Problems with lags? We are gathering data.

hello, dear players!

we have information that some of you might have some problems with lags or disconnects. if you have problems send us the following information please:

1) when has the problem begun?
2) results of the command 'ping geo.l2.eu -l 1000 -n 100'

Spoiler

- press win+r buttons
- type 'cmd' and press enter
- in the appeared window type 'ping geo.l2.eu -l 1000 -n 100'
- right-click and choose 'mark'
- select all and press enter
- paste the received data in your message

3) results of the command 'tracert geo.l2.eu'

Spoiler

- press win+r buttons, type 'cmd' and press enter. in the appeared window type 'tracert geo.l2.eu'
- right-click and choose 'mark'
- select all and press enter
- paste the received data in your message

4) results of the winmtr for geo.l2.eu

Spoiler

- download winmtr tool from attachments or here: http://winmtr.net/download-winmtr/
- run winmtr tool
- type 'geo.l2.eu', click start button
- launch the game. let winmtr work for 20-30 minutes at least and click stop
- make a screenshot of the received data and send us

also please indicate the time when you have problems. it is really important for us.

regards,
lineage 2 team

 

i wish to comment you with the following remarks :
-it's not us that are client of cogentco since cogentco is your data carrier provider or the data carrier provider of the datacenter where your house the server.
-cogent/cogenco is a well know provider to be the cheapest but also have the poorest peering politics in europe. they have multiple problems with opentransit (orange), l3com (level 3), teliasonera and italia telecom who are the 4 main peering providers for western europe zone. cogent have 80% market part in income revenu on the transit market but only take care of 20% of the transit in europe, wich represent 85% of the total worldwide transit. this mean cogent only care of 15% of the european transit but with overloaded routes.
-a "ping geo.l2.eu -l 1000 -n 100" will not show you something really "acceptable" for statistics use since cogent is well know to have put icmp routing rules priority as a pigeon hole argument over their customer. "see, the pings are good with no packet loss, it's not us that are faultly" ...
-a "tracert geo.l2.eu" is also not as good for information as a ping command. since cogent also route the icmp used in tracert with special rules in their bgp. pigeon holed again.
-a better way to get better stats is "pathping geo.l2.eu -p 200 -q 100 >"%homepath%\testgeol2.txt"" wich will make a pathping = a tracert + ping each node 100 times with 200ms interval between each ping and put the test log in a file named testgeol2.txt located in your used home folder.
-winmtr is a bit tough for a standard user to use.

and finally, i've already spotted that using cogent as a peering provider is really bad with some of my data in the following post : https://forum.4game.com/showthread.php?t=4354&highlight=cogent

ps : i also prefer using wireshark to monitor my network connection wich provide better statistics for advanced users but i doubt you would allow me to post infos about what i've found in the data, i also have access to 2 as, one from opentransit and one from teliasonera wich show bgp rules from cogent that are **** for icmp routing and the dpi they've put on their network.

edit : my latest result of pathping with my standard provider :

Spoiler

why am i going from aubervillier/paris (france) to london (england) to go back to paris (france) to go to reims (france) to reach the luxembourg ? too many jumps. and it seems one of your last node (assuming it's the main router of your datacenter) have 3% loss, 2% loss on your server and 1% loss in the cogent route (the 100% are well configured routers to reject icmp request, wich it's non factor). for a total 5% loss, the average rtt (round trip delay, ie : double way ping) is too high for a well play of internet games.

next is the pathping of a vpn provider i use for personnal purpose (including play l2 in good condition and such other things), it's not passing by cogent :

Spoiler

0% loss, better rtt, no weird route making an europe trip equalling the earth diam lenght ...

 

dear brume,
first of all i'd like to thank you for provided data .
perhaps, pathping utility is more informative. but we need to have trace and ping information from users in this case.
as for the rest we are working on the issue and your information is really helpful for us.

p.s. could you pm me your wireshark and winmtr results please?

sincerely, nemezida.

 

there is some weird things in the wireshark result that show some packet fragmentation in the tunneling made by frost, with different mtu size (seems that the double encapsulation is not really well supported since some of the packet are more than 1600 to 1700 bytes, especially in the evening/early night, between 18h00 to 23h00 gmt)
and winmtr have as result some packet reject by cogent router, no mention of my personnal router that is setted to not accept icmp by default (security based, but in its logs, i doesn't show any loss-it's a cisco running cisco ios so i have detailled logs and i trust them), and also not mentionning the main router of my fai (the one that stand as main registry manager for bgp/as that doesn't answer to icmp, since it's an rfc requirement).

i have to mention that it's not a problem from our provider since cogent is not our provider but yours, or the provider for your datacenter, our provider can't do anything on the route cogent use or the qos on cogent network. maybe you should try to ask your datacenter to route directly to opentransit rather than cogent.

 

update from 17/4/2012 :

Spoiler

here is the solution :
https://stat.ripe.net/query/widget/as51497?widget=asn-neighbours
https://stat.ripe.net/query/widget/as3215?widget=asn-neighbours

https://stat.ripe.net/query/widget/as51497?widget=routing-history
https://stat.ripe.net/query/widget/as3215?widget=routing-history

cogent is not our provider but yours (or your datacenters). you have 3 possibilities, makes us routes via p&t lux (more power). our provider have enough routes (as shown for orange on my as)
last but not least, a visibility map :
you https://stat.ripe.net/query/visibility-inrdb/as51497 => 52% of the ripe peers, this mean half your european playerbase have problems with routing to you. (ripe is the organisation that allow ip for european area if 52% of the ripe peers can't reach you, it's a problem)
my fai https://stat.ripe.net/query/visibility-inrdb/as3215 => 92% not 100% but enough for me, i'm not providing services.
my vpn provider (for purpose of only showing what a service provider, internet or other must look) https://stat.ripe.net/query/visibility-inrdb/as30094 => 100%
a well known game server of another game https://stat.ripe.net/query/visibility-inrdb/208.94.26.137/32 (even if based in the us) route 92% of the european peers without problems

 

thank you brume. our network engineers are working on it.
that's why we need as much information from our user as possible.

sincerely, nemezida

 

maybe you can try to use the ris tool from ripe database to investigate your routes to the inx that are most used in europe ? they have a bogon tool that is very usefull

sincerely, one of the last that have bgp as 2nd most used language. (also slackware integrist, with bearb, dirty t-shirt with wrote u