Page MenuHomeFeedback Tracker

Insane bandwith usage / Server chokes
Closed, ResolvedPublic


Since the 1.42 update arma3server.exe uses an insane amount of bandwith.
Where 70 players would use about 10-15Mbit on 1.40 they now use up to 30-40Mbit on 1.42.
And after about 30 minutes after startup the server chokes and sometimes crashes.

We run A3Wasteland. {F25988} {F25989} {F25990} {F25991} {F25992}


Legacy ID
Unable To Duplicate
Dedicated Server
Steps To Reproduce

See logs and screenshots.

Additional Information

Getting this on both my servers. Different hardware, different connection, different os, different max amount of players.

Event Timeline

loudnl edited Steps To Reproduce. (Show Details)Apr 8 2015, 10:06 PM
loudnl edited Additional Information. (Show Details)
loudnl set Category to Dedicated Server.
loudnl set Reproducibility to Random.
loudnl set Severity to None.
loudnl set Resolution to Unable To Duplicate.
loudnl set Legacy ID to 2563283538.May 8 2016, 11:51 AM
loudnl edited a custom field.
loudnl added a subscriber: loudnl.May 8 2016, 11:51 AM

Also server bandwith usage is through the roof. More then doubled from before the update. Usually had around 10-15Mbit upstream, it's now up to 30Mbit.

It seems that both my servers are having network issues since 1.42. This did never happen before this update. Both servers are on different internet connections, have different hardware, different os etc.

Added more logs from both servers.

Adam added a comment.Apr 9 2015, 9:02 AM

You are running mods which could be the source of your issues. Please try reproducing the issue on unmodded version of the game and if the server crashes attach crashdumps found in C:\Users\<Name>\AppData\Local\Arma 3\

(crashdump is rpt + bidmp + mdmp file with same name).

loudnl added a comment.Apr 9 2015, 9:07 AM

Could you point me to which mods? As I'm sure I don't run any.

loudnl added a comment.Apr 9 2015, 9:08 AM

Or do you mean the client side mods which the server loads the bikeys for? e.g. CBA, JSRS & Blastcore?

Adam added a comment.Apr 9 2015, 9:22 AM

Yes, the client side mods. Thank you.

loudnl added a comment.Apr 9 2015, 5:22 PM

Running without any entry in basic.cfg makes no difference:

Running without client side mods makes no difference:

Bandwith shoots through the roof when player amount rises until it reaches a choke point and the bandwith drops down and takes FPS/CPS with it.

Bandwith usage used to be around 10-15mbit but goes up to as high as 30-40mbit just 25 minutes after a restart:

loudnl added a comment.Apr 9 2015, 5:22 PM

Server RPT logs shows nothing significant.

loudnl added a comment.Apr 9 2015, 5:41 PM

And there goes the performance without any mods:

loudnl added a comment.Apr 9 2015, 7:49 PM

Added #captureFrame logs (with profiling branch) from both servers including ASM screenshots from the same time in logs3.7z.

loudnl added a comment.Apr 9 2015, 8:04 PM

I've changed the player slots on my servers. On one from 70 to 50 and on the other from 50 to 40. This seems to hold em steady but it's not like it should be. Bandwith stil goes bananas.

Do you have enough information now? My servers keep becoming unplayable...

There is definatly something wrong here.
In the next picture the top two are my main server on extDB and the bottom two are my test server on ProfileNamespace saving. Seeing the amount of players * the bandwith usage would give the same amount as the top one.

And this is my second server on extDB2:

So changing the following settings in basic.cfg seems to make the bandwith usage a bit better, but is still not good. This never happened before 1.42:

MinErrorToSend = 0.001;
MinErrorToSendNear = 0.01;



This is how it was before (top 2):

This is the result:

MGTDB added a subscriber: MGTDB.May 8 2016, 11:51 AM
MGTDB added a comment.Apr 12 2015, 4:03 PM

Seeing the same massive increase in bandwidth when I hit 70+ players running EpochMod, crashes a couple of times a day with syswow kernel as fault, hit 70 players, ping increases to over 1000 m/s, server cpu usage starts fluctuating between 10 and 60%, player count then drops <70 and server returns to normal

Same issue here, 70+ players running Altis Life, crashes a few times a day, server networking rockets then drops all the way down to around 2MB-10MB. Causing alot of lag for players playing, as there is no way that the server can send the amount of data needed to keep a smooth experience. When it drops back to normal, everything recovers and becomes better again. Until the issue happens once again. This is really hurting large servers and needs to be resolved as soon as possible, as players and hosts are really having issues.

I am having the exact same issue. No mods other than EXTDB.
Server runs fine, 40+ FPS for all users, then starts to drop a little... then hammers down to 1-3 fps for all players on the server. Works fine until update.

Had same problems with servers choking when running Breaking Point Mod also.

hobart added a subscriber: hobart.May 8 2016, 11:51 AM

Having the same Issue running A3 wasteland and extdb. this is not an A3 wasteland issue this is a BIS broke something within the last update issue.

Why is this still on need more info? How much more info do you need?

Fix your game stop blame the mod it is the same on altis
Stop dlc

Can we please have an update ?

I have several servers crashing multiple times a day since 1.42! this needs a urgent hotfix!!!

The S3K Wasteland server is having the same issues ever since the 1.42 patch. We typically run with 100 players on Stratis, and our bandwidth usage ranged from 15-25mbps, our CPU usage was about 10-20%.

Since update... Approximately 1-2 hours after a restart we get a spike in CPU and Bandwidth. The CPU goes to 30-100% usage and the bandwidth jumps to 40-50mbps. The error log is spammed with "Server: Network message f8817a1 is pending". It comes in waves, initially it seems to subside after 10 minutes, but every time it happens it gets worse and lasts longer until restart.

For example last night the error started at 22:58:42, and by 23:04:15 it had written that error message to the logs 219529 times. That's less than 6 minutes and makes up about 50% of the log file and the instance had been running for 5 hours at that point. Earlier in the log file there are shorter bursts of the error which seems to start about 1-2 hours after restart.

Please resolve asap.

Unfortunatly not completely solved in perf2.
Server doesn't seem to crash yet on 70 players. But fps still drops and bandwith is still high.
This is on:
MinErrorToSend = 0.0015;
MinErrorToSendNear = 0.015;

This needs to be resolved ASAP, Multiple servers are having the same issue.

Adam added a comment.Sep 7 2015, 10:19 AM

Closing, If the issue persists please create a new ticket with Repro and more information. Thanks.