A DC Problem I Cannot Understand Can Be Happening

[Kraggy]

Details:

This is happening on one of my two broadband connections, Virgin Media in the UK and only started around 6 weeks ago. So far Virgin have been unable to explain it.

Playing FFXI the connection fails with a POL-4001 or POL-4002 "Server reported timeout" error (other the DC is the more usual 3001/3002) around 6-7 minutes after logging in the first character.

I have three PCs on which the game is installed and I've had very, very few DCs in all the years I've been playing (12+) until in end of May or early June. It happens on all three PCs, two of which are Windows 7 and a newer one on Windows 10 Pro.

I have both Virgin Media 150Mb (Superhub 3 in Modem mode fronted by a TP-Link AC750 for the last two weeks and before that a Superhub 2 in Router mode) and BT Infinity's 80Mb service using an ASUS RT-N66U in ADSL Mode.

I never have problems with FFXI when using BT at all. Given my setup I simply change the Default Gateway on a PC to immediately swap from one ISP to the other so the only difference is the ISP/routing I'm using.

So it would appear it's a problem with my VM connection BUT I am having NBO PROBLEMS with any other game or indeed any other internet activity using Virgin, at all so it doesn't seem plausible to blame my broadband connection to the ISP. Aslo, since the problem started before the VM router was changed then it's pretty inconceivable it's a router fault.

Moreover, given I don't have this problem when using BT it doesn't seem reasonable to blame SE's servers either.

Which seems to boil down to it being a problem with Virgin or Virgin's routing which is extremely specific to FFXI's internet activity which is way beyond my knowledge to understand.

Steps:
1. Log into a character
2. Do anything at all, or do nothing, for around 5 or so minutes
3. See the red DC icon appear to the left of the nameplate
4. While this is happening the S/R indicators still show activity for about a minute before R0 appears
5. While this is happening I can use chat and do other things like Search just fine .. so not all activity has stopped between the POL.EXE and SE's servers
6. While this is happening TRACERT shows no problems reaching the IPs which Windows Resource Monitor shows POL.EXE is connected to
7. After a couple of minutes from the DC icon appearing the 4001/4002/3001 appears

Date & Time:
Any

Frequency:
100% using Virgin Media, 0% using BT

Platform:
Windows

PC Manufacturer & Model Number:
Generic custom configured PCs using I7s and ASUS or Gigabyte boards.

OS:
Windows 7 and 10

Graphics Card:
Nvidia GTX980 and GTX1070

ISP:
Virgin Media

Type of Internet Connection:
Cable

[Sirmarki]

Virgin were recently on Watchdog, and are having big problems with over utilisation (basically they have over sold in areas).

I remember an issue with BT a few years ago that only seemed to affect FFXI. Maybe something similar is happening with Virgin at the moment.

I too am on Virgin Media and have noticed an increased amount of R0's in the game recently. Strangely enough, not to the point of DC though. The connection would drop to 30% status or so, and then regain itself.

[Kraggy]

I think I saw something about that issue with VM though in my area the utilisation isn't very high apparently, though it's interesting you mentioned your increase in R0.

I got a reply from a VM guy on their forums saying that he wanted to have an engineer come and look at the signal strength .. he clearly didn't read my report carefully otherwise he'd have seen that the problem is only with FFXI and is so entirely reproducible that it's impossible that it's something local at my end; I'm not hardware expert but no way can router/modem hardware have a fault that's so specific, not to mention the fact as I noted the Superhub was replaced some weeks after the problem started.

[Kraggy]

This is my latest post to Virgin, I know it's not SE's fault but I need SE to tell me what network 'fault' could cause this very specific scenario, otherwise I'll never get this fixed.

=========================================

Okay, I just did some more experimenting and what I found can only be explained it seems to me by Virgin doing some sort of 'traffic shaping' or other protocol fiddling, this scenario is 100% repeatable.

I'll explain exactly what I'm seeing.

1) In the game itself there are two indications of connection status, one is a digital display marked Send and Return which provide a measure of latency, they're not 'ping' times as such but indicate the general state of data transfers in each direction. The second indication is an icon relating to the character, normally this has nothing to do with network activity but when the client is not getting some kind of reply from the server it changes to a red 'DC warning' icon; I have no idea the precise nature of the 'missing' server response.;

2) After loading a character and having it sit doing nothing at all, using Windows 10's Resource Monitor I see the following network connections listed for POL.EXE (the game client):

a) a TCP connection to 202.67.54.124 (the last quad varies from time to time and is I think the server for the game zone the character is in)

b) a UDP listener

So, as usual, after 5/6 minutes the character icon changes to the red 'DC warning' and the game client stops moving the character ... BUT ...

1) The S and R indicators continue to show network activity with unchanged rates for some time, only after a minute or so does the 'R' figure begin to dopr and a packet loss indication is seen.

2) I can do many other things in the game unaffected, for example I can list characters in the zone I'm in or on the server as a whole, I can use in-game chat, and I can initiate trade with other characters .. all activity which by definition requires client/server interactions.

3) using a command shell I can PING the listed IP address just fine right up to and beyond the moment when the game client errors off with the error (in this case 4002 "Server timeout", which as I've noted before is not what happens when the server 'disappears' with a dropped connection).



From all the 18+ years of experience I have with networking (from my first internet connection in 1989) this cannot IMO be explained by any hardware fault at my end. The fact that it's 100% reproducible after a very precise time lapse after the commencement of activity seems only to be explicable by some sort of protocol problem and one which before April I had never seen in 12+ years of playing this game.

And to repeat, this never happens with BT so there's no chance it's caused by Square Enix's systems.

[Kraggy]

Thanks for the help SE .. er, okay, I'm on my own I guess.

[Jile]

As odd as this is going to sound.... Try disabling your windows firewall then turn on XI.