Here's a couple typical traces to rt1 and rt2...
RT1...
Host Name IP Address Hop Ping Time Ping Avg % Loss Pkts r/s Ping best/worst
* Unknown Host * xxx.xxx.xxx.xxx 1 40ms 39ms 0% 90 / 90 10ms / 230ms
rc2so-ge11-0.cg.shawcable.net 24.64.0.8 2 20ms 23ms 0% 90 / 90 0ms / 131ms
rc2ch-atm8-0-2.il.shawcable.net 204.209.214.150 3 40ms 54ms 0% 90 / 90 40ms / 150ms
ge-3-16.a01.chcgil01.us.ra.verio.net 129.250.223.53 4 50ms 66ms 0% 90 / 90 40ms / 241ms
ge-1-0-0.r02.chcgil01.us.bb.verio.net 129.250.27.29 5 40ms 57ms 0% 90 / 90 40ms / 141ms
p16-5-0-0.r01.chcgil01.us.bb.verio.net 129.250.2.220 6 100ms 60ms 0% 90 / 90 30ms / 130ms
p16-2-0-0.r01.chcgil06.us.bb.verio.net 129.250.5.70 7 80ms 58ms 0% 90 / 90 30ms / 130ms
p16-3-0-0.r01.dllstx01.us.bb.verio.net 129.250.5.85 8 71ms 82ms 0% 90 / 90 60ms / 170ms
ge-1-0-0.a10.dllstx01.us.ra.verio.net 129.250.28.164 9 120ms 86ms 0% 90 / 90 60ms / 171ms
fa-8-0-0.a09.dllstx01.us.ra.verio.net 129.250.28.170 10 60ms 96ms 0% 90 / 90 60ms / 340ms
rt1.applink.net 206.50.88.40 11 120ms 103ms 0% 90 / 90 70ms / 191ms
...and RT2
Host Name IP Address Hop Ping Time Ping Avg % Loss Pkts r/s Ping best/worst
* Unknown Host * xxx.xxx.xxx.xxx 1 40ms 33ms 0% 71 / 71 10ms / 110ms
rc2so-ge11-0.cg.shawcable.net 24.64.0.8 2 20ms 23ms 1% 70 / 71 0ms / 120ms
VLN.Serialx-x-x.GWx.POPx.ALTER.NET 157.130.157.41 3 20ms 22ms 0% 71 / 71 10ms / 100ms
116.at-6-0-0.XR2.CAL1.ALTER.NET 152.63.137.158 4 20ms 20ms 0% 71 / 71 0ms / 90ms
192.at-1-0-0.TL2.CAL1.ALTER.NET 152.63.136.82 5 20ms 19ms 0% 71 / 71 0ms / 80ms
0.so-4-0-0.TL2.DFW9.ALTER.NET 152.63.0.182 6 80ms 77ms 0% 71 / 71 60ms / 150ms
0.so-7-0-0.XL2.DFW9.ALTER.NET 152.63.2.182 7 70ms 78ms 0% 71 / 71 60ms / 170ms
0.so-7-0-0.XR2.DFW9.ALTER.NET 152.63.102.2 8 70ms 80ms 0% 70 / 70 60ms / 140ms
184.ATM7-0.GW5.DFW9.ALTER.NET 152.63.100.45 9 70ms 83ms 0% 70 / 70 60ms / 130ms
savvis-dfw-gw.customer.ALTER.NET 157.130.155.234 10 71ms 87ms 1% 69 / 70 60ms / 150ms
savvis-internet.cntx04.usdlls2-bsn.savvis.net 64.241.96.94 11 110ms 91ms 0% 70 / 70 60ms / 220ms
* Unknown Host * 0.0.0.0 12 0ms
Only real difference between the two is RT2 doesn't reply to ICMP (what's the main peer point clients come in over anyway, RT1 or 2?). I have no idea what the setup is or how the code forces these pesky switches over to TCP. One would hope that is not an issue although anything is possibile. Nevertheless, both paths are fairly decent but I think its a given that the route across the Net isn't the cause. Something is triggering the push over to TCP at the point of hitting the arena server itself or coming into the local network. The problems with warps is of course that TCP is a poor gaming protocol and will cause problems for anyone coming in over anything less than a stellar connection. Stop that forced switch and make sure UDP remains the protocol being used, and you fix your problems.
Since this all started when the Stalingrad terrain was put in place, what happens when you switched the terrain back to the previous terrain? Does the problem still exist? If you haven't tried this I would suggest you do. I have no idea how or why the terrain would force you to TCP, but clearly that switch is triggered in the game code somewhere, so who knows by what. How or why a map would do it, I dunno. It certainly should be looked at if it hasn't already though, for nothing else just to eliminate a piece of the puzzle.
Also what are the packet sniffs saying? Someone should be doing these with the arena entry session. You should at least see from those what event is triggering the switch to TCP. That might give some insight.
Is the server side setup identical to the last session? Or were changes made there recently?
Vortex
