|
Author Topic:   What's up with the goofy lag? Network geeks, err... techs needed. :^)
Irascible
Member
posted January 09, 2004 02:43 AM            
There are two problems that may or may not be related.

1) My tank stutters rapidly back and forth a small amount. It's independent of any single computer, cable, or even game. A player in counter strike will jerk around too. I've reset the modem, bypassed all home network devices, tried different computers, tried different cables - everything that I can think to do locally. The (n)ever helpful online tech guy for my ISP had me ping a couple servers with no problems resulting, so he's not convinced.

A short movie of it. Folks on dial-up will want to do a "save target as".

Video 2

This one is redundant. But it's always nice to blow up Max.

2) Then there's the play on LDA. I'm getting an unusually high amount of lost packets resulting in teleporting. A 1000 packet test to LDA averaged an 8 percent loss. A 10,000 packet test to 8 of the 9 pingable servers yielded the results below. All servers were flawless except LDA. The dropped packets occur even when no one is on LDA or LDA comm.

Any ideas? I haven't a clue what's up with the tank stutter. It must be my isp. The packet loss seems to be all about the LDA server.

[This message has been edited by Irascible (edited January 11, 2004).]

IP:

KiLlEr
Member
posted January 09, 2004 01:16 PM            
I had similar problems as well a month ago with the LDA server. It took a nearly 2 weeks for it to clear up. It is usually caused by a bad router out on one of the branches of the internet.

If you do a tracert to the LDA server, then working backwards from the LDA server run that ping test to see at which machine the lost packet percent is 0. Then you know the problem lies between that machine and the one you tested immediatly before it.

You can then talk to your ISP about it, as you know at what point the packets are getting lost. They might be able to contact whomever owns that machine and have the trouble adddressed.

IP:

Irascible
Member
posted January 09, 2004 02:28 PM            
"If you do a tracert to the LDA server, then working backwards from the LDA server run that ping test to see at which machine the lost packet percent is 0. Then you know the problem lies between that machine and the one you tested immediatly before it."

TY Killer. Done and done. There's that big ole piccy right above! The problem with the lost packets is LDA itself.

IP:

Sailor
Administrator
posted January 09, 2004 02:49 PM            
I think what Killer is saying is that it is probably a router that is in a location prior to sending the packets to LDA and that if you move back and ping the address before the LDA server then see if you still have packet loss. If so then move to the next adddress back and test it. If it is at the final router prior to LDA then their ISP may be the one with a bad router - not that the LDA server itself is bad. Hope that makes sense.

------------------
May you always have a following Sea a cool breeze at your back and the comfort of those you hold nearest to your heart.
Sailor's Home Port


IP:

Irascible
Member
posted January 09, 2004 04:05 PM            
OK.

Man.

Killer, Sailor, you're both great. I don't think I need to tell you what an asset you guys are to this forum. But so help me God, if one of you tells me to do that which I've already done one more time, I'm going to blow up!!!

Now I know I'm not very good and speaking network talk. So maybe I haven't communicated well. Statements like the one I made above: "All servers were flawless except LDA" - well, perhaps that wasn't technical enough. And the picture showing a tracert - showing all servers at 100 percent except LDA, which had 6 percent dropped packets - well, perhaps you're stuck on a 14 inch monochrome and can't really see it. I'm trying to be understanding. We'll get through this together.

Kidding aside (and yes, underneath the heavy layer of sarcasm was the intention to "kid"), unless 6 percent loss is normal, LDA has a problem.

IP:

Blue n Gold Sue
Member
posted January 09, 2004 04:13 PM            
Maybe you need to be a non-tech head to understand... I got you perfectly Ras. You pinged LDA and got the loss, so you pinged each address that was listed in the first test and got 100 % good results - that is UNTIL YOU HIT THE LAST ADDRESS, which is LDA's.
So you think the problem is a LDA.
I think Sailor is trying to tell you that while LDA's machine MAY be the problem, it still may not be at that location, as there is still network connections outside of LDA and to the network... At least that is what I am thinking that is what he is saying. Capiche?

IP:

Irascible
Member
posted January 09, 2004 04:27 PM            
In terms of routers that are revealed by the tracert, we can eliminate all of them except LDA. Some routers don't respond to ping requests. But the two closest to LDA in my test did respond and had no dropped packets. Of course, tracert doesn't reveal anything that's not ping-able. Obviously there could be switches, hubs, bridges and who knows what else between LDA's router and the closest router to LDA. So while the problem may not be LDA's actual router, it's close to them for sure.

IP:

KiLlEr
Member
posted January 09, 2004 04:27 PM            
Yeap Sailor's got me interpreted correctly. The problem lies between nodes/gateways or between nodes/gateways and a server. It is possible that the LDA server has a bad network card, but it can also be a bridge or router between the LDA server and their ISP.

You don't have to do it again, as long as you have the IP of both machines between which the failure occurs.

I got:

Ping statistics for 206.222.78.139:
Packets: Sent = 61, Received = 55, Lost = 6 (9% loss),
Approximate round trip times in milli-seconds:
Minimum = 50ms, Maximum = 119ms, Average = 57ms

for the LDA, and 0% for 66.46.34.110

the TTL (time to live) on the reply from LDA was 107, and the 66.46.34.110 machine was 235.

Seems like there is alot of hops inbetween LDA and their ISPs gateway.


This is my path to the LDA machine (ignore the first line, thats my LAN's gateway).

Tracing route to 206.222.78.139 over a maximum of 30 hops

1 * * * Request timed out.
2 20 ms 24 ms 15 ms 10.18.128.1
3 13 ms 13 ms 14 ms dstswr2-v2.rh.lybrny.cv.net [167.206.33.34]
4 13 ms 15 ms 12 ms r4-ge9-2.mhe.hcvlny.cv.net [167.206.33.5]
5 21 ms 14 ms 14 ms 65.19.104.18
6 12 ms 13 ms 12 ms 65.19.104.2
7 14 ms 17 ms 13 ms r2-srp5-0.in.nycmny83.cv.net [65.19.96.54]
8 13 ms 14 ms 14 ms so-2-0-0.gar2.NewYork1.Level3.net [167.206.8.62]
9 15 ms 34 ms 13 ms ge-1-3-0.bbr1.NewYork1.Level3.net [209.247.9.205]
10 14 ms 13 ms 16 ms ge-6-0-0.edge1.NewYork1.Level3.net [64.159.4.146]
11 14 ms 13 ms 15 ms att-level3-oc48.NewYork1.Level3.net [209.244.160.130]
12 1437 ms 15 ms 15 ms tbr1-p010401.n54ny.ip.att.net [12.123.3.57]
13 42 ms 41 ms 41 ms tbr1-cl1.cgcil.ip.att.net [12.122.10.2]
14 43 ms 40 ms 41 ms gbr6-p40.cgcil.ip.att.net [12.122.11.46]
15 39 ms 38 ms 40 ms gar2-p370.cgcil.ip.att.net [12.123.5.221]
16 48 ms 52 ms 48 ms pos9-0.gwy2-tor.bb.attcanada.ca [12.125.142.6]
17 48 ms 51 ms 48 ms srp0-0.core1-tor.bb.allstream.net [216.191.65.241]
18 49 ms 49 ms 49 ms pos1-0-0.bgp1-tor.bb.allstream.net [216.191.67.42]
19 49 ms 49 ms 52 ms fe0-0.adsl4-tor.bb.allstream.net [216.191.67.242]
20 53 ms 51 ms 72 ms 66.46.34.110
21 52 ms 51 ms 52 ms 206.222.78.139

Trace complete.

[This message has been edited by KiLlEr (edited January 09, 2004).]

IP:

KiLlEr
Member
posted January 09, 2004 04:37 PM            
Maybe I better explain what I mean.

You see, there are 21 hops between this machine I'm using and the LDA server, including LDA, as reported by tracert. The pings reportes a TTL of 235 from the gateway LDA is connected to. that means that 255 (maximum TTL possible) minus 235 is 20, which is exactly the number of hops from my machine to that gateway ( hop #21 is LDA).

However, LDA's server had a TTL of 107.

255 - 107 = 148 hops when it was suppose to be 21 (or really close to 21)

This means that 127 hops took place between the LDA server and thier ISP's gateway on good packets.

THis most likely indicates a problem between the LDA machine and the Gateway, and not the LDA machine itself (if the network card in the LDA machine was having troble, the TTL of the returned ping packets whould be 234, not 107).

[This message has been edited by KiLlEr (edited January 09, 2004).]

IP:

Irascible
Member
posted January 09, 2004 04:53 PM            
127 extra hops? Sounds like a beer commercial gone bad: "Now even more flavor with 127 extra hops!"

I thought time to live was... time. In other words, I thought a packet was given so much time to do it's thing. If the TTL expired, then the router that has it when it expires would flush it.

But, it's been a while since I've taken that course. So assuming that ttl is the max allowed hops, I would assume that those hops could be representing some of the same equipment getting hit more than once - right? I can't imagine there actually being 127 pieces of equipment in addition to the 21. So it must be bouncing around circuitously, yes?

IP:

KiLlEr
Member
posted January 09, 2004 04:57 PM            
Yeap. Seems that way. a pair of routers bouncing the packets between themselves or a router bouncing the packet between the LDA server and itself.

The TTL is set to 255 when the packet is initially sent. Everytime it gets routed, the TTL is decremented by 1 before being sent. When it hits 0, it gets dropped.

Tracert works by starting the TTL at 2, sends out a ping. Then tracert increments the TTL by one, and sends out another, etc for the maximum number of hops you specified. In other words, it sends out a packet with the destination set to the IP you specified to all network appliances that are 1 hop away, then 2 hops away, then 3 hops away, etc, effectively mapping out the path your packets would take to thier destination.


127 extras hops? thats good beer! LOL

[This message has been edited by KiLlEr (edited January 09, 2004).]

IP:

VorDesigns
Member
posted January 09, 2004 04:58 PM            
The old location for the Dinochrome server (moved it yesterday) had a lag problem even thought the ping was usually very good.
I traced the problem to a router immediately adjacent to the location ISP and there was nothing they could (would) do about it.
Essentially, the only service that was really sensitive to bad routers was TM and only itermittently mail duplications for Outlook users.
So, I moved the game server to another circuit.
Bad router info is common on larger ISPs.

Oh, by the way, dinochrome is living in a new loation, maybe people will start using it again.

[This message has been edited by VorDesigns (edited January 09, 2004).]

IP:

Irascible
Member
posted January 11, 2004 06:57 AM            
Hey Vor,

Could my ISP be causing the jerking motion that's in the video? Unfortunately it has no connection with packet loss. Nor does it seems to have any connection to any thing local after the modem.

IP:

Irascible
Member
posted January 11, 2004 10:56 PM            
Need help with the jerking crap! The second video is what I was dealing with tonight. It comes and it goes.

IP:

Paranor
Member
posted January 12, 2004 01:43 PM            
The problem is clearly on the LDA side. If I read all of the posts correctly both Killer and ras have lost packets. That's not good.

IP: