dedicated-servers

[quiet cairn]

Does your router support "NAT loopback" because if it doesn't your gonna need to use your local IP instead

[frosty gale]

It's not locally. It's running on GCP

[quiet cairn]

The port is confirmed open by my testing.

[frosty gale]

But still.. Can't connect 😦

[quiet cairn]

The last part of that error message is concerning.

No, Null, Null

Reach out to host

Give them that error message

[frosty gale]

I don't think Google Support can help me here 🙂

[fossil stump]

Anyone had unexplainable train collisions?

I've had a few recently where trains that should not have been able to collide because of signalling, have collided and de-railed leading to my supply chains failing...

[desert walrus]

You're hosting with google?

[desert walrus]

Probably improperly configured signals

[fossil stump]

I assure you it's not

two trains coming into one junction with path signals on both legs, there should be no way two trains can enter the junction at the same time.

and I have watched it work hundreds of times, then suddenly it doesnt.

[desert walrus]

Try block signals instead, see if that fixes it

@fossil stump

[fossil stump]

I have literally sat and watched the junction work fine hundreds of time with path signals, this is not a case of incorrect signalling, this is the signalling suddenly failing to work at all for some reason.

replacing it with block signals doesn't prove anything if the failure is intermittent.

[desert walrus]

Can I see a screenshot of the junction with the signal overlay?

[fossil stump]

it. is. not. a. problem. with. the. junction. It has worked hundreds of times with the same signalling and hundreds of train journeys through it. It's a random failure of the singalling.

[desert walrus]

I don't know why you'd bother posting a question if you refuse to even consider any answer other than "the games broken"

[fossil stump]

because I'm trying to figure out if it's something to do with my server specifically or the game

[frosty gale]

Yep

[loud minnow]

Hahaha is probably just your set up lol… because I can’t see how trains signals not working can effect the server in ANY way lol and if is not working then just fix it

[fossil stump]

yes... because that would explain why it works hundreds of times... and then randomly doesn't.

I have literally sat there and watched the trains go through the junction without issue, then randomly when I have been elsewhere the junction has failed. It's almost as though like... there might be a bug with the server... you know.. it being "experimental" and all.

[blazing halo]

Here are two explanations coming to my mind:

a) the trains are bugging out due to some random lag spikes (much like the whole 'passing through solid objects' with the cannons), or more likely
b) there's a tiny, tiny timing issue somewhere, and it works properly 99.9% of the time... but eventually fails

How do you even get collisions? Using a common track? Looping around?

[fossil stump]

two tracks combining into one.

[blazing halo]

Huh... Do the signals work like a block signal?

(the proper term currently eludes me, sadly,

but it's a system where the signalling detects that there's a train on a particular block,

and it locks out any other oncoming traffic)

[fossil stump]

path signals allow trains to reserve a path through a junction, the system should not allow two trains into the junction at the same time

[blazing halo]

Does it check if the block right after it is free?

[fossil stump]

yes

[blazing halo]

Hmm...

Then no clue.

[fossil stump]

exactly, it HAS to come down to the game fucking up, but whether that's something that just randomly happens or a performance issue on my server is the question, and that's why I asked here.

[blazing halo]

What you can do is sit your ass down at the junction, record your screen, and leave it there until it happens again.

You should also record the perf data for the server, to correlate.

[hot flame]

Unfortunately, we cant tell you that without your server logs and exact times when it happens and performance data of your computer and network at those exact moments. Nor can we tell you if its a signaling issue or not because you refuse to even entertain the possibility. My suggestion would be the same as Kad's.

[blazing halo]

Are you able to connect to any Internet Satisfactory game? Having logs from the server and client might shed some light on this.

[fossil stump]

or you know... if it was happening to other people... that would give some indication that it's NOT specific to my setup, that's why I asked...

[hot flame]

why are you getting mad at me? I am just stating what I would recommend. Do it, or don't. You may have better reach on the Satisfactory Reddit

[blazing halo]

As far as I've been monitoring this channel (and I tend to scroll up), I haven't seen this particular issue raised.

And I've been here before U6 (I joined sometime late April or so?).

[fossil stump]

because it's simple logic, if it was a signalling issue, it would happen repeatedly and frequently, not randomly. It's also a very VERY simple junction, two tracks into one, with Path signals on both inputs and the output, there's no way if the signals are working there should ever be two trains in the junction at the same time.

[hot flame]

So anyway, down to why I am here: I just set up a dedicated server for me and a friend. We are only at tier 2, and after running for a few hours, the server is pulling 6.5GB of memory. Is the memory usage pretty linear or should I be worried that it's using so much so soon?

[blazing halo]

That looks about right for a U5 DS.

Also, I wouldn't worry about memory consumption. You're going to run into save time limitations long before RAM usage becomes an issue. The only save I've seen that has significant memory consumption (to the tune of 26 GiB by the process) takes about 3 minutes or so to save completely. I wouldn't recommend loading it on the server.

[hot flame]

Yeah, I read your reply to the other guy. Hence my reply stating that you wouldn't entertain the fact. What I have seen suggested on Reddit for trains that randomly crash is to ensure that no 2 trains are ever going different ways on the same track. Personally I have had better luck using Block signals than Path signals because block signals seem to ALWAYS work, and Path signals seem to only work when they want to. I had a 10 train setup my last save and didn't need to use a single Path signal. If it is a simple setup with "simple logic" as you say, you may want to drop the path signals and just use block signals so it never happens again.

[blazing halo]

You both realise that without any further data, y'all can argue the point 'till the cows come home, right?

[hot flame]

Just offering suggestions I have seen work.

I 100% agree with you Kad, which is why I am not arguing, just offering help

[blazing halo]

You're offering mitigations, which might or might not work. The only way to properly determine and resolve this is to record the perf data and go from there. Being able to see it happening would also be valuable.

So yeah.

[hot flame]

Well, the problem is, at 6.5GB of memory usage, that is 70% of my memory being used (including windows processes) and the game is running as a service on an otherwise idle machine at tier 2. That is why I was worrying.

[blazing halo]

That's the minimum amount of RAM it's going to consume.

The maximum (well, the reasonable maximum) is somewhere around 8 GB, I think.

[hot flame]

well right now its consuming ~2.2MB (not sure why the memory was so high earlier)

[blazing halo]

I don't honestly know, because I haven't seen big server-side saves yet.

My own consumes about 7.6 GiB of resident memory.

[hot flame]

what tier?

[blazing halo]

Food Court.

But it doesn't matter. It scales with structures built, not tier.

[hot flame]

yeah but tier gives me a ballpark

[blazing halo]

It really doesn't.

For most core players, the game begins once everything is unlocked.

[hot flame]

Yes, it does. If you tell me "I just finished tier 8" - I know you at least have nuclear setup. If you are still on tier 5, you probably don't have a huge or small turbofuel farm or plastic factories going yet.

[blazing halo]

Not necessarily. Turbofuel works, nuclear is complicated. You can push towards endgame with turbofuel perfectly fine.

But like I said, it doesn't matter.

Memory use scales with structures built.

[hot flame]

I have a general understanding of how many buildings each tier needs, I can use that general understanding to guess how big (or small) your build is, giving me a ballpark.

[blazing halo]

That's the thing. You can't.

By perfectly mundane means, you can get to end of phase 4 within 2 active days, incl. the time needed to produce everything.

[hot flame]

So if I say I am at tier 2 - how many buildings do you think I have? Hundreds? Thousands? Hundreds of Thousands?

[blazing halo]

I have no clue.

If you decided to build iron production out as far as you possibly can, you can do that.

Resources are unlimited.

And even if you assume minimal builds and full conservation of resources, geared towards optimal resource consumption per unit of time,

once the game reaches phase 4, all bets are off.

The game really gives you a lot of freedom, in the end. There are goals you can go for, but you don't have to do them right away.

[hot flame]

yes, and if I decided NOT to build iron production out as far as I possibly could and only build what I needed, this gives you a RANGE of how many buildings I could have. Then you use other context clues (such as he said he only setup the server this morning) to further narrow down the possibilities. Using what you know, you can reasonably guess, or as I stated earlier, ballpark.

you COULD be completely wrong, you are right.

[blazing halo]

That's true, but the way I see it - that range could stretch between your low ballpark and potentially infinity, is what I'm saying. But sure.

At any rate, including foundations and power lines, I currently have something like 500 000 structures built. My game consumes 7.6 GiB on the u5 server, and is about 3 MiB as a save file. That should give you some context.

It takes about 13s to save it on an i7-4770.

[hot flame]

but that isnt the point of "ballparking" figures. If I said I was going to the grocery store to buy hotdogs. I could potentially buy $1 million dollars of hot dogs, or I could buy a single pack. You can reasonably guess I am not going to spend $1 million on hotdogs and reasonably guess that I am going to spend less than $10 dollars.

[blazing halo]

I can, assuming rational spending and reasonable needs. In the game, building is essentially free (constrained only by time), and needs are irrelevant - you build because you can.

[hot flame]

thanks, that is about what I was thinking. Thats not bad

[blazing halo]

The game can take quite a bit, to be fair. That was sort of the point.

[hot flame]

Can I have built out all of the iron production at tier 2 to max in a matter of a few hours (remember I setup the server this morning)

[blazing halo]

I have no clue. You having set the server up just this morning tells me nothing about your prior experience with the game - only prior experience with the dedicated server. Further: the game uses the same saves on both server and client, so you could've loaded a save you've been working on locally up until now. ;]

[hot flame]

again, reasonable guess

[blazing halo]

I prefer to ask rather than assume.

What I'm trying to say is that the ballpark you may come up with may be wildly inaccurate, you know?

Under any other circumstances, I would actually agree with you entirely.

But the game was built with millions of structures in mind.

The ballpark is larger than you think it is.

But it's also a bit of a moot point, because while the RAM use scales with structures, the increment is relatively tiny.

[hot flame]

I did ask, you refused to answer on the basis that it doesn't matter. Yes, I know that estimates can be wildly inaccurate. Using this knowledge that estimates can be wildly inaccurate means I am not going to go out and spend $300 on an additional 16gb of ram just because some saves hit 20+gb of memory usage.

[blazing halo]

This game isn't like ARK, where you get a 1 GB save file after living on the Island for a day. :)

[hot flame]

No, but like ARK it is memory hungry, and like ARK when I played it, it is still in EA and not optimized meaning memory usage can become an issue (it hasn't, but its not out of the realm of possibility either).

[blazing halo]

IT IS NOT MEMORY-HUNGRY LIKE ARK dude :D

NOT EVEN CLOSE :D

NOT EVEN THE SAME BALLPARK XD

A modded ARK server with a long-running save can easily reach 20+ GB of RAM use on startup.

This game? I haven't seen the U5 DS reach beyond 8 GB yet.

[hot flame]

You are a very literal person aren't you?

[blazing halo]

(the U6/Experimental is a little less conservative, starting out at around 10.5 GiB to... about 12 GB I'd say)

[blazing halo]

It's not that. I mean, I am, somewhat, I suppose... but I run both, and these do not behave even remotely alike.

When I say that SF is geared towards having millions of structures, I mean it. ARK would die if you tried something like that. I've tried.

The difference is night and day. If SF is assembly code running on bare metal, ARK is a Java virtual machine running on top of a virtual machine.

But regardless.

The bottom line is this:

while the server is saving the game (which is very single-threaded), it doesn't respond to client communication.

If this time exceeds the timeouts, players start to drop connections.

The default timeout is 30 seconds.

If you exceed this save time, you'll start seeing problems with connection retention.

And you'll reach this long before RAM becomes a problem.

You can increase the timeouts server-side, but I am unsure if client-side timeouts won't disagree.

Besides, there comes a point where it also becomes moot - who's going to sit through minute-long pauses while the server saves the game?

From this analysis, one thing is clear: CPU single-thread performance is the key performance metric for this game.

RAM is a distant second, and disk space is essentially negligible.

[hot flame]

yeah, I just wasnt sure if the memory usage was normalized (always around 6-7gb) or if the memory usage grew with the save size (like ARK servers do).

[blazing halo]

It does grow with size, but it grows in small increments.

The save I mentioned? The one that saves for 2.5min? It has half of the playable area covered in multilevel factories.

I shit you not.

[hot flame]

ok, cool. Thanks for the info.

[blazing halo]

This is what inside looks like.

So yeah... I wouldn't worry. 🤣

A quarter of it is under the map, too... 😂

[hidden rock]

nice piping

i like the funny colors

also patterns. i heavily underutilize those

[blazing halo]

That's Taro's save. I'm still in awe.

[frosty gale]

Didnt meant to have here a full support session, but glad that would go like that!
Will share in a min…

[hot flame]

You showed 2 of the 3 ports that should be open, do you have port 15777 open?

[blazing halo]

He can't show the third, it doesn't reply anything with basic checks like that.

As to whether it's input-only or if there are triggers for it (or it using a more sophisticated protocol), I can't tell.

[hot flame]

Which is why I asked if he had it open, because the query port (Default 15777) is required to join the server, which he is having an issue with, and which he didnt show he had open, which is why I asked.

[blazing halo]

Sure, that's a good question to ask.

[hot flame]

happy I have your approval now that I have given you my reasoning

[blazing halo]

... whatever floats your boat, man. I didn't say that to be antagonistic. In fact, I may've been a bit defensive, since that's my method of checking if the port forward works.

I may not be the easiest person to talk to, but believe me that I don't actively try for having an argument.

[hot flame]

Yeah, I just solved that by port forwarding the range 15000-15777 instead of adding 2 seperate rules since the 15000 beacon port automatically increments until it finds a port thats open

[blazing halo]

That is not good advice. I wish people stopped advocating for port-forwarding entire swaths of ports. :/

[hot flame]

I know, you have been nothing but helpful since I got here - I said that being playfully sarcastic - just forgot to punctuate it with the proper emotes 🤣

[blazing halo]

Personally, I think that the server auto-incrementing its ports (instead of failing when the port is not available) is bad form on the developers' part.

[hot flame]

It isnt an entire swath, it is 777 ports, which is fine especially when it is on a single protocol, and its not bad form, its literally part of the engine.

[blazing halo]

ARK fails when the requested ports are not available. Conan also fails. DnL also fails.

It's up to the developers of the game.

I believe that, in this particular instance, they're doing this for the sake of their internal "router" thing, which I also find problematic.

From a sysadmin/netadmin perspective, a service should not silently fail and self-reconfigure like that. It should fail, with a clear error message.

Especially since if you run multiple instances, it's impossible to know ahead of time which port will actually relate to which instance, as it's launch-order-dependent. That's a big "no" as far as I'm concerned.

[hot flame]

well considering it is still in EA, and that value cant be manually changed, the workaround is to open ports within the incrementation so that it can successfully launch.

[blazing halo]

To clarify: from an admin perspective, everything is allocated resources according to requirement. "A place for everything, and everything in its place." would be the guiding principle. As such, the server will already be assigned its proper place in the system. It shouldn't attempt to self-reconfigure; instead, it should clearly communicate failures so that precise admin action can be taken. I sincerely hope they straighten this out before going Release.

This sort of functionality (along with the internal router and so on) can be considered "clever", I suppose. But if you're an admin, you don't want "clever" - you want "reliable" and "replicable".

If the server is stated to use input ports A, B and C - I expect it to listen on A, B and C... and not listen on A, and then make loopback connections to B and C via 127.0.0.1. If the machine is output-limited, that will fail and it won't be immediately clear as to why.

[hot flame]

I dont look at it as clever, I look at it as a temporary solution until they can dedicate more time to the dedicated server.

[blazing halo]

Maybe. But someone actually wrote the internal router, and then wrote in the mechanics to disable it via the -multihome= switch. Which I also have some choice words for, by the way.

It's clear to me that the server is doing things this way so that it can be "more easily accessible" to the general public. And while I can understand and appreciate the sentiment, there's a number of concerns when running any sort of public services that go beyond what the service is.

As such, I feel that the developers shouldn't be taking this into their own hands. Networking is difficult. Running services is difficult. There are many concerns that go far beyond running the game itself. By creating shortcuts, they potentially hide these (and other) difficulties from people. I think that's a potential disservice.

I'm fully aware how it sounds to say that something is "just not done". But these things aren't done for a reason.

Also, I'm pretty sure the game is doing something funky regardless of the settings, because I can't get it to work over my NAT no matter what I do. :-/ Neither server nor client. It also refuses to work over tunnels, for whatever reason, and I think these issues are related.

[blazing halo]

And?

Though tbh I don't expect it to work... I posted my issue on the QA site, and I got a reply from someone sharing my issue... and they're running on GCP. Symptoms seem to align: can get replies via nc, but can't connect.

( See: [h]ttps://questions.satisfactorygame.com/post/62e2e201ca608e080351af06 )

[quiet cairn]

Strangely enough with my experience with Garrysmod. It will also auto increment. Maybe source related and not satisfactory's fault?

And how steam handles P2P connections

[blazing halo]

As far as I remember, goldsrc hlds doesn't do auto-incrementing.

[blazing halo]

I'm not touching anything steam-p2p-related with a 10-foot pole.

It's not straight-up IP. Thanks but no thanks. :D

Also, that's Steam-specific then.

Classic hlds wants 27015/udp.

From what info I can find, there is a self-incrementing port, but that's for VAC. So, like I said, Steam-specific.

[quiet cairn]

But maybe they applied the same rules from P2P to dedicated. For whatever reason, I don't know why

[blazing halo]

They should not, it's a completely different beast.

[quiet cairn]

But here we are

[blazing halo]

If they did, it would explain why it's broken.

You'd have about as much success with using bike wheels on a car.

[subtle swallow]

hi could i join a satisfactory experimental server

[blazing halo]

How do you mean?

[subtle swallow]

like a server you can join?

[blazing halo]

No. What is meant is dedicated servers - a piece of software you can install and run, to be a host for up to 4 players.

The key advantage is that it can continue running when everyone logs off, doesn't require a GPU to function, and behaves (mostly) like other system services.

If you're looking for people to play with, try #looking-for-group-old

[mental locust]

Yessss

Own the satisfactory.dev domain 😜

Hi kadd

[bold jetty]

is there a way to run a dedicated server on a arm64/v8 procesor architecture

[loud minnow]

No

[bold jetty]

not even with shit like box64

or is there a way to like amulate

to amd64

[loud minnow]

Satisfactory doesn’t support ARM base architecture

So simple answer is no

[bold jetty]

bruh are they ever planning to make a arm build

[loud minnow]

Idk why don’t you go ask them lol

[lilac juniper]

A couple of things that seem like bugs. (North Forest)

• When you walk past a node you hear a Miner but there isn't one there.
• If I start to manually gather ore from a node and walk away my character continues to harvest.
  Not sure if these are Dedicated server issues or experimental in general.

[quiet cairn]

There are a few ways to do such things with x86 programs. One such way is with QEMU. But even if you somehow jerry rigged something after hundred hours of researching it and trying. I guarantee the overhead will be so high anyway. Only other is Apple's Rosetta. It doesn't work well for games apparently though

[quiet cairn]

So the short answer is definitely no. Why do you need to anyway?

Even if you could get it to run, they don't even have the proper specs to run the game anyway. Only like the MAC could, so it's just foolish anyway.

[bold jetty]

a friend and i have a headless server but it runs on arm

[blazing halo]

The game has some elements that are heavily biased towards single-core performance. Assuming your emulation's cost is exactly equal to the improvements ARM may offer in terms of instruction efficiency (and being realistic, it's likely much higher), you're still looking at something like 1.8 GHz per core. ;P

[quiet cairn]

What could this possibly be? 🤣 A snapdragon or Qualcomm chip? You out of your mind if you think those little boys could run a game server

[blazing halo]

I wouldn't be so quick to judge, to be fair... ;]

... but the truth is, the CPU clocks might be a bit too slow, apples to apples.

[bold jetty]

ye cpu not best bnut i think we hae like 24

gigs ram

its like verry good for mc

but not satis ig

[quiet cairn]

If you got satisfactory to perfectly run with all the crazy virtualization and no hit to performance theoretically then it would just barely be able to run. In my opinion

[blazing halo]

For instance, as a general comparison, that machine I posted pics of manages 150% of the performance of my i7-4770 (1 thread) as repoted by sysbench.

[blazing halo]

Probably... though I could test this... xD

[quiet cairn]

Excuse my sarcasm @bold jetty but my suggestion would be to invest in a different setup. For something cheap and similar size then id suggest a hp or Lenovo Tiny PC. You could find an old one on eBay for like less than $100 and it would be much better and easier than anything you're trying to attempt.

[blazing halo]

It occurs to me that I have this machine, and it does run Linux... xD

[bold jetty]

its a vps

[quiet cairn]

I didn't even know that was possible

[blazing halo]

The only real answer you can get would be a "try it and see".

[quiet cairn]

Seriously tho. I had no clue you could. It's usually KVM virtualization anyway so why wouldn't they just allow you to do x86

Who would sell that

[blazing halo]

Honestly? Most common stuff has arm64 binaries in packages, and if it doesn't - you can always compile from source. The only problem is closed-source binaries.

I'll look into it later. Getting qemu for static emulation isn't an issue in itself...

[vale parrot]

Its probably free tier Oracle cloud VPS. Those run ARM A1 and 24GB of memory.

[blazing halo]

Well, this is turning into a bit more of an issue than I expected, for me... I'd have to rebuild my kernel with binfmt. :D

[loud stream]

How do I get a server

[frosty gale]

Sorry for the delay... I was afk and didn't come back to PC...
Here are the logs:

For some reason, I have beeing able to connect once to the server. But after I stopped the VM to do some maintenance and run it again, it never worked.
I even recreated the VM from scratch a couple of times with the same startup script. No success.

Also on the Satisfactory client side I tried the DNS, tried the direct IP, nothing helped. The message is always the same:

[quiet cairn]

I think your VM is broke. Try running a different game server.

Only way to find out

[frosty gale]

I tried. Couple of times already. Same result.

[quiet cairn]

Well then you found your answer

[frosty gale]

What?

[quiet cairn]

I said...

If any game is broken in your virtual machine on that vps...what are you doing in the satisfactory discord

[frosty gale]

I mean, I only tried with satisfactory... Would be too much effort to try to make another game server there... I would prefer take this effort to find the issue for satisfactory...

But, for example, I have a TS server running on that vps

works great

[quiet cairn]

That's what I asked. Idk why you said same result above

[frosty gale]

Sorry, misunderstood the question

[quiet cairn]

I was asking if that was the only application that had an issue with ports facing including game server

[frosty gale]

Not really... Even with the current server, if I test the ports, I have some results:

Doesn't seem to be any port issue

[quiet cairn]

I apologize for my rudeness. If I had a $1 per time a person on here asking questions regarding ports or VM when they don't work with anything including satisfactory simply because their ISP or VPS blocks it, I'd be a millionaire

[frosty gale]

Completely get your point. I tested the port connection and seems that it passthrough the firewall.

[frosty gale]

This is from another network, ofc

[quiet cairn]

Is it Linux or Windows

[frosty gale]

Linux

[quiet cairn]

Steamcmd?

[frosty gale]

Here is the startup script for the fresh ubuntu 22.04 minimal image:

(The IP might change, let me know if you want to test it directly to get the latest IP)

[quiet cairn]

I found an error message similar to yours on the satisfactory forums with no solution unfortunately

[frosty gale]

Looks like they are not really reacting to those bug report posts in their forum...

I was also searching around and found some more posts with similar issue and no official reaction...

That's how I end up here.

[oblique ice]

Which forum are you talking about? The QA site? They just got back from vacation last week.

Which could also explain elsewhere.

But Dedicated MP has been in experimentail for a LONG time.

[frosty gale]

I have seeing some posts from 1y ago with no real official answer...

[blazing halo]

@frosty gale yup, looks like you're having the same issue I am.

If you've been here https://questions.satisfactorygame.com/post/62e2e201ca608e080351af06 - that's my post.

[frosty gale]

Yep! I'm the only one commenting on it

[blazing halo]

You're marcosvr?

[frosty gale]

Yep

🙂

[blazing halo]

Well that would've been good to know before I brought it up. xD

[frosty gale]

Sorry... Too many nicknames 😄

Right now I'm trying to follow the configs from this repo: https://github.com/wolveix/satisfactory-server
I'll let you know if it works in anyway

[desert walrus]

@blazing halo I upvoted 😄

[blazing halo]

Yay. :D

[blazing halo]

For me, it's a routing/iptables problem, so I'm not sure how reconfiguring the service will help...

[frosty gale]

Just trying out some more possibilities 🤷‍♂️

It doesn't even read the configs from the ini files 😦

[soft chasm]

dont know if anyone else has experienced it or not but running on a dedicated server I never had rain in any of the biomes...took that same save file and went single player and got rain.

[frosty gale]

Have you tried to run the server with the argument -DisablePacketRouting?

[frosty gale]

Does anyone know why it's trying to use another port (56084)?

[desert verge]

Did you play for more than two hours on the server? There was a bug about the random seed for the weather system that got reset to the same value everytime someone started the game

[frosty gale]

@blazing halo I could just reproduce the same behavior locally. I'm running this docker container image in another PC (Linux) in the same network as my gaming PC. Same timeout issue. Nothing really blocking the communication between both PCs.
The docker image: https://github.com/wolveix/satisfactory-server
You mentioned in your post that it worked for you locally. Could you share your setup?

[blazing halo]

Direct install.

[blazing halo]

That looks like the listening bit of its IP4-to-6 tunnel. (the existence of which I don't fully understand)

[crude steppe]

Hello, I have a small problem with the Satisfactory DS on Epic Game, I launch the application but in Satisfactory my localhost is on two different versions? How can I do it?

[blazing halo]

Yeah, EPIC currently has the versions borked (both options install the same version). Use steamcmd.

[crude steppe]

Oh ok thank you for your answer, do you have a tutorial to follow for SteamCMD ? Or the wiki page?

[blazing halo]

Wiki page. It's in the topic of this channel.

[blazing halo]

But there is something that could be blocking the communication - the interface between the virtual machine, and the physical machine. Do you need to either bridge the network connections or forward ports internally with Docker? Because I know you usually do with other virtualisation software. I'm admittedly not good at all with Docker.

[frosty gale]

Actually I don’t know that deep about Docker too. But I would assume that there is some kind of interface between the Docker network and the host network forwarding the requests from the ports assigned to the container. You are most probably right!

[blazing halo]

One word of note: the ports must be mapped 1:1, at least the beacon port needs to be (don't know about the others). This is because the game actually hands out port information to connecting clients, so they must match up (since the game won't know about any mappings you may do).

[frosty gale]

All ports are 1:1…

For example, for the Docker container:

satisfactory-server:
image: 'wolveix/satisfactory-server:latest'
ports:
- '7777:7777/udp'
- '15000:15000/udp'
- '15777:15777/udp'
environment:
- MAXPLAYERS=4
- PGID=1000
- PUID=1000
- STEAMBETA=false
labels:
traefik.enable: false
restart: unless-stopped

[formal vigil]

Our server had a crash and there are no enemies spawning except for the bees. Anyone have any ideas?

[blazing halo]

There aren't actually any enemy spawns, really. The devs are reworking enemy AI in Experimental.

As far as I can tell, the depopulation is the combined result of no new spawns, environmental hazards (which now affect critters as much as players), and fauna infighting.

Essentially: they kill each other or die in the toxic gas, and no new ones are made.

[formal vigil]

Oh ok. So this is a common problem?

They were just all gone after the server got back up and the bees endlessly spawn back when there's no factory nearby

Some of my friends worry if the save is broken. But from the sounds of it, this will be fixed in the near future?

[iron tinsel]

need some help with DedicatedServer on SteamCMD. Server is running on a server on my LAN, i'm able to connect via steam on another computer on my LAN, friend I'm trying to play with not on the LAN cannot connect. I've checked everything ports 15777, 7777, and 15000 are all open on my router but he still cannont find the server. I tried sending him an invite from the session but I can't send him an invite as session is "private" which I cannot change to "friends only". I can connect from my computer to the server via both the IPv4 address and IPv6 address

[blazing halo]

Have you verified that you have a public IP?

[iron tinsel]

Have you ever had one of those moments where you thought you were smart and then realized you were dumb?? Thanks, that fixed it

[blazing halo]

... you were giving out the wrong IP, weren't you? 😆

[iron tinsel]

lol yes, yes I was

[blazing halo]

But you're also wrong about something: intelligence is the ability to perceive and parse information; making the correct choices is wisdom. And that we all have problems with, every so often.

[iron tinsel]

well said

[blazing halo]

And here I was, leading with the public IP thing in an effort to not make it seem like I expected you to be an idiot. xD

But I guess it paid off, because you took one look at that question and solved the issue.

Relax, I'm just ribbing. :-) It works, that's what matters. Go forth, have fun!

[iron tinsel]

lol right! well I appreciate it

[brittle viper]

It seems that my map markers do not get saved. Is this a bug or am I missing a step when placing markers down?

[marsh dawn]

iirc, it's a bug.

[frosty gale]

So, coming back here to add one more finding:
Looks like it's not on the server side but rather on the client side. My friend was connecting to the server without problems yesterday and playing it for a couple of hours.
While I still can't connect. 😦

[blazing halo]

So you have the client side of my issue.

[frosty gale]

Looks like yes

[blazing halo]

If you're the only one with an issue, I know VPN works. You can try setting up something like OpenVPN on the machine, and connecting through that.

[frosty gale]

https://tenor.com/view/sad-cat-sunakook-tired-exhausted-gif-24687868

[obsidian rune]

What's your router, though? I'm curious.

Unless it's a Windows issue of some sort, but we'd be seeing a lot more of it if it was...

[frosty gale]

Standard german Vodafone router... Vodafone Docsis 3.1 (Arris TG3442DE)

I had my first success connection today after a couple of days. Played for some minutes. Than I closed the game and open again, bam. Timed out

[plain sand]

is there any way to reduce the autosave interval? in multiplayer the autosave constantly causes me to lag during building which breaks the elements or during longer hyperloop travels I keep getting stuck in the pipe.

[frosty gale]

https://satisfactory.fandom.com/wiki/Dedicated_servers/Configuration_files#Autosave_slots_and_Autosave_interval

[plain sand]

cool thanks!

[marsh depot]

how come nothing is coming up when i try to connect to my server

im using steamcmd on a windows server 2019

[frosty gale]

So many possibilities... Port blocked by firewall, wrong IP, some error on the server...
Could you give us more details? Logs? Tests? Debugs?

[marsh depot]

how do i get latest logs and send here

"Port blocked by firewall" im using the local IP, but maybe its something on the windows defender firewall but nope i didn't even see satisfactory or steamcmd in the firewall list

[frosty gale]

Should be somewhere in %localappdata%\FactoryGame\Saved\Logs
Or if you are using NSSM or running as a service:
C:\Windows\System32\config\systemprofile\AppData\Local\FactoryGame\Saved\Logs
:\Windows\ServiceProfiles\NetworkService\AppData\Local\FactoryGame\Saved\Logs

[marsh depot]

there is no logs folder or files in that location (also am not using nssm yet)

this is it

[frosty gale]

Most probably it's getting output on the console

[frosty gale]

Nothing come to my eyes here... Maybe is something with the firewall, but I'm not a windows server guy so I don't know how to help here...

[marsh depot]

windows server is the same as windows, not really any changed features just added/removed features

[loud minnow]

Simple way to check if is the firewall just disable windows firewall for a bit see if you can connect and if you can then is your firewall setting are incorrect @marsh depot

[frosty gale]

I know what windows server is 😄

Just don't know how to check the firewall and so on

Does it helps? a2hosting.com/kb/getting-started-guide/internet-and-networking/troubleshooting-network-applications-with-powershell-and-tnc

[marsh depot]

fixed it, I didn't run the port forwarding powershell command provided on the wiki

my bad

[frosty gale]

Good that it worked 🙂

[blazing halo]

Which means it's blocking. It's restrictive by default; if it's not on the list, it's denied.

[blazing halo]

It's not a "port forwarding" command, it's an "open the firewall ports" command.

[empty jasper]

What is the performance like for dedicated servers right now? I tried a vps but had lots of lag and every so often we got put back a bit. I read it needs like a ryzen 5. Do people have good expierences with using VM's?

[blazing halo]

If "every so often" was 5 minutes exactly, that's the auto-save. ;P

[empty jasper]

Nah i put the auto-save to 30 mins, also chests sometimes bugged out with items not loading and stuff

It was on a VM with around 2,3GHZ 6 cores, so i hoped it would work

[graceful mist]

My server just recently logged me out due to the whole Steam servers issue. I now do not remember my admin password that was set. Is there a way to retrieve it?

Nevermind. Got it.

[quiet cairn]

My servers CPU is 9 years old and runs flawlessly. A good dedicated VPS is like $100 a month minimum. Y'all on some crap cloud VPS, even if it says it isn't, it is. Check the hardware information via command line, run a couple of benchmarks. You get what you pay for, nobody should be surprised.

[loud minnow]

Rather just save those 100 dollars for three months and make myself a low power computer and would be way better than a VPS my opinion

Lol

[quiet cairn]

That's exactly what I did. My Lenovo Tiny PC with a i5-3570 was $150

[loud minnow]

Yeah is dumb paying for renting very old equipment and companies try to charge you a lot for renting service

Most of the time that you see rent a sever for this low amount of money yeah your renting probably 10+ year old cpu and very old equipment

[quiet cairn]

It's not that the equipment is old. I imagine it's good equipment. It's just completely virtualized. Spread across hundreds of people, there's certain times when usage is higher then others.

My CPU is literally 9 years old and it's completely fine.

[loud minnow]

Yeah it usually is but most of the time is old Intel Xeon

With very low cpu core clocks etc anyways my point is is better to save the money and buy something cheap for your hosting need than renting out imo

[quiet cairn]

Waste of money. It has it's purposes when your like making a website for example. But if your running a game server...it's almost never a good idea

[loud minnow]

Yeah of course.. for those uses then yeah you have Amazon web service etc for that… but for game server is a waste of money

[quiet cairn]

I wish someone would have told me when I was younger. I would have been able to save up for a better server today

I mean I could afford it. But I've already wasted that much

[loud minnow]

Like my server is a 3950x with 64GB 3466 cl16 dominator ram kit, with a 980 m.2 drive and other SSD for game storage.. is way more powerful than anything I could rent out and it was all made with left over parts from upgrading my main rig

[quiet cairn]

haha

Definitely

[loud minnow]

Yeah renting for a game server is a waste just go on eBay buy your self older laptop or even does old dell pc and just make that to a game server and just add much ram as you can because that is the main thing with hosting game server is how much ram you have and cpu single thread speed

If you like hosting server if you just want to play with your friends for a weekend than yeah go and rent one out nothing wrong with that but for the long term not worth it

[quiet cairn]

Only time when it actually makes sense to rent a server, is when your opening a public community. You don't want to get dos or ddos. At that point your taking donations to run it, so that's fine. If it's just for your friends, which is what most people do for this game, old equipment with the recommended specs. Is the move.

The recommended specs being (wiki states)

Processor Recent (comparable to the i5-3570 [Intel] or Ryzen 5 3600 [AMD] or better) x86/64 (AMD/Intel) processor. No 32 bit or ARM support. The server favours higher single-core performance over multiple cores. Memory 6GB minimum, 8GB RAM is recommended for larger saves or to host more than 4 players.

[loud minnow]

That ram requirement will need to be increase for update 6 tho that for sure because it won’t start with 6gb or 8Gb you need at least around 9-12gb for update 6

[quiet cairn]

Yeah that hasn't been updated yet

[loud minnow]

Yeah they going to say 12-16gb requires

For a game server to run fine

[quiet cairn]

I think their Ryzen recommended specs is a bit high as well. Ryzen 5 1600x I'd bet would work fine too.

[loud minnow]

Yeah that’s true I think is because the game is very the single threaded heavy and those 1600x single threaded performance is not really that good

And i5 from that gen would be faster than a 1600x

In terms of single thread performance

[quiet cairn]

Passmark says it's equivalent to the Intel i5-3570 in single thread performance. I think it's a bit AMD biased

[loud minnow]

I think I would be fine personally but don’t have one to test it out lol

[quiet cairn]

Fair

If anyone here has a spare Ryzen 5 1600x let me know. So we can fix the satisfactory wikipedia

[loud minnow]

I only have a 5950x, 3950x, 3300g I use to have a 5800x but sold it to a friend lol

[quiet cairn]

My main PC was a 2600-> 5600. I don't why you would need anything more for gaming tbh

To each their own

I still can't believe I bought my PC in 2018. And did a CPU upgrade with no hassle. Intel fanboys could never

[loud minnow]

Well for me is a hobby really I like buying the new new to check it out my main is a 5950x, 3090, 128gb trident z neo 3600 cl18 ram kit, 570 prestige motherboard and 980 pro etc… I pretty much maxed out the x570 platform lol

H

[quiet cairn]

The new am5 socket will be cool. But requiring new motherboard + ram as well...it's just not worth it for how expensive it is. The new generation is only going to be a 10% lift starting.

Max am4 will be future proof for awhile

[loud minnow]

Yeah I’ll be waiting a couple of gens before I jump into ddr5 etc this first gen is not really that impressive the speed that they are getting from ddr5 later gens of ddr5 will be faster

[quiet cairn]

I read that even though AM5 will exist. They will still make some AM4 chips

Some with small generational improvements

Until am5 is fully fleshed out

[loud minnow]

I mean if amd give me a 5950X3D I’m buying that in a heartbeat

Lol

I don’t think would happen but you never know

And I would put my 5950x on my current sever and I would built another with the 3950x lol 😂

[frosty gale]

I'm running the dedicated server on GCP, more precisely e2-standard-2 machine.
It has an Intel(R) Xeon(R) CPU @ 2.20GHz, 8GB RAM and 30GB SSD.
So far it works like a charm 🙂

Depends on which region you choose, you can run it for $8~15 Monthly (Spot price)

cloud.google.com/compute/vm-instance-pricing?hl=en#e2_predefined

[zealous mural]

Sorry to reference a message from so far ago, just curious if there’s been any more development on the query protocol?

Mainly has player count been added?

[zealous mural]

actually I just checked, looks like it's still just status, version and query port 😦

[rose valley]

No problem, I guess some sort of a status check was kinda overdue anyway. No changes were made to the dedicated server since it's launch since we had to redirect our efforts to other areas. However, we have loads of cool stuff planned and at some point in the future dedicated servers will get a second pass. As for why that specific protocol isn't documented, it's still subject to change just as much as it was back when we first released it. Remember the whole dedicated server is considered experimental.

[zealous mural]

Cool, thanks for the details

I referenced https://github.com/dopeghoti/SF-Tools/blob/main/Protocol.md and wrote a decoder in JS - if for some unholy reason someone wants to do the same thing, you can decode via Buffer (Node)

const protocol = {
    messageId:          (buffer) => buffer.readUInt8(0),
    protocolVersion:    (buffer) => buffer.readUInt8(1),
    userData:           (buffer) => buffer.subarray(2, 10).toString('ascii'),
    state:              (buffer) => buffer.readUInt8(10),
    version:            (buffer) => buffer.readUInt32LE(11, 15),
    beaconPort:         (buffer) => buffer.readUInt16LE(15, 17),
}

was hoping to add Satisfactory to my discord bot (as an example from another game)

but the only details would be queried is whether the server is responding, and which version it is running 😭

I don't even know if it's of interest, but I might look into building a light mod to support pushing this data out

[atomic mortar]

wh-wtf

[loud minnow]

just left over parts lol

[atomic mortar]

yeah but like

god fuckin damn

[blazing halo]

> leftover parts

I think we have different definitions of "leftovers", my dude.

[blazing halo]

Is there anyone I may message about the server's networking? Or perhaps the game's in general. It doesn't seem to want to work (server nor client) over NAT for me at all and I wanted to ask some questions about how it handles that, and I wonder if there's any rationale I could read for the self-incrementing ports? Because that last bit is weird and very unexpected from a sysadmin/netadmin perspective.

I already made a post about the former on the QA site, but it seems like a very specific issue that likely won't see a lot of attention. Not sure if I can do more.

[frosty gale]

Well, if it helps it to give some more attention: also happens to me, but on the client side.

[blazing halo]

To be honest, I'm mostly curious as to why. :D I mean, it must be a quirk of my setup, somehow. But what. :D

[rose valley]

It's vanilla Unreal Engine behaviour, in fact. As is pretty much most of the networking code. You can, however, override that port via a cl parameter. Specifics elude me now as I'm off the clock atm.

There was a change, however, that made it so you only need to forward/open one port only, the query port and all traffic is routed through that port. That is the default behaviour.

In general, when it comes to networking/NAT issues, there is little specific applications can or cannot do about it. They all boil down to your setup, network configuration, etc. However, there are a lot of people here that are willing to help troubleshoot networking related problems.

[quick mist]

I do really want to expand the utility of my fledgeling toolset once the API stabalizes somewhat

(also welcome back; hope your vacation time was great, Bogdan!)

[blazing halo]

Yeah, nobody really has a clue as to why the Client would keep sending handshakes after the Server sent its reply... or why it wouldn't reach the Client. :) It's very specific.

[rose valley]

Could be some form firewall that is too smart for it's own sake, flagging that behaviour as malicious and thus blocking it.

[blazing halo]

Not really. Both ends (server and router) Debian Linux.

[rose valley]

But that is speculative. Could be your ISP who has this smart kind of behaviour.

[blazing halo]

Except every other game and servce I port-forward works just fine. Only Satisfactory is affected.

[rose valley]

I assume they are not in the same network?

[blazing halo]

And this affects NAT loopback as well.

[quick mist]

the only real way to know would be to sniff UDP traffic at each hop along the way, and most ISPs are disinclined to help with that sort of experimentation

[blazing halo]

Funnier still - I tested the ports individually, and they work.

So all I'm really left with is Satisfactory doing something funky.

[rose valley]

Exactly. Most times, there are easier explanations but with networking, you can only speculate from the outside.

Well yeah, Satisfactory may behave in a way that can falsely be flagged as malicious.

Bit as long as we don't know specifically how, we can't do much about it.

[blazing halo]

Except the only place where it would be would be on my Windows machine (where the Client is), using Defender as-is.

And as far as I can tell, it's the replies that aren't making their way back.

If I had more information on how the game handles networking, I could look into more things.

I host other UT-based games (ARK, DnL, Conan) and they don't exhibit this behaviour.

But eh. I primarily wanted to know if there was someone I could talk to about this stuff. I didn't want to pull you into an impromptu debugging discussion. I'm well aware that you're all busy and it would be counter-productive to help solve one guy's issues. I can take "not really" for an answer. :)

[rose valley]

You're in the right place. While not an official support channel, developers sometimes help out as well and specifically with this issue, I was the one who set up the dedicated server networking so that should answer your question. However, the community gracefully does most, if not all of the helping here.

[blazing halo]

I know. I'm proudly a part of it. :)

[quick mist]

I've been almost as fascinated with Kad's long-running saga of troubleshooting eir issue as Kad has been (:

[blazing halo]

;]

[rose valley]

So, you have a server and your client cannot connect to it. Can other clients from other networks connect?

[blazing halo]

No. I port-forwarded correctly, I'm sure of it.

[rose valley]

But have you tested by having other clients connect?

[blazing halo]

I have. No dice.

[rose valley]

Alright, so then we can assume it's the server that's the problem in your case.

[blazing halo]

Right now I'm working around that by using a VPN server others connect to.

The server is on my LAN, and I can connect to it via LAN just fine.

[rose valley]

And I suppose you have a router at the edge of your LAN?

[blazing halo]

I do, yes. It's running Debian 11.

Packet filtering via iptables.

[quick mist]

(and as I recall we've tried stripping those tables as bare as can be in the past)

[blazing halo]

Yes, we did. Down to "forward all ports to the server machine", even.

[rose valley]

And, hear me out, this might sound like a bit too much but for the purpose of troubleshooting, have you tried having the server (or a VM) be connected to the WAN directly?

[blazing halo]

That I have not tried, but I fully expect it to work. It works over LAN, after all.

I know it's a NAT problem, I'm nearly 100% sure of it, one way or another.

[rose valley]

Well, if that would work, then we'd 100% know that it's your Debian based router that somehow blocks the packets. But I don't think it's guaranteed. LANs and WANs are quite different from each other.

[blazing halo]

I also cannot connect to any Internet games from my Client (neither EX nor EA). The router is the only common point for those two.

The symptoms are the same in both cases.

Server-side: LogDTLSHandler: Warning: Ignoring handshake packet received after completion.

Client-side: LogDTLSHandler: Attempted to send packet during handshaking, dropping.

[quick mist]

have you at any point contemplated (backing up and) burning your entire firewall setup to the ground and starting from scratch with an alternate firewall manager (e. g. Shorewall or ufw)?

[rose valley]

Interesting as those use a completely different net driver. nvm, though you meant regular multiplayer sessions, hosted by EGS.

[quick mist]

(I know it's all iptables at the end of the day)

[blazing halo]

I know neither at all, so it would be an exercise in frustration. :D Besides, iptables is the most direct, and I don't know if I need any extra modules for my kernel for those. But I may try that at some point... however difficult wiping it might be in the first place.

[rose valley]

I would go back to the recommendation of bypassing your router altogether or temporarily replacing it with a commercial one if you have it, just to rule it out.

[blazing halo]

That's something I can do, yes... temporarily, at least. I would need to find someone relatively close to me to test both ways though (my usual players aren't available at the drop of a hat, and neither can set up a server).

[quick mist]

When most other possible variables have been eliminated, than one of the presumes constants, no matter how certain, must be the crux.

[rose valley]

Yes, temporarily. It's just troubleshooting. We rule out parts of the net route until we know where the problem is.

[blazing halo]

Hey @quick mist would you be willing to help me test? Where you at?

[quick mist]

Phoenix.

I can help test when I'm not "at work" though.. about four hours hence

[blazing halo]

Oh. Distance might play a small part (Poland), but connecting should show something at least...

[quick mist]

I can test with sfcheck at will though.. it's a full client connection that would have to wait

[rose valley]

I can help you. Sweden to Poland should be 100% doable.

[blazing halo]

The first test would be for me to connect out, I suppose. Any known-reachable SF server would do.

[quick mist]

you can at least try to connect to mine; stand by and I'll light it up

[blazing halo]

I still need to reconfigure my emergency spare router, so - no hurry.

[quick mist]

I have taken the liberty of DMing you my server address

[blazing halo]

Thank you. I assume the port is 15777? EA or EX?

Ah, EX. But this also pretty much confirms that it's my router's doing. I can see that game, and get the version mismatch error. Very nearly instantly, too.

Yes, I can now see the game and have the option to "Authenticate".

So that confirms outgoing communication is OK with a different router.

Anyone want to try connecting to mine, now?

[quick mist]

shoot me the address and port

[blazing halo]

Sent. Expect EA (Update 5).

[quick mist]

looks happy:

[blazing halo]

> $thehost

... nice. ;]

[quick mist]

putting it into a variable is easier than having to redact the IP from an image

[blazing halo]

Indeed it is.

[quick mist]

(or even text)

[blazing halo]

I could use a second pair of eyes to be sure it can be seen by the client, but this is promising.

[rose valley]

I can fire up an EA game and try to connect.

[blazing halo]

Sure. DM the connection details? Expect EA, as seen above, build 188609.

[quick mist]

my tool uses the exact same communication the client uses to "ping" a server from the menu AFAIK

[blazing halo]

Hm, please DM me so I can DM you, or something. I'm blocked.

Sent.

Interesting... now when the server accepts the connections, I can see it printing the remote IP.

So, this is interesting. When the join happens, I can now see the remote IP of the player. Previously, it was the IP of the router.

[quick mist]

ah, thar's the rub

[blazing halo]

Yup. The game doesn't want to work with tunnels at all, and that seems to be why.

[quick mist]

if the server thought it was the router connecting, it would then try to respond to the router as if it was the client, and whoops there goes your connection

[blazing halo]

But the question then is: why is it doing that? Every other game has no issue with that...

[quick mist]

heightened sunspot activity?

[blazing halo]

Bogdan was able to get into the game. That confirms my suspicions.

Now the question is why.

Specifically, why does the game see the router's IP, when it should be seeing the remote IP.

Or, perhaps, why doesn't the router recognise returning traffic as it should?

[rose valley]

The address the game sees is the sender address that is part of the packet header. Your router changes that address during packet routing.

I am a little confused (but it's been months since I was in any networking code) as to why we can't handle that well in the game though.

[blazing halo]

Well, iptables is mangling the packets... so replacing the src/dest address is actually par for the course. Question is, why doesn't it recognise returning traffic?

[quick mist]

UDP is connectionless

all the router knows is "application directing a packet at me"

the router should not be mangling the data to the point where the server does not know the actual source address

[rose valley]

I don't think my router for example replaces the IP addresses in the packet headers, only the ports. But I didn't check it right now tbh.

[quick mist]

your router is somehow trying to act as a proxy server for UE4 and that doesn't fly

[rose valley]

Yep, your router is doing a little more than just routing. It's likely a form of enhanced security but it's clearly going wrong somewhere.

[quick mist]

the router should be mangling the IP for the destination of the incoming packet (that's what IP masquerading is), but the source address data should be untouched when it arrives at the server process

[rose valley]

At least that's how routing is performed conventionally.

[blazing halo]

Okay. I dumped iptables from the emergency router. I'll study them to see if I can understand what I'm missing.

I will also put together a simple service that will show me what IP I'm seeing internally (after the translation).

It's likely that this has been an issue ever since I started out, but it never really affected anything...

... until I tried returning UDP traffic in either direction.

Now, let's see what merry hell I'm in for. Rewiring for the main router. :D

[quick mist]

at least now we have a new and very insightful data point into your issue!

[blazing halo]

I do.

Okay. The network is back up.

Okay. I have removed an entry. I need a volunteer to try and connect with the game client.

I think I've identified the rule that's been causing my problems.

I can now connect to Ghoti's game, from what I can tell.

[rose valley]

Happy to oblige

Nope. Server is offline.

[blazing halo]

Different IP. New data in DM.

Same server, new IP. My infrastructure is back online, so I can now use my DNS name.

[quick mist]

was gonna say, connection timed out for me, but if the address changed that stands to reason

[blazing halo]

Indeed.

Please do a full connection check, if you're willing.

[rose valley]

The server is still offline (as seen from within the game client)

[blazing halo]

That's... not good. :/

Still nothing?

[blazing halo]

Thank you. I will continue to debug this issue later. There's progress however.

[devout bronze]

how do i find the address of the server? im new to this.

[frosty gale]

I feel like the dumbest person on this conversation... I have some slightly knowledge about what you are talking about but still can't really help.
Nevertheless, here comes the dumb question:
I can confirm that I'm also having the timed outs because of my router. I just tried to use my 4G from my phone to connect to the server and it works flawless.
But for me, the problem is on the client side (My own PC and network). The server is running in the cloud and my friend can connect without problems.
Now the question is: What can I do with my standard router from Vodafone about it? AFAIK I can't change that many settings there. Any idea already how to fix this?

[frosty gale]

Are you running the server on your PC? If so, you need to find the address of your network (https://www.google.com/search?q=What's+my+IP) and you will need to forward the ports from your router to your PC.

[devout bronze]

thanks bro

[rose valley]

That sounds even more strange. Do you know if you have a public IP address? (I'm not sure if we can handle CGNATS if that's not the case).

[quick mist]

can you clarify exactly what the question is?

Are you having trouble connecting to a server in the cloud with your client behind your Vodaphone router?

[frosty gale]

No, I don't have a public IPv4 address... This is also one of the reasons why I decided to host the server on the cloud...

[quick mist]

AFAIK CGNAT should not prevent a client connection but I can see how it might

[frosty gale]

The question basically is how can I connect to my server from my network (Behind a standard Vodafone router)

[quick mist]

Do you know what the medium of your internet connection is? 'Vodafone' implies cellular but I can see how it might well just be DSL or Cable etc.

[frosty gale]

Cable

[quick mist]

okay, that's good. A normal cable internet connection should pose no problem connecting a client behind it to a dedicated server on the greater internet.

[rose valley]

Yeah, I'm not sure if our net code (specifically the parts that are custom made, the query port) can handle it, which is why I'm asking. Unfortunately I'm only working with vague memory recall at the present moment.

[blazing halo]

He's seeing the same issue I'm seeing - the same log entries.

Except in his case, he's in a different network, and is trying to connect to the Google Cloud.

In my case, I'm fairly sure my NAT hairpin configuration is the problem. It works fine for TCP, but falls hard on its ass when Satisfactory's UDP is considered. It works fine for literally any other test I can run... but it's still broken somehow, obviously.

[frosty gale]

Another interesting fact: After some tries (around 15x) I can connect one time to the server.

Now that I figured it out it's my router, I connect with the following steps:

1. Connect my PC to my phone that routes its 4G network
2. With it, I can see and connect to my server
3. Disable the 4G routing, the computer fallback to my WiFi and I'm still connected to the server. I just keep playing...

[quick mist]

okay that is just wild

seamlessly failing over from the mobile hotspot to the ISP connection? Wild

[frosty gale]

Exactly... It just works 😄

[rose valley]

Yeah, I can attest our net code is not that elaborate. Some dark magic is going on there.

[quick mist]

is your phone somehow sharing your Wifi connection? That'd at least make some sort of sense even though that'd be.. what, triple NAT?

[frosty gale]

No... The phone is routing the 4G network through the WiFi hotspot it creates

I also have ExpressVPN, but I also can't connect to my cloud server through the VPN (The server isn't in the same VPN). So I'm pretty sure it's my router

[marsh dawn]

It's "a kind of magic" - Queen

[frosty gale]

One dream, one soul
One prize, one goal
😄

[rose valley]

Hey, not a lot of people I know listen to OSI!

[blazing halo]

Also, @quick mist your python tool can obviously test the server's up-ness... but it doesn't actually say anything about whether the client will see it. :/

That's the fun part. It responds on the ports, so you can read the response... and from the client - nope, can't connect, server offline.

[quick mist]

it would actually because if the tool doesn't get a response back it doesn't see the server state and version

the tool, primitive though it is, requires at least one two-way conversation to not be considered a timeout. If message 0 doesn't make it to the server, that obviously goes nowhere, but the message 1 reply has to make it from the server back to the tool

[blazing halo]

Yeah... but for some reason, the handshake replies don't make it back, but everything else does?

[rose valley]

That plays into the entire setup and it's weirdness. There are three ports: Query, Beacon and Game. Query is what Ghoti's tool pokes, that seems to work. It seems to break down for the other communication channels (that the games uses to establish a connection). Try disabling packet routing (check older patch notes for details on how to do that) and see if that makes any difference. Note that all of this will likely get a big ole rework at some point.

I need to sign out. Good luck in your search! I'll check back tomorrow to see if you got anywhere.

[quick mist]

At some point I'd love to be able to get the sftools package to the point where it can query a server, get the server / session name, number of players, time-in-game etc that are shown by the client prior to actually logging in for a game session. I know that would mean exposing the auth mechanism to scrutiny so I'm not holding my breath for that though (:

[blazing halo]

I have an idea...

@rose valley Hey... does the handshake, as implied, use TLS? Would part of that use the source/destination IP somewhere in there? Following that: is it possible that the handshake doesn't come back, because the reply is going to {router IP} and not {actual client IP}?

Would imply that the handshake does come back, but is ignored due to address mismatch... and if the address is used in any way in the crypto, then it would outright fail crypto.

Would also explain why the game doesn't work with tunnels - they change the source address.

[blazing halo]

If "LogDTLSHandler" is any indication and that "TLS" part in there means what I think it means, it might not matter without the proper encryption key.

(and that "empty certificate identifier" error would strongly imply that it does mean what I think it means)

[zealous mural]

+1

[rose valley]

An API will be built at some point. We still have that in the plans. You'll be able to do quite a bit with your cool little tool.

[rose valley]

The secure channel is built on DTLS, which is basically TLS over UDP. Among other, the peer address is used to generate the cookie for the DTLS session.

[blazing halo]

And how might that react to a double-masquerade situation? Because NAT hairpin is exactly that - it's masquerading the server's address (because behind NAT), and the client address (because hairpin). Would it affect it in any way? Logic would say "should not affect" (everyone is theoretically behind double NAT, in a way), but I have to ask. Because the communication is clearly happening - except it doesn't work.

Oh, I did resolve my Internet server reachability issue. What I had was never an issue before, but now it's fixed. Now I'm only really left with NAT hairpin.

[rose valley]

I don't think it could. As long as the sender address is consistent. In any case, it'd be really hard for me to tell without setting out to do an actual investigation into this.

[blazing halo]

Thankfully it's not a showstopper, and I mainly would want this resolved for completeness - every other service I host doesn't mind, after all. But I won't cry over the fact that I have to connect over LAN instead of looping around my NAT. :D

Just as long as you don't end up doing what ATLAS did - which necessitated this setup in the first place (ATLAS servers pass on the IP as well as ports, so even if you connect via LAN - you receive the public IP in return; it makes some sense that they do that, since you actually connect to a cluster).

[rose valley]

Since I don't know exactly where things go wrong, I can't say that we will or will not address this use case (although we will for sure improve reliability/connectivity of the server). So please make sure to add all relevant information to the QA site. I know it looks like nobody is checking but we are. When the server's turn comes to get some love, I'll look through all the QA posts and I am particularly interested in your setup.

This just occurred to me but did you check that the port number stays consistent for a specific client and doesn't change over time?

[blazing halo]

To be fair, the connectivity looks fine. All of this so far has been misconfiguration on my end (which I generally expected) - though a very particular situation indeed. I still have issue with the self-incrementing ports (since in the "server world" things should fail instead of silently changing its configuration to something that doesn't match the rest of the network's setup), but that's about it from my IPv4 view.

[blazing halo]

I didn't look at that, no, mainly because I don't really see the initial (to NAT) connection. Would have to dump some logs on what comes in, and what goes out again.

But you're right in that if the ports are different, it might affect it.

The NAT would still forward it back-and-forth just fine, but... the ports would mismatch between client and server.

[rose valley]

Yes, and since it's UDP, the router SW might decide that it doesn't need to keep them consistent since UDP is on paper connectionless.

[blazing halo]

It is connectionless.

[rose valley]

But Unreal's net driver is in fact establishing a connection and expects a client to always send it's packets through the same port.

[blazing halo]

It's just not sessionless.

[rose valley]

See my follow up. UDP is. Unreal's net stack isn't.

[blazing halo]

It can't not be. I understand what you're saying, though.

It's much like TCP over IP, your approach.

[rose valley]

More Unreal's but yeah. I really wonder at times if it is indeed more efficient to essentially reimplement TCP yourself rather than just use the real deal.

[blazing halo]

And I can see why you do this - it gives you the benefits of TCP, without the drawback of having to suffer stuff you can't switch off.

(like retransmissions or waiting for the whole segmented thing to be reconstructed on receipt)

[rose valley]

Have you by any chance tried to disable the packet router btw?

[blazing halo]

(especially when the packets are obviously no longer relevant)

[blazing halo]

I do use -multihome=, and if the Wiki is right in that it disables it - yes.

Not sure if I need to, now.

[rose valley]

It would be great if you actually tried to enable it then (by not using multihome).

[blazing halo]

I'll give it a test later, sure.

There's no real explanation of what that does and how it does it, so I played it safe by getting rid of it (especially when I start seeing stuff like it trying to loop back to itself over IPv6...).

Perhaps if there was a paragraph or two for admins' benefit on the Wiki?

We generally like to know what to expect from looking up open ports by process. ;)

[rose valley]

It's just a simple router built into the server. With it enabled, all packets from clients are sent to the query socket and the query socket simply forwards them to the actual receivers. A lot of people complained about having to open/forward 3 ports and this was an attempt to address that by reducing it to only one port, without having to rewrite the whole thing (including parts of the engine).

There was an explanation in the release notes albeit maybe not very comprehensive.

[blazing halo]

That's in stark contrast to what I'm used to, to be honest... Most Wikis just tell people to open like a thousand ports on both UDP and TCP... I've seen writeups that tell you to forward TCP/UDP 7000-8000 or so, or worse. :D

(when it's pretty clear which ports are used if you just look at what the networking is doing when you launch the thing)

Still, most game servers require at least 2 ports, and RCON adds a 3rd, so I'm not sure what "people" are grumbling about. In fact, if you enable raw UDP, it brings it up to 3 (4).

I have a whole numbering scheme for ARK, for example. :D

[rose valley]

That is my experience as well but at the end of the day, routing everything through one port is frankly an easy thing to do and it can make it easier for the non experienced admins to set up their servers.

[blazing halo]

It can, yes... unless you're running a very locked-down server, and then you're scratching your head for an hour. ;)

(until you remember to check what it's actually doing, and what the Wiki didn't mention, to find it initiating new connections to itself)

[rose valley]

Gentle reminder that the wiki is community driven. The only official documentation that we have atm is unfortunately in the patch notes.

[blazing halo]

I appreciate that the DS is experimental on top of experimental, and I thank you for providing it (the general trend today seems to be to skip it... which is unfortunate). Still, this needs to be clearly communicated as important. At the end of the day, networking is hard, and no amount of "making it easier" will ever obscure this fact completely.

[warped sapphire]

hi does someone mind helping me out.

ive been trying to get the answer from the history trying to set up a dedicated server for me and my mate , but whenever he tries to join the name keeps pending and he cant join , i checked and double checked all port forwarding in both the firewall and router all should be good.
not sure what is is keep getting im completely fine joining myself from my local ip4
and my mate tried both ip4 and public ip both same result

[blazing halo]

That screenshot doesn't really tell us much... Let's recap your efforts so far.

a) you are sure you have a public IP, correct? I assume this is IPv4 and not IPv6.

[warped sapphire]

yea i used to run a ark server so same IP

[blazing halo]

a') you are sure it hasn't changed/is a static assignment?

(it's a relatively easy check, google "what is my ip")

[warped sapphire]

yea thats what i use usually

[blazing halo]

Okay. b) What ports did you forward?

[warped sapphire]

i cant find mine if i use the public IP either ... not sure that might be normal

[blazing halo]

Connecting to your public IP from inside your LAN requires NAT loopback, and a) your router needs to support it and have it enabled, and b) Satisfactory doesn't seem to like it anyway.

[blazing halo]

Okay, that's a lot of ports. Considering that the game only really uses three...

By default, these are: 7777, 15000, 15777.

[warped sapphire]

ive tried specific ports even

but ill change that sure

[blazing halo]

They are also all UDP, so you don't need TCP for them.

Hey, while you're still in that router UI, can you verify that your IP is public? ISPs do all sorts of weird shit from time to time, like switching you to a CGNAT "transparently" and without prior notice.

Humor me please.

[warped sapphire]

mind hopping in a private conversation so i can just stream it ?

[blazing halo]

I don't, but I'd much rather do this here for everyone's benefit.

[warped sapphire]

uhm yea some stuff i probably shouldnt screenshot thats the thing 😛

[blazing halo]

You may send the screenshots via DM, I don't mind. Please go ahead.

Okay, I can already say this: for the duration of these tests, go ahead and switch your Windows firewall off.

You will reenable it afterwards; do this to remove it from the picture.

[warped sapphire]

i did

[blazing halo]

In the course of troubleshooting we have determined that @warped sapphire 's IP is not public.

(it's in the 10/8 range)

@warped sapphire please contact your ISP and request a public IP.

[plush token]

Just did a clean install of exp dedicated server on steam, I get a crash when I attempt to launch:

Verified game files, setup port forwarding, no change

[blazing halo]

What else shows up afterwards? Because that should not be causing a crash.

If you could also specify if you're running EX or EA, how much RAM the machine has, and if you're running anything else on that machine (like, the game client perhaps?) - that'd be great.

[oak crane]

I'm working on a experimental server, I can connect locally, and I'm 90% sure my port forwarding is working as I can get a connection notification in the serverlog

I've seen some forum posts stating that this error can be a result of timezone mismatch, which would make sense given that my server's timestamps are about 7 hours ahead of my timezone

and for refrence, logs against sytem time

I dont think this is a satisfactory issue, but rather a windows issue, but I'm not really sure how to fix it, ive set my timezone manually but that didnt seem to change anything

[quick mist]

The times in log filenames typically reflect UTC, so that's unlikely to be the issue. The error you posted looks more like a network issue. Check the Wiki for directions on ensuring you have the correct exceptions created in your system's firewall.

[blazing halo]

The timestamps are actually broken. As long as the PC has the correct timezone, things should be fine in that respect.

I would be interested in seeing the server logs from around the connection attempts.

[blazing halo]

Any update on the questions I posted?

[oak crane]

the serverogs dont output anythen connecting through port 15777,
Iam running the Game Port on 25565 since my modem doesnt want to forward 7777

Connecting to the server on port 25565 yeilds a spam of these

[blazing halo]

If you only intend to run one instance, you could technically run the server on port 15001/15002... ;P

(though 15010+15011 makes a bit more sense)

Define "doesn't want to forward"?

[oak crane]

it gives an error regarding conflicting with other ports, never seen something like it before tbh. and it's only with port 7777

[blazing halo]

Do you have any sort of UPnP or port triggering set up?

[oak crane]

Ooh yea it seems to be the UPnp messing with it

[blazing halo]

STEAM likes port 7777. ;]

UPnP deserves to die, I think...

"Let's make it easier for people who have no clue what they're doing to shoot themselves in the foot."

[oak crane]

lmao

[blazing halo]

It's WPS all over again.

You don't need UPnP. You don't need WPS. What you need is a solid grounding in terms of networking education. 🤣

But eh.

Incidentally, both can be used to compromise your defenses from the inside. :D

[oak crane]

so now I have ports 7777, 15777, and 15000 forwarded, the server is using those same ports, still getting the timeout error connecting with port 7777

given the default setup, which port should I be putting in to connect?

[blazing halo]

Be a darling and add -multihome=0.0.0.0 to the launch command. Let's see what happens when it's not trying to route shit via IPv6 all around itself.

[blazing halo]

The default it shows when you try - 15777.

[oak crane]

ooh, that worked

[blazing halo]

So...?

[blazing halo]

What worked?

[oak crane]

the multihome commandline argument

[blazing halo]

Yeah, okay, but what did it do?

Can you now use all the server offers "worked", or did something change "worked"?

Or now it's showing a different error "worked"?

[oak crane]

I'm able to connect to the server externally now, heres my port bindings before and after the multihome argument

[blazing halo]

Ah. So that is what happens...

I have IPv6 disabled since I'm not using it at all, so I've never seen that before.

In theory, the internal router thingie should work with just port 15777 forwarded.

As to why it didn't want to in your case... no clue.

@rose valley ^ a wonderful example of the internal router in action ;-) I assume it's a configuration issue, but it still looks weird as heck. Why would it spam those NotifyAcceptingConnection messages?

[frosty gale]

Mods working yet for servers? just curious

[loud minnow]

no

[mental locust]

@frosty gale never

[frosty gale]

[blazing halo]

To clarify: the work to make mods compatible is ongoing, but it's all volunteer work so it's slow-going. It'll happen Eventually™.

[frosty gale]

take yo time, was just wondering brotha!

[blazing halo]

Hey, I'm waiting too. ;-)

Just... don't hold your breath.

[frosty gale]

no promises

i tend to get my hopes up

[blazing halo]

Well, it might be potentially unhealthy... ;P

[frosty gale]

i feel the oxygen deprivation

[upper trellis]

[warped sapphire]

im getting some weird visual glitches where the conveyer belts kinda teleport recourses .. is this a common thing for dedicated servers and something i can fix ?

like this supposed to be a consistent line but its not

[quick mist]

That is a fairly common visual glitch. if you look at the throughput of the device at the end of the belt you will see the true rate of input from the belt.

[warped sapphire]

alright thx

[blazing halo]

Yeah, the glitch is purely visual. The items on the belts are populated based on predictive algorithms, so they don't always match up in position. As you can see (by items disappearing right before their destination), the game has an input priority - meaning, if it needs to have an item, and it should be available according to the hookup, it'll just take it from the belt wherever it is.

[blazing halo]

@low thorn Okay, I tcpdump-ed both sides of the NAT (since for whatever reason, TRACE doesn't like iptables-legacy...). I'm in the process of reconstructing the events, but from what I can tell at a glance, the port numbers match up on both sides of the NAT. So it looks like the traversal doesn't mangle the port numbers.

[blazing halo]

Hmmm... Say, how important for the server's auth is it that the handshake packets arrive in the correct order?

Mainly because

1661540388.666844: 10.0.1.22:15000 --> 10.0.1.1==PUBLIC-IP --> 10.0.1.2:49815, size: 325
1661540389.658790: 10.0.1.22:15000 --> 10.0.1.1==PUBLIC-IP --> 10.0.1.2:49815, size: 91
1661540389.658828: 10.0.1.22:15000 --> 10.0.1.1==PUBLIC-IP --> 10.0.1.2:49815, size: 766
1661540389.658849: 10.0.1.22:15000 --> 10.0.1.1==PUBLIC-IP --> 10.0.1.2:49815, size: 326
1661540389.658863: 10.0.1.22:15000 --> 10.0.1.1==PUBLIC-IP --> 10.0.1.2:49815, size: 84

1661540389.674754: 10.0.1.2:49815 --> PUBLIC-IP==10.0.1.1 --> 10.0.1.22:15000, size: 63
1661540389.674755: 10.0.1.2:49815 --> PUBLIC-IP==10.0.1.1 --> 10.0.1.22:15000, size: 766
1661540389.674762: 10.0.1.2:49815 --> PUBLIC-IP==10.0.1.1 --> 10.0.1.22:15000, size: 290
1661540389.674767: 10.0.1.2:49815 --> PUBLIC-IP==10.0.1.1 --> 10.0.1.22:15000, size: 19
1661540389.674778: 10.0.1.2:49815 --> PUBLIC-IP==10.0.1.1 --> 10.0.1.22:15000, size: 66

I know these are likely not the query/reply (mainly because the time order is reversed), but could this provide a clue?

Also, I am 100% sure that the NAT keeps the correct port numbers on both sides.

Now, I didn't record session IDs, so I don't know if I matched the traffic correctly (all I had was IPs and ports). All I can be reasonably sure of is these clusters, and the fact that the back-and-forth happens on the same port numbers.

The order shouldn't matter, though. It could end up being messed up each time the communication passes through the same NAT in both directions; packets could be held up and their order could change at any moment.

I honestly hope it doesn't. :D I mean, it's UDP. Just because they left the server (or client) in a specific order, doesn't mean they'll arrive in that order. Relying on timing would introduce unpredictable race conditions, unless they also carry either timestamps or IDs (or sequence numbers, if my earlier detective work trying to crack these packets is any indication...).

I suppose I can't really debug this any further without having knowledge of the protocol (and the requisite keys, if any).

[frosty gale]

@blazing halo you are an absolute gigachad

with how hard you work in here

[mental locust]

@frosty gale gigachad... lame

[maiden roost]

Is there any advantage to running a dedicated server if I'm playing solo? Like will my production continue if I'm offline?

[mental locust]

Well yes...

If you set the server to run with you logged out

But why....

[blazing halo]

Hey, I play that way. :D

Don't bash it! ;D

I have limited time, and I don't always want to set up continuous production. Sometimes I just want to set something up, go to sleep, and then have a full container next evening.

[blazing halo]

Admittedly, all that shit (by me) is about solving a problem I personally have, and not many people are likely to encounter. But sure. 🤣

[hidden smelt]

I've just moved a save from a U5-DS to U6 Local and am getting a bug with empty train platforms, there isn't any track on the 'empty' platform and when I try to dismantle it the game crashes

Is this something to do with the save originally being hosted on a U5 server and is there something I can do without using save tools to replace the broken items?
Edit: I can add track to where its supposed to be and trains will traverse/power transmit

[blazing halo]

Hmm... personally, no clue. Might qualify as a general game bug.

Try this: load the EA save in local EA, save under a different name, and then load that save in EX local.

There should be no difference between local and server in terms of save output, but there may be additional checks if the game is saved in single-player.

Worth a try, anyway.

@hidden smelt

At worst, you'll confirm that the bug happens in single-player too, which would qualify it as a general bug.

And if that doesn't work - can you load the EA save and remove the offending structures before moving to EX?

[hidden smelt]

Ok, will try that and see what happens, at least for now being able to place normal track and complete the loops will get things working and hopefully it gets fixed by the time I want to dismantle an empty platform haha

[hidden smelt]

I could try that too, its a 300h save and there will be a few though

[blazing halo]

If you load it into EX as-is and then proceed to add stuff, it won't be easy for you to go back to that EA save afterwards. So... don't delay. ;)

And please do that check I mentioned before, okay? It would answer a fairly important question.

(is it just server-related, or does the bug happen on single-player/in the client as well?)

[hidden smelt]

gotcha, luckily the first order of business was to figure out why power was out and trains were all buggin'

[blazing halo]

True that...

[hidden smelt]

Game loaded in EA-SP with most recent Autosave
traversing to suspected problem location...

[blazing halo]

Heheh... Transit in progress...

[hidden smelt]

Sorry for the delay
weird, platform is broken in U5-SP but not in U5-DS, U5-SP also terminates when dismantling it too

Looks like I'll have to use save tools to remove the offending platform then rebuild it

[blazing halo]

Can you remove it on EA-DS?

It's entirely possible it'll bug out there, too... which would indicate that it's just plain bugged. ;]

[hidden smelt]

Its functional as a train tract/station section but the DS crashes when I try to dismantle the platform

[blazing halo]

So it crashes on EA-DS, it crashes on EA-local, it crashes on EX-local...

... so basically, that platform is plain fucked regardless of env.

[hidden smelt]

Seems so, how odd XD

[blazing halo]

Happens. :D

Hey, with how complex this game is, I'm actually surprised this shit doesn't happen more often really... 🤣

[mental locust]

[blazing halo]

So less "there are bugs left, the devs are shit", but rather more "there are so few bugs left, the devs are gods". 😆

[mental locust]

@blazing halo

[hidden smelt]

Honestly, same. The overall stability of the game is downright fantastic, I keep telling people that the 'buggy' experimental is more stable than most AAA full Day-One releases

[blazing halo]

Certainly fewer issues than you'd normally expect from an EA game... Or perhaps that's the EA games I've seen. xD

[blazing halo]

"full" :D

[hidden smelt]

Especially EA Games games XD - Early Access titles too

[blazing halo]

Used to be games got a full release, then DLC, then DLC, then content pack...

Now you make a full game, cut 1/3rd off, and chop that into DLCs... xD

Or, it feels that way.

[blazing halo]

#shotsfired

[hidden smelt]

it really does, not a fan of that whole theme myself

[blazing halo]

But honestly, with how every new release by EA Games is essentially the same old, you'd think they would've worked that shit out by now.

[hidden smelt]

Lets be honest, they won "Worst company in America" for a reason....

[blazing halo]

They say good software takes ten years. Well, EA Games sure has a stable codebase at this point, right?

[hidden smelt]

Truuueeee XD

[blazing halo]

Certainly with their sports games, anyway.

And yes, I say "games", plural, because they actually do have multiple titles.

I count FIFA as one game, essentially. xD

It's less "buying a new game" and more "piecemeal subscription model" to be honest.

[hidden smelt]

Oh and don't forget you have to re-purchase your team each year with the same abysmal 'chance' rates even though its the same damn game (almost)

[blazing halo]

Contrast Satisfactory: if you want to, you can load a save up in SCIM and unlock everything. There's even a handy "unlock all" option.

I know it's not exactly the same, but... xD

Well, I suppose you can't unlock Phase completion that way.

But in theory, you don't have to. It'll nag you to complete the phases, sure, but with all tech unlocked - you'll be able to build everything the game offers.

All you'll "miss out" on is the golden cup.

Well, whoopee. Form validation passes. Now on to submitting.

[hidden smelt]

100% man, OG style "you get everything in the tin" games are by far my preferred options, I've been loving the indie scene pretty much since the time Satisfactory came to EGS

Just finished checking the whole rail system for faults, only 1 other platform had that same weird bug - I could see it from the tool as there was a break in the rail line but there was a 'platform' designated

[frosty gale]

Shit on my parade pal

[mental locust]

@frosty gale

[plush token]

Running EX branch, the machine has 16gb of RAM and is running nothing else. It's a bit tough to catch anything else, the window closes itself after scrolling a lot of text. This was just the first block of red text

after reading a bit more about dedi servers we decided we'd just keep going with a non dedicated one, sounds like they will get tuned up later maybe as part of 1.0

[blazing halo]

Holy shit, you live.

🤣

[plush token]

lol, yeah sorry for the wait

[blazing halo]

Anyway, the server should actually write to a log file.

But if you don't want to bother with that, start cmd and launch it via CLI. The CMD window should remain and let you scroll at your leisure.

[plush token]

where's the log file for windows installs? can't find any reference on https://satisfactory.fandom.com/wiki/Dedicated_servers

[blazing halo]

Try in the server folder.

[reef gull]

Correct me if I am wrong, but I thought it is in the appdata folder of the account running the server application similar to the client version

[plush token]

ah I think i found them, under factorygame/saved/logs

[reef gull]

in the install?

or in the appdata?

[plush token]

in the install

[reef gull]

okay, gotcha

thanks for the notes

[plush token]

yeah no prob, like i said we kinda gave up on dedicated for now but i'm still curious why a clean install would fail

if anyone wants to look, here's the crash log

[reef gull]

[2022.08.25-18.19.06:155][  0]LogNet: Created socket for bind address: :: on port 15000
[2022.08.25-18.19.06:155][  0]PacketHandlerLog: Loaded PacketHandler component: DTLSHandlerComponent ()
[2022.08.25-18.19.06:155][  0]PacketHandlerLog: Loaded PacketHandler component: Engine.EngineHandlerComponentFactory (StatelessConnectHandlerComponent)
[2022.08.25-18.19.06:155][  0]LogNet: IpNetDriver_2147482526 IpNetDriver_2147482526 IpNetDriver listening on port 15001
[2022.08.25-18.19.06:156][  0]LogLoad: Took 0.050014 seconds to LoadMap(/Game/FactoryGame/Map/DedicatedserverEntry)
[2022.08.25-18.19.06:156][  0]LogInit: Display: Engine is initialized. Leaving FEngineLoop::Init()
[2022.08.25-18.19.06:156][  0]LogLoad: (Engine Initialization) Total time: 3.21 seconds
[2022.08.25-18.19.06:156][  0]LogCore: Engine exit requested (reason: EngineExit() was called; note: exit was already requested)

I haven't gotten info SF Server logs yet, but this is interesting... got a second instance running?

[2022.08.25-18.19.06:098][  0]LogServer: Bound Server Query Socket to address '[::]:15777'
[2022.08.25-18.19.06:098][  0]LogServer: Bound Server Query Socket to address '0.0.0.0:15777'
[2022.08.25-18.19.06:098][  0]LogServer: Error: Failed to initialize server. Could not bind any addresses to port '15777'
[2022.08.25-18.19.06:098][  0]LogWindows: FPlatformMisc::RequestExit(0)

[plush token]

Nope, I just attempted to run from the steam GUI

[blazing halo]

Yup, that's the issue --> LogServer: Error: Failed to initialize server. Could not bind any addresses to port '15777'

[plush token]

did I not open the port properly?

[blazing halo]

Most likely didn't get permission to.

Funny thing is, it reports that it bound it to both 0.0.0.0 and ::

which is obviously bullshit.

Those lines should most likely read "Attempting to bind..."

[reef gull]

heh... forgot about the permission thing... haha, usually not what the problem is when I get that, but you wouldn't be wrong

[blazing halo]

But yeah, it could also be because 15777 was unavailable.

Doesn't the Windows version of the DS just kill the window while keeping the server running?

So if you launch it, and the window disappears... you think it crashed, launch it again, and then it craps out (because a server is already running on those ports).

[reef gull]

idk... last time I played with DS was with some modding testing, I haven't touched them since

[blazing halo]

I haven't tested the server on Windows, ever.

[reef gull]

I think I did Windows because of the multicore thing... and RemoteMonitoring can eat resources on REALLY REALLY LARGE saves

[blazing halo]

@plush token can you verify that the server actually stays running when you launch it, even if the console window goes the fuck away?

If it does, that's another pebble in the "idiotic dev choices" pool...

[plush token]

I can, but probably not tonight... i made the mistake of starting windows patching on that machine

[loud minnow]

You can see it on task manager if the window goes away and just end the task

[blazing halo]

Yeah, pretty much. But you have to know it's happening, or at least suspect.

[reef gull]

reboot would clear any lingering bindings...

[blazing halo]

To any uninitiated gamer, this would look like a crash,

and every subsequent launch attempt would crash,

further reinforcing the idea.

If it's by design (and I seem to recall that -log is implied on Linux only, meaning it might not keep the window around on Windows), then it's yet another suspiciously thoughtless design choice which was probably aimed at making people's lives easier, but backfired...

IF.

[loud minnow]

It doesn’t you need to add -log

I run it on windows

[blazing halo]

Does the Windows flavour of the DS just automatically background and kill the window, while continuing to run?

[loud minnow]

Yeah

[blazing halo]

WHAT

THE FUCK

DEVS?!

[loud minnow]

Yeah it will just close it and run in the background you need to add -log to keep the window open

[reef gull]

Question now is that a UE choice or a CSS choice...

[blazing halo]

@plush token so your crash is being caused by the server pulling a Copperfield on ya...

[loud minnow]

Lol

[blazing halo]

I dunno man... If I saw my code doing that,

my first priority would be to make sure it doesn't.

As a programmer, you're supposed to have this skill

this pessimistic outlook on life

trained, of course, to be worth shit and not just be general pessimism.

You're supposed to look at how thing X may fuck up.

• and then write code to prevent it.

That's kinda the point, or one of the points. You're supposed to think ahead and consider how feature X will be mishandled, or how it may be misunderstood.

[loud minnow]

well it is on the wiki

[blazing halo]

Yeah

but actually no.

It says "forces the server to display LOGS"

not "prevents the server from terminating the console window".

Very different things.

[plush token]

odd that it's default for linux but not windows

[loud minnow]

yeah so if you dont add -log it will display the window for a second then closes but server still runs in the background

[blazing halo]

Probably because on Linux, the server has no authority to kill the console.

On Windows, it's as simple as freeing the handle and telling the system that you did.

It helps that on Windows, a console window is basically a window like any other window.

On Linux, it's a connection to an existing terminal (tty) that doesn't go the fuck away when you disconnect.

[plush token]

This blurb on the wiki lead me to believe the window should stay open:
Manual - Steam GUI
Simply launch the Dedicated Server from Steam just like any other program. You will be able to find it in the Tools category rather than games. It will open a terminal window in which you can review the logs in real time. Do not be alarmed if the window stops showing information after a while. This usually means the server is up, running, and waiting for connections.

i'll test this out sometime though, to see if the server is actually up after a first run

thanks for taking a look fellas

[loud minnow]

well i dont use the steam GUI to launch it

[blazing halo]

STEAM should allow you to add command line switches, so go ahead and add -log to it.

[loud minnow]

never tried it like that

[blazing halo]

Yes, because people that have hosted a server before know to launch it in an existing cmd process, so that logs don't go into the never-never when it dies.

[blazing halo]

Sure thing.

I feel like strangling a dev. :D

[reef gull]

Glad to help

[reef gull]

SA Reporting for Duty! salutes

Haha, 🤣

[blazing halo]

I mean come on. The existence of a -log switch implies that it's possible to not have the window freed.

So... why wouldn't they... make it the default behaviour...

I just... I can't... what?

It's really high up there, right next to ports that self-increment without telling anybody.

Because port forwards work dynamically, right guise?

_<

arrrrrrrrrrrrgh.

[reef gull]

hands Kad a beer

[blazing halo]

I hope it's a proper beer, yeah?

Not the pale piss people seem to be in love with.

You know, as it happens...

[reef gull]

Haha, I prefer microbreweries personally... but whatever flavors your fancy, 😅

[blazing halo]

Yeah, currently drinking a beer from a nearby "craft brewery" or whatever they call themselves.

A peated RIS, apparently.

Whatever that means.

[lime seal]

I'm looking for through the files and folder for dedicated wondering about other things myself, like how one might load up custom code to the server, aka mods but there isn't even a mod folder, so it looks like we are waiting for some very basic support on that one unless I'm really missing something

[reef gull]

mod support is a WIP atm, and I would suggest checking with the modding discord for updates

[lime seal]

alternatively reversing engineering the base dll files and recompiling the dll that i know are being loaded, but i don't think thats the best idea hmmm

[drifting jay]

modding discord 💀💀

[lime seal]

exactly💀

its like suggesting, here is a lake of fire, go jump in💀

but any rate lack of mod folder does indicate the css didn't provide a basic level of support for mods so I don't think its possible with some reverse engineering and injecting it into the existing code base's that are already loading hmmmm

or maybe a slight modification to them to force the server to look at a mod folder and load up those code bases also might work, essentially putting in a hook functionality sense it doesn't really look like that exists

[blazing halo]

The server supports mod loading 100%.

It's the mods that don't support the server.

@lime seal

I know it sounds idiotic, but the truth is - the server can load the same mods that the client can (you have to make the /mods directory yourself though). But the one mod to rule them all, the so-called mod loader, which every other mod depends on and will not work without, was built before the server was a thing.

Until they can make the "mod loader" compatible with the server (and provide tooling to build binaries for Linux, because that's missing too AFAIK), no mod will load on the server.

The guy is working on it, but it's just one guy.

[oblique ice]

Beer aged in peat maybe? In the barrels obviously.

[lime seal]

So your saying one has to make the mods folder in the factorygame folder as it would be for the loader? Very interesting it isn't obvious that is the case.

[loud minnow]

Yeah but it still wouldn’t work… because the “mod loaders” doesn’t support dedicated server that’s is what he is saying @lime seal

[lime seal]

It would if I made a loader that worked with server, this is a engineering problem after all😉

and as everyone engineering have solutions, no matter how complex or simple

[loud minnow]

go for it.. knock yourself out my man @lime seal

[graceful badge]

Setup an experimental U6 dedicated server to play with some friends (4-5 of us max).

Running it via steam.
Specs of the server are pretty high. Game runs very smooth.

But we’re noticing items on belts are getting stuck visually. Doesn’t actually cause any production issues.

Has anyone else seen this? Any tweaks that can be made to resolve it?

[oblique ice]

I've heard of that bug mentioned here in the server and in the QA forum, so, it's not just you. No idea about resolving.

[crystal eagle]

The belts thingy on dedi server have been like that since the they introduced servers, still think it’s a sync error somewhere

[blazing halo]

Of course it is.

And I doubt it'll ever be resolved.

Let's see... 30 extractors pushing ores through an average of 5 full-length belts (let's say each belt holds 30? items per length post-to-post, so 150 items) would demand there be 4500 ore units on the belts at all times (when full). Depending on how often the server needs to send out position updates, per 4 players, that might be anything between 18 000 and say 300 000 updates/second.

If the server did naive updates, that's the sort of rate it'd need to be pushing to ensure all players have the same state at all times.

Even if we assume that the most economical way to send that update would be to give the belt index (let's say 4 bytes) and position on said belt (1 byte), that's 5 bytes per item, or between 88 KiB and 1.5 MiB - per second. The nice thing about doing it this way is that it would only scale with count; it would always be 5 bytes per item. xD

Just for the belts. :D

Now obviously, that's bullshit. The game isn't doing this in a dumb way. I don't know how it does it exactly, but I expect it sends out bulk updates with some deviations (like if there's an item missing), perhaps with some sort of encoding on item types or so (there must be a finite number of item combinations you could send down a belt), and the rest is likely done by predictive algorithms which assume some things, extrapolate over time, and then get corrected every so often.

[slate hound]

this was covered in a dev video at one time

[blazing halo]

What did they say? I haven't seen all of those. Link me up?

[plush token]

I think I have a server up and running now; how do I connect to it from a second machine? What IP would I use?

my 192.168.x.x local address?

[slate hound]

essentially its an array per belt, they dont track exact positioning and i dont remember which video it was discussed in, but it tracks the data as an array over time, so displayed items are just a visual of whats in the array roughly.

[graceful badge]

If your on the same network, then yes, the local IP. If your not on the same network, then use your public IP

[graceful badge]

Thanks for that detailed info

I’d love to install the mod for covered belts, it would solve everything 🙂 for now, we can live with it

[blazing halo]

If from the local network, whatever local address you're using. If you're using 192.168/16, then I guess so?

[frosty gale]

The goat @blazing halo back at work I see 😤

Just a curious question how many hours do you have on satisfactory 😭😭

[blazing halo]

... goat? 🤔

And 435, so not that many.

[plush token]

Hmm, still can't connect to this thing... this is a screenshot from google fiber port forwarding, pretty sure i got that part right:

assuming my dedi server has IP ending in .235

which it does according to ipconfig

And here's my in game server when I try to connect to the above IP address at port 15777

Last screenshot, the latest from the server window:

[blazing halo]

Uhhh... you are in the same network as your server, right?

[plush token]

yeah, both plugged in via ethernet

[blazing halo]

You're port-forwarding, so I assume you want people from the Internet to reach this server, yes?

[plush token]

yeah

[blazing halo]

Right.

[plush token]

though right now I would accept if just I could, as a small victory

[blazing halo]

So, is that IP the reserved DHCP static lease assignment for the server machine?

[frosty gale]

“Greatest of all time”

Bro 435 hours and you know all of this

[plush token]

I wouldn't say static, but it is the current IP:

[frosty gale]

Jesus

[blazing halo]

I'm not sure I'm comfortable with either. ;]

[blazing halo]

And what will happen when that IP changes?

[plush token]

I can't remember what I did 1-2years ago, but that IP persists on reboot

[blazing halo]

You may want to make sure that it's assigned as a static lease to that machine. Otherwise your port forwards will stop working when it changes.

Moving on.

Yes, you should be using IP 192.168.1.235, port 15777 to connect to your server.

If you can't, it means there's something blocking you.

Have you disabled the firewall on the server to test this?

[plush token]

oh, let me check

looks like it's off

i've used this machine to host other dedi servers, so that makes sense

[blazing halo]

Right.

What does Satisfactory say when you connect to the server?

[plush token]

after about minute of attempts, a time out error

[blazing halo]

I'm a programmer by trade, and a network engineer by hobby. ;P I've also done modding for UE before, and run multiple UE gameservers besides. I wouldn't say I know them "inside and out", that's silly, but let's just say I've seen shit.

[plush token]

that's a good combo

[blazing halo]

Right. I assume you can ping the machine, and talk to it in other ways?

Please verify for me that you used the correct IP - remove the address from the list and add it again.

I would also like to be sure that the server is actually running those ports you specified.

[plush token]

I can ping it from the machine I want to play on

and yeah, I did remove it and re-add it a couple times. Seems really weird that the server manager masks the IP while i'm typing it...

[blazing halo]

You can click 'Show'.

[plush token]

I know, just ... what a weird thing to code in

[blazing halo]

It's there for streamers.

[plush token]

oooh

that makes sense... i was like that took X hours they must have had a reason

[blazing halo]

Right. So dig up the latest FactoryGame.log on the server machine (should be somewhere in the install directory), and read through it.

Find the lines where it says that it bound to ports. Show 'em.

[plush token]

and another similar line for 15000