea yet, since the stuff is single threaded and will most likely break fast with a few players.
though i think your client kinda saves added stuff?
like i can see factories we built forever from far away when i was there once. even when relogging i think. just stuff that existed on the map is visually there again and get removed when going close

since servers are in alpha state, i guess stuff like that can get better and more optimized in the future.
the time just hasnt come yet

[hollow atlas]

I'm not asking for 24/7, just once, when loading.
Purely so that you can see things in the background, or you get less pop-in when approaching.
Heck, even just having some clientside cache after relogging would be great. You say this was there before, but it's definitely not there now (nor do I personally remember this, but I could be wrong there).

But, sounds like such an option isn't there right now.

[deep turret]

it would basically streaming save file kind of
one down side would be high ram usage on both server and client
you'd have to make a lightweight version of all object that are just visual

when you are running single player it is actually here simulated

"lite way" Broken English 3000

smh

sorry for that

[torn finch]

The best he could do as a “fix” imo would be to run it on a VM completely within a RAM Disk if he had 64gb (probably a bit more) of RAM

The “issue” is the way the game loads/streams assets. They did change it from the old “tile” based system in u8 to something a bit more granular, but it’s still slower than u7 due to either the engine upgrade requirements to run, and experimental branch issues being fixed before EA release

I don’t really understand why everyone seems to use NSSM for windows dedicated server. I just use a .bat script along with task scheduler. Perhaps NSSM gives an easier to use interface? For me NSSM is nowhere near as stable, compact, or consistent compared to the built in OS tools.

[misty solstice]

@bleak flare you prob also need to switch to expermimental on your dedi ...

[deep turret]

what do you mean, we were talking about what informations are rendered, tile are either loaded or not, this also applies to foliage etc

[deep turret]

it doesnt run with the same environment from windows system behavior point of view, it doesnt receive the same signals when windwos shutdown etc..
I dont know what are the pros and cons for satisfactory dedicated servers but it certainly is not the same as just putting it in task scheduler, especially priority wise

[ocean terrace]

Can someone help me with this?
Whenever I open the server I created, it starts and then crashes.

Nevermind, it fixed itself on it's own ._.

[visual onyx]

got a dedicated server working for a few hours with people, is there anything major to know before committing to a full play through on a dedicated server?

[green pagoda]

Define high tickrate and save regulary xD

Seriously, we are 3 on server we need to put 400 tickrate minimum to have less crash (client or server side)

Vehicles or not much recommended (tractor, explorator, truck) too much risk of crash or bug for now
Train and drone seems ok for now
Hypertube is buggy but still okay since last update

[deep turret]

vehicle crash as been fixed last patch though mostly

[torn finch]

NSSM is basically an app wrapper for python code to load as a windows service. It runs in the same environment. It just boots on startup and does what it’s told with a nice interface. You can make your own services in windows without it is all I mean

To me it’s just excess overhead and not native so unneeded. For the average person it’s probably best as it does make everything in one spot and doesn’t require deeper understanding or learning of windows bs, it just does it for you. It’s a nice tool for sure, but I’m confused why you say it’s “different from a windows behavior point of view”? How so? It’s an executable, that starts a windows service. You can start your own windows service too…

Without NSSM*

[deep turret]

starting a software from task scheduler != windows service

[torn finch]

No it isn’t. It’s just part of the behavior I want/needed that and most are familiar with it

[deep turret]

then you can get that a service is not tight to same environnment connection, like when windows is down, when user log out, or other similar things, it is not 1:1 to any application that is casually launched

I got that it's probably useless for satisfactory though

[torn finch]

Yea? I guess it just doesn’t apply in my case as I have a dedicated headless server I run 24/7 just for satisfactory

Fair enough

I get what you meant now 👍

[deep turret]

yes, questionning using nssm for satisfactory then I have nothing to say, but questionning nssm as a whole thing because you can launch from task scheduler I dont get

[torn finch]

Never meant to claim the second. So I agree with ya.

But I realize now the reason for NSSM being suggested, can’t believe I didn’t before. It dies indeed make the process easier for the average person trying to run a game server. Running deficated satisfactory servers on win 11 arm through a VM isn’t exactly supported let alone common lmfao. I can be so dumb sometimes. Apologies for any negativity.

@deep turret

[lilac juniper]

Is there an API we can use to gather server information such as current users on the server, list of users that have been on the server, current Tier, perfomance metrics, etc.?

[green pagoda]

noice i will try as soon i can thx for the intel

[deep turret]

the only "api" currently existing is the game client's server manager's console

and is very limited

[deep turret]

#patch-notes message
• Temporary fix for audio log spam on Dedicated Servers when using vehicles which would lead into clients being kicked out of the Server
I believe it is mostly it

[fleet scarab]

Anyone else having trouble with not able to cut down trees in a self hosted server?

[lethal bridge]

anyone else got this problem with latest experimental? server console says -> Opening a new route from '[::ffff:46.167.26.216]:51658' to '[::1]:15000' but failed every time

[supple anchor]

disable ipv6 on your server

worked for me

[lethal bridge]

yeah i assumed that too

[supple anchor]

you can try using -multihome too

[lethal bridge]

what a bad solution but hey thanks 👍

ipv6 still useless nowadays xD

[supple anchor]

don't know why it's using ipv6 in the first place

but here we are

[torn finch]

Ue5 default as ipv6 was supposed to be fully implemented a decade ago and everyone still dragging their feet on that

Imo I’d give up on dedicated server for u8 till a big dedicated server patch. Anything else is an exercise in frustration.

[fleet scarab]

I cant cut down trees in server?

As soon as i upload my save to my server all the trees are back and i cant cut them down?

[torn finch]

Me and my friends use a startup script that checks a GitHub repo for a newer save and launches it after downloading if there is, and uploads the save on game exit. So long as we play together when online, it doesn’t have an issue with multiple different versions

And just use client hosting. Or the extra pc as a server/client. But not everyone has a second pc and account on hand

It’s not the best at all but it works for now

[green pagoda]

bug with sync, sometimes cutted trees reappear :/

It's a random issues

[deep turret]

it's better using ipv6 for network compatibility and new standards,
but the problem is it is bugging with it, and mixing things up

only in u8 though

u7 didnt have most of these problems

[fleet scarab]

I allready spend a day cutting down trees in the factrories when update 8 released on exp

[torn finch]

Is it the same LOD bug as everything else at play, where it doesn’t load the close LOD and things that are gone still show? Like the power shard slugs are my best example

[green pagoda]

Yep

[torn finch]

Either way, the devs did say that dedicated will be pretty broken for u8 for awhile

Like, they’ll start fixing it after u8 goes to live branch iirc

Dedicated server is just lower on the priority list sadly

[fleet scarab]

Im thinking about getting myself a second account to host a client on a low spec grapichs setting on a second pc

[green pagoda]

Understandable if not much dev to work on every side

[torn finch]

Which I hate. Because I see quite a performance boost when I run a local server on a different pc. Makes sense imo as it’s got the server work offloaded

And also is nice to have a persistent shared world with friends

That GitHub workaround I made is messy at best

[fleet scarab]

I want to keep de session online for friend yes

[torn finch]

Can always leave it running in the background staring at a wall

Lol

[fleet scarab]

But i dont want to keep my gamer rig running all the time.

[torn finch]

Why not? Is more harmful to a pc to turn it on and off daily than run 24/7

Though both are negligible so it didn’t matter aside electrics busy or heat

[fleet scarab]

I keep it on but i dont want to keep my graphics card maxed out all the time

[torn finch]

Set a low frame cap

And put everything on low starting at a wall in a corner

[fleet scarab]

I want to play on it as well

[torn finch]

Then when you want to play, turn settings back to normal

[fleet scarab]

And i can use a second account to send me items from main storage to a new base im building

[torn finch]

Idk, I’m spoiled by my m2 Mac mini running the server, as it’s 15w of power and heat are nothing compared to my gaming pc.

But for now the easiest way to keep a server running is to just have a spare account and pc…at least imo. That’s not feasible for most though, including me. I only have one gaming pc, I’m sure most do lol

There is/was some way to bug it out and have the “host” die, allowing 4 active players. Not sure if it still works in u8. I could never make it work

[fleet scarab]

Im back to work on my new plastic factory.
Thanks for the input

[visual onyx]

what is the default tickrate and how do I change it? Also thank you you've given me some nice clarity

[spare orbit]

default is 30
on u7 we took it to 60, helped some visuals, nothing more.
u8 we reverted this because everything else than 30 was just unstable as hell and while trying to reach 60 the server dropped to 15

[chrome pumice]

I have been running on 60 ticks just fine. It only lowers when somebody connects then go back up.

[visual onyx]

Is there some metric where I can track the tickrate or do you just visually notice

[chrome pumice]

I believe it shows on the server manager. I am not able to check this right now. I was able to notice items on the belts almost never disappear or stop moving with higher tick rate.

[spare orbit]

yeah, server says it ingame
and yes, thats the only positive about higher tickrate we ever saw. stuff on belts getting less stuck

but we cant get it stable on u8

but default works better than u7 default tbh

[lapis frost]

Howdy kind folks! A friend of mine was running a dedicated server and it started acting funny when I tried to connect. I connect past loading screens, and then satisfactory crashes after showing a single frame of ingame

I tried setting the server up on a machine I had and got to that point and now can't connect to the server at all.

The console has a lot of mention of failing to initialize the SteamAPI

And now I am getting the below error

[elfin oyster]

There have been some issues with how the client is handling ipv6. There have been a few different ways to deal with this posted over the last week or so.

One was to specify the ipv4 as if you were running a multi tenant setup another was to disable packetracjng.

Your call on which one you are more comfortable with 🙂

Sadly I have stopped tinkering with U8 experimental dedicated servers as I couldn’t work out how to use advanced features without transplanting a local save file and session.

Does anyone know if these options can be toggled through the console shell?

[marble berry]

have you tried loading an auto-save and check the box when doing so?

[elfin oyster]

Hmm. I’ll give it a go.

I did try loading a local save through the server manager, but that only applied to the single session I set up and didn’t persist

[lapis frost]

Can you tell me more about specifying to use ipv4?

[forest anchor]

Anyone know if you can still enable multi core or not cuz one of my server has a slow cpu so im wondering if multi core would help

[desert walrus]

The server will use as many cores as it can
Multi core is always on

[forest anchor]

I mean i have 24 cores buts only used 1

[sharp bison]

add multihome=0.0.0.0 to the startup arguments in the service file

[deep turret]

if you load your server save with the option on it will be on
then you have to save that as well to persist

either auto-save disconnect-save or manual-save
as long as you saved your game in advanced mode

[deep turret]

there are some problems with these kind of arguments in u8 though

waiting for a patch if it ever come (dedicated server isn't priority)

[spare orbit]

actually no problems to that.
just that ue4 defaulted to ipv4 while ue5 wants ipv6 now.
disabling ipv6 on the machine lets it behave as before

because there is no ipv6 and it uses the old defaults

would also work if you set up ipv6 routing properly. but many servers and firewalls cant do that

[deep turret]

getting synced secure ip with ipv6 is harder than with ipv4

and authorities have tried to artifically put limits on ipv6 forwarding, which means it is less tool supported

(btw, fundamentally, connection establishment is by local forwarding, even though nobody put that officially yet)

[spare orbit]

i get this

then this

everythings correct.
but client somehow errors now

in loading sceen

wasnt before, changed nothing

[spare orbit]

hm, it somehow corrupte the save being unable to load it. using another one worked

[near rapids]

You can try to load corrupted save in to single player, see if can connect normally and move back to server if works

[deep turret]

127.0.0.1 is localhost, it cant be routed

it happens the same with ipv6

network is broken

[spare orbit]

localhost is correct

the server needs to bind to the local network address

[deep turret]

you mean you are running server on your machine along with the game client ?

[spare orbit]

this is the route from external to localhost. because the server obviously binds to the local network adapter of the system its installed on

its my private ip to the servers local ip

[deep turret]

localhost isnt the server local ip

[spare orbit]

it looks this way when a server is behind a firewall

[deep turret]

nope

[spare orbit]

it does here always. because the server cant know its outside ip

obviously

[deep turret]

local ip isnt localhost

you have both public ip, local ip AND localhost

that's the loopback network stack interface

[spare orbit]

right.
127.0.0.1 is the local ip
localhost is the localhost

[deep turret]

nope, 127.0.0.1 IS localhost
localhost name records 127.0.0.1 and ::1

[spare orbit]

no, a server behind a firewall only has its private ip

[deep turret]

127.0.0.1 is not a private ip

it's loopback ip

every machine have 127.0.0.1

even the router

[spare orbit]

yeah, but thats where you bind stuff on

[deep turret]

you bind on any local ip

[spare orbit]

it has to be an internal ip

[deep turret]

or even sometimes on external ip if your network stack allow it

[spare orbit]

and there usually is just one. thats 127.0.0.1. except you made additional virtual ones

[deep turret]

that's why you have binds to 0.0.0.0: and [::]:
that are placeholder ips, basically : you bind to every ip

[deep turret]

nope, also 127.0.0.1 is technically a virtual one as it is software loopback

[spare orbit]

yeah, but binding to 0.0.0.0 simply results in binding to 127.0.0.1 except you made additional internal ips

[deep turret]

you dont make any ip by binding anything

[spare orbit]

0 is to all. normally there only is this one

[deep turret]

ips are setup there already

[spare orbit]

yeah, you can bind only to existing stuff and the literal only existing efault one is 127.0.0.1

[deep turret]

0.0.0.0 means all the local ip, and loopback is one of them, a special one

you cant communicate with any other machine only running on 127.0.0.1

that's why it's called localhost

[spare orbit]

thats wrong

you cant send with it, but every input gets there

thats why the server makes a route from it

because when you arrive at a server, you arrive there

[deep turret]

no input dont get there

inputs get on local ip interface, the normal ones

[spare orbit]

so you mean the server im literally playing on doesnt work actually?

[deep turret]

like 192.168.1.56 for instance

[spare orbit]

no, you ask for that, but thats not where you arrive

you connect to it with this ip, but the server gets your message at the internal ip

[deep turret]

no firewall on earth would behave correctly if any packet arriving was then injected into loopback interface

[spare orbit]

actually its the opposite.
nothing could work wen you bind everything irectly too your door outside instead of having it internal

[deep turret]

the server gets the message as it arrive on socket it binds for, which is any local address the machine have if they binded to 0.0.0.0

[spare orbit]

0.0.0.0 defaults to 127.0.0.1

[deep turret]

0.0.0.0 defaults to any

[spare orbit]

any internal
and only internal is 127.0.0.1

it doesnt bind to external

[deep turret]

any LOCAL

[spare orbit]

both are local, the term is useless

[deep turret]

exactly both are local 0.0.0.0 binds to any local

[spare orbit]

it is critical to not bind directly to outside

[deep turret]

and 127.0.0.1 is one of THEM

[spare orbit]

no, any internal, not any local

[deep turret]

man, you are trolling me I cant

[spare orbit]

as said, security doesnt exist when binding directly to outside connected

[deep turret]

why do you think 127.0.0.1 is an IP ADDRESS and why do you think we talk about loopback INTERFACE, the terms are there, its comming from "outside"

127.0.0.1 is just a virtual outside specific to the host

[spare orbit]

so again, you say it cant work while it works and we literally do it that way in a large company because it is literally forced by law

[deep turret]

if you do it that way in companies it's called a security hole

[spare orbit]

you are not allowed to bindd any software directly to the outside ip

[spare orbit]

what you do is bypassing every security

[deep turret]

because localhost have more access to the network than outside coming trafic

[spare orbit]

ok, i end this here.
you have clearly no idea about the topic

[deep turret]

you are still confusing local ips and loopback ips, the host doesnt even HAVE public ips, it could even not know about it

local ip are there

• for ipv4

because we are talking ipv4

[spare orbit]

simple question to disqualify you.
give an example for an internal and an external ip

[deep turret]

on ipv6 you can have public-local ips

[deep turret]

I already gave it to you
192.168.1.54 is local ip
127.0.0.1 is loopback (internal) local ip

[spare orbit]

that was not the question

dont dodge it

[deep turret]

I didnt

[spare orbit]

i said internal and external.
you just marked stuff as local (both, so remove that word) and internal

[deep turret]

that are not standard defined terms but lets is it then:
192.168.1.54 is local(internal and external) ip
127.0.0.1 is loopback (internal) local ip

[spare orbit]

internal and external
saying this is like self disqualification

it cant ever be both

[deep turret]

you are then confusing what communication is

only external would be routing

if you get the message at some point and read it, it's going inside your communication pipeline

local ip can be routed on loopback when they are used as such

and then it's internal behavior

like loopback 127.0.0.1

[spare orbit]

a pc has an external ip were you can contact it, this ip is solely for communication with the outside.
so you contact this ip, the pc gets it, and reroutes you to a specific port of the internal ip, on this rerouting, stuff like firewalls etc happen and can decide to either block or let you go to the internal one.
then you arrive at the internal ip and here you are allowed to talk to something thats waiting at this specific port. this place is where you bind software to.
when you dont do this and rather bind the software directly to the outgoing ip, no internal rerouting and packet filtering can happen, so you are completely unprotected. this is in fact illegal here.

when you ever configured a hardware firewall you would know, that you always reroute the connection from the outgoing ip to the internal one and eveything inside is connected to this one and never to the outgoing

and the ip you said like 192.168.1.54 is the external one. binding to this bypasses every protection this pc could have. its half okay when you still have an external firewall before that, but its still oof

[deep turret]

green machine doesnt do routing because it's a user machine

red machine is a router

[spare orbit]

yeah, thats the disqualification for thinking a user machine and a router are technically different

[deep turret]

did I say they are technically different ?

one is doing the cyan thing, the other isnt

that's what qualify a machine for being a router or not

[clear lagoon]

That explanation is so convoluted and incorrect that I don't even know where to begin.
A computer has some special loopback addresses that always send network packets to itself, i.e. 127.0.0.1/32 or ::1/128
On top of that, the network interface cards (NICs) of the computer can bind to one or more IPv4 and/or IPv6 addresses.
Due to the scarcity of IPv4 addresses, new technology was introduced called Network Address Translation (NAT) that can translate one single publicly assigned IPv4 addresses to multiple internal network device address ranges, kind of like internal phone numbers within a company PBX.
These ranges are 10.0.0.0-10.255.255.255, 172.16.0.0-172.31.255.255, and most commonly in home networks: 192.168.0.0-192.168.255.255.
When not connecting to the internet by means of NAT your device can have a publicly valid IP address that is directly reachable.

Any packets that arrive at the IP Address of the computer will be handled by the OS Kernel as appropriate, usually filtered through firewall rules as configured. There is no way to skip past this software firewall by using some sort of magic internal IP address.
Because of the nature of NAT, it needs to translate incoming packet data to the correct destination device in its internal network. Because new sockets initiated by an external party only connect to the publicly known IP Address and a port, the NAT device can only establish the connection if this port is configured in a Port Forwarding table. This coincidentally makes the NAT device effectively function as a firewall for new inbound sockets.

So NAT is just a work-around to make a single IPv4 address service multiple computers, and is not how IP networking in general is intended to function.

With IPv6, there are again enough addresses to directly assign to all individual devices. This leads to the effect that a router will perform its original function again: your router becomes responsible to route packets for the address space assigned to your router and all devices behind it. This means there is no distinction between public and private addresses. Your router can send any packets destined for a specific address directly, without having to "translate" addresses to internally assigned ranges.
This does not prevent your router from still performing firewall functions. It is still responsible for routing all packets, can read their headers just fine, and can drop any packets as it so wishes. If your router does IPv6 addressing and is in any way worth its salt, it should still be performing firewall functions just as incidentally happens for inbound sockets on IPv4 with NAT.

[spare orbit]

congrats to completely dodging the topic and explaining something that was never even mentioned

also a skill, i guess

[clear lagoon]

I was responding to your explanation of external and internal IP addresses and how it makes no sense how you explained it, and you state that packets sent to a public ip address are entirely unfiltered

[spare orbit]

they are

[clear lagoon]

I responded with an explanation that provides why they are not

[spare orbit]

and you talking about being a filter in front of it

[deep turret]

I'm sorry my maniac side showed up, there is only a very tiny bit off : it's 127.0.0.1/8 instead of 127.0.0.1/32 even though 127.0.0.1 is the one mostly used

[clear lagoon]

there is always a filter between the kernel and user space

[spare orbit]

well yeah, but thats not really useful actually

[deep turret]

binding is basically telling the kernel "I want to know about that so tell me" the kernel is managing this because there are so many pieces of software running on a single machine

[clear lagoon]

... are you calling kernel firewalls not useful?

[spare orbit]

well, they do stuff, but when someone actually tries, they hold nothing.
they only prevent the most stupid attacks

[clear lagoon]

hardware firewalls are just running an OS that still does firewall in software you know. it's all the same devices. computers, routers, switches, same difference

[clear lagoon]

firewall rules are the most important part of a good network setup. heuristics firewalls are a far more advanced topic

intrusion detection is not what I am trying to talk about here

[deep turret]

"There is no way to skip past this software firewall by using some sort of magic internal IP address." unless you manually add rules that punch that hole into it, but you dont want to do that as I said already
ps: not against you, I'm just linking my messages around yours, just going a bit further into it

[spare orbit]

but anyways. about the actual topic.
you may try to bind the server to like -multihome 10.0.0.24
it will still bind to only 127.0.0.1 because it is supposed to do that

its impossible for the engine to bind to the external one

doesnt matter what ipv4 address you give it, it will always bind there

[clear lagoon]

? what are you talking about

[spare orbit]

just try it, you wont be able to. because that would be exactly the hole mentioned here

[spare orbit]

the engine will always refuse to bind directly to the ip if its the external ip of the system.
probably because thats disallowed by many countries laws

[clear lagoon]

then how would it get network packets?

[spare orbit]

i explained that

[deep turret]

"when you dont do this and rather bind the software directly to the outgoing ip, no internal rerouting and packet filtering can happen, so you are completely unprotected. this is in fact illegal here.

when you ever configured a hardware firewall you would know, that you always reroute the connection from the outgoing ip to the internal one and eveything inside is connected to this one and never to the outgoing"
are you trolling ?

software binds to any interface they want as long as kernel allow them to, and then the packet isnt "rerouted"

[spare orbit]

im asking myself the same about you for quite some time.
but im kinda sure you just really dont know

[clear lagoon]

okay so the server binds to 127.0.0.1, which means it will only receive packets from the kernel that were destined for that IP address specifically. The router, at 192.168.0.1 dispatches packets for the game to my computer, at 192.168.0.107. How does the kernel that receives the packets end up in the userspace of the server only bound on 127.0.0.1?

[deep turret]

though you can reroute it by special rules but that would be stupid

[spare orbit]

explained that

[spare orbit]

here

[clear lagoon]

where? I can't find specifically where you make the jump from 192.168.0.107 to the userspace process

[deep turret]

you literally said "rerouted"

[spare orbit]

because English has no precise word for that as far as i know

need to pick stuff thats close to the meaning. sadly the language is really limited

[deep turret]

I think it have but you want to put a convoluted thing inbetween for no apparent reason

[spare orbit]

as i said, try to start the server with binding ot to the external
that wont happen

[clear lagoon]

your userspace process is bound to 127.0.0.1 so the kernel will find no userspace processes to deliver the packets on the server port for that are destined to 192.168.0.107. The packets are dropped. your connection would fail

[clear lagoon]

literally how I am running my server, but sure 😛

[spare orbit]

so our whole working infrastructure doesnt work.
interesting

[deep turret]

Nachtnebel are you running windows or linux ?

[spare orbit]

give me the shell output

[deep turret]

so I can directly give you command to gather listenning

[clear lagoon]

you obviously see I have a missing link in my explanation, but you can't explain what the missing link is?

[spare orbit]

it will say 127.0.0.1

[spare orbit]

depends what you exactly want.
have everything with few windows, few linux, some bsd, etc

im not sure what you are asking for

[clear lagoon]

UDP bindings as follows:

85.10.195.11:7777
85.10.195.11:15000
85.10.195.11:15777

[spare orbit]

nono.
what does the server say. not the pc

so the shell output of the software

not to what the pc translates it

[clear lagoon]

[2023.06.22-22.14.19:074][596]LogNet: Created socket for bind address: 85.10.195.11:15000

[spare orbit]

no, the routing adress

not the socket

connect with a client and tell the route

[deep turret]

man

socket is kernel <-> in the software already

[spare orbit]

just do.
it shows the other address then

because thats the actual route

[clear lagoon]

I swear if this entire tangent is because UE5 has some stupid internal packet router that binds to the public ports and just retransmits packets to localhost, this is going to have been such a waste of time because both are part of the same software in userspace

apparently I have to install an update. give me a min

[deep turret]

whatever do UE5 that may be stupid does not mean it cant read packets directly

it's just the software doing what it wants to

and I was saying it is not supposed to do that

[wide elm]

Well, the latest video on dedicated servers ( https://youtu.be/sLPX6TovVNU ) explains, I guess, why I experience such a high number of crashes on my pc when I'm playing on my own server. It's like experimental * experimental... which is a shame because sometimes it completely makes it impossible for me to continue playing (3 UE5 crashes in a row within minutes of restarting the game is no fun). Anyway, check out their latest update (the one about them going on a well-deserver holiday for the month of July) on YouTube for a bit more info. This is not meant as a sneer by the way, they do very much deserve some time-off. I might just revert to release 7 for my multiplayer sessions. This is early release and experimental at that.

[spare orbit]

like you get this for whatever you told them via -multihome

and when a client connects, it gets routed to 127.0.0.1 nonetheless

i blurred my private ip for obvious reasons

thats my real external ip

[deep turret]

that's UE5/dedicated serve internal, it is not machine networking

[spare orbit]

an it will always get route there. its impossible to prevent that

[clear lagoon]

[2023.06.23-12.31.34:413][212]LogBeacon: OnlineBeaconHost_2147401330[IpConnection_2147401058]: Beacon Join FGServerBeaconClient Steam:(Steam)76561198057327031 (unauthenticated)
[2023.06.23-12.31.34:772][223]LogBeacon: OnlineBeaconHost_2147401330[IpConnection_2147401058]: Handshake complete.
[2023.06.23-12.31.36:508][275]LogNet: UChannel::ReceivedSequencedBunch: Bunch.bClose == true. ChIndex == 0. Calling ConditionalCleanUp.
[2023.06.23-12.31.36:508][275]LogNet: UChannel::CleanUp: ChIndex == 0. Closing connection. [UChannel] ChIndex: 0, Closing: 0 [UNetConnection] RemoteAddr: 84.193.155.129:50756, Name: IpConnection_2147401058, Driver: IpNetDriver_2147401328 IpNetDriver_2147401328, IsServer: YES, PC: NULL, Owner: FGServerBeaconClient_2147401046, UniqueId: Steam:(Steam)76561198057327031
[2023.06.23-12.31.36:508][275]LogNet: UNetConnection::Close: [UNetConnection] RemoteAddr: 84.193.155.129:50756, Name: IpConnection_2147401058, Driver: IpNetDriver_2147401328 IpNetDriver_2147401328, IsServer: YES, PC: NULL, Owner: FGServerBeaconClient_2147401046, UniqueId: Steam:(Steam)76561198057327031, Channels: 9, Time: 2023.06.23-12.31.36
[2023.06.23-12.31.36:508][275]LogNet: UNetConnection::SendCloseReason:
[2023.06.23-12.31.36:508][275]LogNet:  - Result=ControlChannelClose, ErrorContext="ControlChannelClose"
[2023.06.23-12.31.36:508][275]LogNet: UChannel::Close: Sending CloseBunch. ChIndex == 0. Name: [UChannel] ChIndex: 0, Closing: 0 [UNetConnection] RemoteAddr: 84.193.155.129:50756, Name: IpConnection_2147401058, Driver: IpNetDriver_2147401328 IpNetDriver_2147401328, IsServer: YES, PC: NULL, Owner: FGServerBeaconClient_2147401046, UniqueId: Steam:(Steam)76561198057327031
[2023.06.23-12.31.37:544][306]LogNet: NotifyAcceptingConnection accepted from: 84.193.155.129:40624
[2023.06.23-12.31.37:610][308]LogNet: NotifyAcceptingConnection accepted from: 84.193.155.129:40624
[2023.06.23-12.31.37:611][308]LogNet: Server accepting post-challenge connection from: 84.193.155.129:40624
[2023.06.23-12.31.37:611][308]LogNet: IpConnection_2147401022 setting maximum channels to: 32767
[2023.06.23-12.31.37:611][308]PacketHandlerLog: Loaded PacketHandler component: DTLSHandlerComponent ()
[2023.06.23-12.31.37:611][308]PacketHandlerLog: Loaded PacketHandler component: Engine.EngineHandlerComponentFactory (StatelessConnectHandlerComponent)
[2023.06.23-12.31.37:611][308]LogNet: NotifyAcceptedConnection: Name: Persistent_Level, TimeStamp: 06/23/23 12:31:37, [UNetConnection] RemoteAddr: 84.193.155.129:40624, Name: IpConnection_2147401022, Driver: GameNetDriver NetDriverEOSBase_2147481943, IsServer: YES, PC: NULL, Owner: NULL, UniqueId: INVALID
LogGenericPlatformMisc: Error: SetEnvironmentVar not implemented for this platform: SteamAppId = 526870
[2023.06.23-12.31.37:611][308]LogNet: AddClientConnection: Added client connection: [UNetConnection] RemoteAddr: 84.193.155.129:40624, Name: IpConnection_21474dlopen failed trying to load:
steamclient.so
with error:
steamclient.so: cannot open shared object file: No such file or directory
[S_API] SteamAPI_Init(): Loaded '/home/sparxy/.steam/sdk64/steamclient.so' OK.  (First tried local 'steamclient.so')

no 127.0.0.1 in my logs. just my home ipv4

happy now?

[spare orbit]

yeah, because you literally copied the part right after that what i sent

nice try

[deep turret]

note : we are kind of investigating bug, you dont investigate bugs if you consider any behavior as a feature

[spare orbit]

thats not a bug because its a literal law

[deep turret]

you made it so, when it is only in UE5 software here

[spare orbit]

i already said, that almost every existing software behaves that way

most time it just doesnt tell you that it does

[clear lagoon]

there. the start of the connection handshake

still no 127.0.0.1

[spare orbit]

thats way before handshake

its right below what you sent before

that creating a socket

[clear lagoon]

I'll just upload the entire log file and redact my login ticket

1 sec

[spare orbit]

like this close to the message

[clear lagoon]

here's the full server log. point me at it

[spare orbit]

isnt that quite a bit short?

like its few thousand lines here

[clear lagoon]

that file is 3863 lines

[spare orbit]

ah, display full file doesnt mean full for discord ....

so you really are allowed to commit a security breach. okay

[clear lagoon]

from your screenshots all I can assume that there is some piece of software between the unrealserver binding and your public ip that I'm not aware of

there is no security breach

go at it

[spare orbit]

yeah obvious, because direct bindin is illegal here

as already said

but it should be the engine not allowing it. because its nothing you do manually

[clear lagoon]

so what software is it you're running that indirects the network connection?

it's probably something to do with windows defender or firewall since your screenshots look windowsy

[spare orbit]

there is none.
maybe the engine tells what county you are and has different defaults?

[clear lagoon]

what country you in, if you don't mind me asking?