Factorio Forums

For the purposes of this bug report, the word 'server' may be interpreted as 'headless server'.

Back in 13.8, a bit more than a month ago, a modded server with 41 mods was hosted for a few people and it had the problem of more than 3 players not being able to connect without causing the server to lock up. I decided to see if I could get more than 3 players to connect to a heavily modded 13.17 windows server on my local computer and got 6 duplicate clients connected fine, so I contacted the old server host and asked if we could do some testing. The modpack I am using has 59 mods, and if more than 55 are enabled, I kept getting the "couldn't establish network communication with server" error when trying to connect to his server, which he later mentioned was running on Linux. It doesn't matter which 55, just that fewer than 56/57 were enabled. When the 55 mods were enabled, the 3 player limitation was still in effect.

In summary: A client with more than 55 mods enabled cannot see the equivalent Linux server, and also more than 3 clients cannot join that same Linux server. If more than 3 clients try to join said Linux server, the fourth client causes the server to stop responding.
Here is a link to the modpack. It currently has 4 (semi-random) mods disabled, the amount necessary to see and connect to the Linux server.
I could get 6 duplicate clients with the full 59 mods connected just fine to my local windows server, and said server host could also connect to my local windows server for a total of 7 clients + 1 headless.

Also, the exact number of mods may be 55 or 56, but 57-59 kept getting the "couldn't establish network communication with server" error, for said Linux server.

I am in central US and said server host is in Europe, but I don't think that's relevant for this bug, since 6+him could connect to my windows server but him+3 of my duplicate clients caused his Linux server to lock up.

Please post the log of the server, if possible of a client as well.

Loewchen wrote:Please post the log of the server, if possible of a client as well.

When the server stops responding, the clients do as well; I've just been force closing them. I will contact the server owner and will see if we can do a more prolonged test to see if the clients ever start responding again, and get the server log. There are no client logs of this that I know of, since the clients do not crash.

What about the 56+ mod thing?

Ambaire wrote:When the server stops responding, the clients do as well; I've just been force closing them. I will contact the server owner and will see if we can do a more prolonged test to see if the clients ever start responding again, and get the server log. There are no client logs of this that I know of, since the clients do not crash.

The log is always created, on the server as well as on the clients.

Loewchen wrote:

Ambaire wrote:When the server stops responding, the clients do as well; I've just been force closing them. I will contact the server owner and will see if we can do a more prolonged test to see if the clients ever start responding again, and get the server log. There are no client logs of this that I know of, since the clients do not crash.

The log is always created, on the server as well as on the clients.

Understood, found where the logs are stored. It may be a few hours before I post them since I need to do the test over again and the person who was hosting the server isn't available right now.

Loewchen wrote:

Ambaire wrote:When the server stops responding, the clients do as well; I've just been force closing them. I will contact the server owner and will see if we can do a more prolonged test to see if the clients ever start responding again, and get the server log. There are no client logs of this that I know of, since the clients do not crash.

The log is always created, on the server as well as on the clients.

Alright, I have some logs for you. Surprisingly, this time, it hung after the 5th client tried to join, not the 4th. These were created using his server and 5 cloned clients running on my local computer, with authentication off. I have a log file for each client and for the server. The clients joined in order from 1 to 5, waiting until each client was fully loaded and active on the server before the next one attempted to log in, with player names test001 through test005. Link to zip file with the server log and the 5 client logs.

Edit: And a pastebin link to the server log.

The clients were non responsive to attempts to move around or exit the game, with the menu greying out when an attempt was made to exit. The map could be accessed normally and the game could be saved to a file. The fifth client hung on the Factorio background screen. Imgur album below of the 4th and 5th client screens.
http://imgur.com/a/rY0If
Also, even after the server process was killed, the clients remained unresponsive.

Edit2: When I say that a client with 56+ mods cannot see Linux server, I mean that with 55 mods it was telling me I didn't have x mods that were on the server, with a random extra enabled, it went straight to 'cannot communicate with server'.

Were any of you using IPv6 to connect?

HanziQ wrote:Were any of you using IPv6 to connect?

Not that I'm aware of. I was just typing a standard x.x.x.x:x ipv4 address.

i can confirm that the bug persists as of v 14.2 headless linux edition.

Regallion wrote:i can confirm that the bug persists as of v 14.2 headless linux edition.

14.2? Are you sure?

Naturally i meant 14.1, that was a bit of a typo.
Notably the server talked about in this report is mine.

This should be fixed in the 0.14.5