TF2 Client Timeouts

[Zennik]

We're having a strange issue with clients randomly timing out on our TF2 Servers en masse. Up to 80% of them at once.

A few things to point out:

1. We have 3 CS Servers and a DoD server that do NOT have this issue. DoD server even runs on the same server as 2 of our TF2 Servers.

2. We're running the 32-man bypass file, but noone else seems to have
trouble using the bypass. We also run srcdsfpsboost. Running the servers at 66 tick. fpsmax of 600.

3. The servers are Core2 E6600s with 2gb of RAM. Not a server load issue.

4. Bandwidth/network connectivity: It's in a datacenter, and we've done basic network troubleshooting. That combined with the lack of issues with CS/DoD leads me to rule this out.

5. Windows 2003 Web Edition 32-bit

Now, my question is, has anyone else come across this issue? The server CPU/Memory/Disk Access doesn't spike when it happens, and traceroutes/pings show no issues between the server and affected clients when it happens.

Also, some people just aren't affected at all. My desktop, hooked up to Comcast Cable, never gets booted. I run default settings, no tweaked console variables. 1024x768 in a window. Running XP Pro... an IDENTICAL computer, same settings etc, but hooked up to a QWest DSL, gets booted everytime it happens.

Now while given that info, signs point to it being some sort of issue between the server and the client... During this entire period of time, the client that timed out from the server was able to maintain a steady ping and an active RDP session to the server, as it timed him out from TF2.

So, all that crap layed out. What could possibly be causing this? Someone suggested the 32 man bypass might have done it, but we dropped the bypass and moved it down to 24 and that didn't help.

Could it be just some bizarre combination of settings we're using that are just causing the engine to be unstable for most clients?

Anyway, here's our commandline:

srcds.exe -console -game tf +map MAPHERE +maxplayers 32 +fps_max 600 -tickrate 66 +ip IPHERE:27015

And the server.cfg:

//START CONFIG
// server logging
log on
sv_logbans 1
sv_logecho 0
sv_logfile 0
sv_log_onefile 0
mp_logdetail 0
mp_logmessages 0
writeid
writeip

sv_cheats 0
sv_maxunlag .5

//Ensure that sv_unlag 1 works appropriately in TF2. Consider in event of lag.
sv_unlag 1
sv_region 1


// round specific cvars
mp_timelimit 0
mp_winlimit 5


//server cvars
mp_autoteambalance 1
host_framerate 0
mp_allowspectators 1
mp_autocrosshair 0
mp_autokick 0
mp_chattime 10
mp_fadetoblack 0
mp_falldamage 1
mp_flashlight 1
mp_footsteps 1
mp_forcecamera 1
mp_friendlyfire 0
mp_limitteams 2
mp_tkpunish 0
setpause 0
sv_allowdownload 1
sv_allowupload 0
sv_alltalk 0
sv_cheats 0
sv_consistency 0
sv_pausable 0
sv_restartround 0
sv_secure 1
sv_stopspeed 75
sv_timeout 60
sv_voiceenable 1
sv_voicequality 5
tv_enable 0
tv_delay 90

//Server Name
hostname "TESTSERVER"
sv_contact email@address.com

// Enter custom config data here
rcon_password "changeme"

//STOP spikes
decalfrequency 1550
decalfrequency 10


//Client Rate Tweaks
sv_client_cmdrate_difference 20
sv_client_predict -1
sv_client_interpolate -1
sv_client_min_interp_ratio 1
sv_client_max_interp_ratio 5
sv_mincmdrate 10
sv_maxcmdrate 101
sv_minupdaterate 10
sv_maxupdaterate 101
sv_maxrate 30000
sv_minrate 3500

//END CONFIG


I hope I didn't leave anything out.

I think it's a config issue somewhere, either on the server or in my actual SRCDS setup.

We've essentially eliminated the possiblity of hardware or network problems.

Help would be much appreciated.

[Zennik]

bumpbump

[Drocona]

Hmm I'm not sure what could be causing it... mostly it's because of a bad link but the other servers run fine =/