[16.36] Server not responding

Anything that prevents you from playing the game properly. Do you have issues playing for the game, downloading it or successfully running it on your computer? Let us know here.
Post Reply
cakemasher
Manual Inserter
Manual Inserter
Posts: 3
Joined: Tue Apr 17, 2018 9:12 am
Contact:

[16.36] Server not responding

Post by cakemasher »

Factorio Community,

For quite some time I've enjoyed playing this game very much. Me and some friends (max. 5 clients connected to the server at the same time) eventually start playing together on a headless server I created on a server I rented somewhere. Recently I've purchased my own server, and placed it in a datacenter. The server has the following specs:
Server Specs
The server has ProxMox 5.1 installed, where I've created a virtual machine containing 4 cores, 4gb of memory and full gigabit (1000/1000) internet connection. I've installed the newest version of the Factorio headless server (0.16.36) where the previous server was running version 0.15.x. We've decided to start over, generated a new map and start playing (no mods installed, nor did we add any enemy bases). For a while now, we're getting random 'Server not responding' messages and enormous lag (for example; upon picking items up from a belt, it takes up to 2 seconds before it's picked up), while the base isn't that big yet (no robots, no trains). I've tried to do some debugging by checking firewall settings, network issues and CPU core overloading, but I couldn't find anything out of the ordinary.

The UPS is stable at 60.0. Pressing F5, is showing me some information, but the values are not really constant. The latency is around 9 - 16 ticks (sometimes raising to ~30 ticks, but just for a short while). Buffer is also not constant as it's ranging between 1/0/0 tot 6/6/2 and other different values. Honestly, I'm not really sure what the values mean that are shown in the F5 screen.

The network and firewall configurations seem to be 100% good and stable. The server is using around 10 - 15% of 1 CPU core during normal playing, and raising to around 70 - 80% while saving the map.

The headless server is running on Debian 8.10 64-bit, using 'screen' to keep the server alive when the terminal is closed.

Does anyone know something we could try to improve the 'Server not responding' message, and lag? Any help is much appreciated!

Thanks in advance.

Kind regards,
Cakemasher


cakemasher
Manual Inserter
Manual Inserter
Posts: 3
Joined: Tue Apr 17, 2018 9:12 am
Contact:

Re: [0.16.36, build 36253] Server not responding

Post by cakemasher »

Loewchen wrote:See: viewtopic.php?f=49&t=52433
I assume you mean, add the log files & save file? I've attached both to this post.
Attachments
factorio-previous.log
Server log of the previous session.
(8.87 KiB) Downloaded 69 times
factorio-current.log
Server log of the current session.
(5.13 KiB) Downloaded 79 times
servermap_save.zip
Save file of the map that has been running since the beginning of the server.
(6.88 MiB) Downloaded 46 times

Loewchen
Global Moderator
Global Moderator
Posts: 8302
Joined: Wed Jan 07, 2015 5:53 pm
Contact:

Re: [16.36] Server not responding

Post by Loewchen »

This seems to be caused by high package loss which suggests a networking issue.

cakemasher
Manual Inserter
Manual Inserter
Posts: 3
Joined: Tue Apr 17, 2018 9:12 am
Contact:

Re: [16.36] Server not responding

Post by cakemasher »

Do you maybe have some tips to check if it is due to UDP packet loss?

I've pinged to the server 1.000 times, sending 1500 bytes of data each ping. Of the 1000 ping request, it dropped only 4 and had some high latency of a couple of them, but the average remains low. See screenshot for the ping results.

Pinging is TCP of course, so I've started a UDP server on the same server as the Factorio server is running on using the command;
nc -u -l -p 1234 > /tmp/100MiB.bin

On the host, I've downloaded a 100MiB.bin file from a speedtest website, and send it to the server using the following command;
cat 100MiB.bin | nc -u [server-ip] 1234

This resulted in a data transfer of 100MiB over a UDP socket, with no data loss. The file on the senders machine is the exact same as the one that was 'uploaded' to the server.

Do you maybe know what else I can test / try or do to check if packet loss is the problem?

All help is highly appreciated! Thanks alot!

Edit #1
netstat -suna
Edit #2:
I've perfomed a 'iperf3' test, which basically opens up a UDP server, and allows a client to send data to that server for testing purpuses. Iperf3 actually does keep track of packetloss.
I have ran a test for 5 minutes, transferring data from the client to the server at the maximum speed the client can handle. This resulted in a data transfer of 3,35GB over a UDP connection, with 0 packetloss (see screenshot).
Unless there is something I'm completely overlooking, I doubt if this has anything to do with packet loss.

- Cakemasher
Attachments
Ipref3 test results.
Ipref3 test results.
ipref.png (9.52 KiB) Viewed 2051 times
Results of transferring 100MiB file over an UDP socket.
Results of transferring 100MiB file over an UDP socket.
udp-transport.png (7.67 KiB) Viewed 2056 times
Results of 1000x ping with 1500 bytes each.
Results of 1000x ping with 1500 bytes each.
ping-results.png (4.39 KiB) Viewed 2056 times

Post Reply

Return to “Technical Help”