![]() |
|
![]() | ||
|
Subject:
Unresolved network packet loss/ ping issue
Category: Computers > Hardware Asked by: shortywop-ga List Price: $50.00 |
Posted:
06 Apr 2004 14:03 PDT
Expires: 06 May 2004 14:03 PDT Question ID: 326214 |
For the life of me I cannot figure out what is wrong with my second computer on my network. I am using: Comcast Cable internet Linksys BEFSR41 Router with built in Switch 2 Computers both using Win2k on the network I have my Comcast HSI on second floor going into linksys router. The computer's connection at this location works fine. My second computer located downstars is connected via a Gigafast switch and a 25 ft CAT5 line running up through the ceiling into the router. Problem: The problem mostly arises with multiplayer gaming issues. 95% of the time I cannot ping game server lists...in the case that I can, the ping times will be extremely high. I regularly use All Seeing Eye to ping game servers on my other computer and it works fine on that one, but no go on the second. I have replaced practically every part ( cat5 line, nic, switch, even software drivers) and I can still not resolve this issue! I have even tried using a different line over a completely different path in case of interference issues, still nothing. I figure it can't be my Linksys router because the other computer goes through the router and works fine. Yes, I have tried different ports. Anyone can solve this problem I will be greatly indebted to you!! | |
| |
| |
| |
| |
| |
|
![]() | ||
|
There is no answer at this time. |
![]() | ||
|
Subject:
Re: Unresolved network packet loss/ ping issue
From: alkali-ga on 06 Apr 2004 23:24 PDT |
Shortywop, This sounds like the old problem of the unnecessarily segmented LAN. The BEFSR41 has a 4-port 10/100 switch built-in. The Gigafast switch downstairs is entirely unnecessary, and it is either causing the problem or preventing you from diagnosing it. If you are using the Gigafast because you have other computers downstairs that also need to be connected to the Linksys, then disconnect them and connect the downstairs machine directly into the BEFSR41. Then test ping to the gaming servers in that configuration: Downstairs Machine ---- Linksys ---- Comcast HSI Do this with the upstairs machine plugged in, then again with the upstairs machine unplugged to see if there is any difference, and if this has fixed it. If you are using the Gigafast because of distance, then get a longer cable to connect the downstairs machine directly to the BEFSR41. Try and do this with a high quality cable. Cheap cables twist pairs by colour pair. This is very wrong. In order reliably to go longer distances with 100Base-T ethernet, you want a cable with the TX/RX pairs twisted. You can tell by looking at the connector ends: cheap cables will go something like orange-orange green-green blue-blue brown-brown sequentially across the pins. Good cables will switch pins 4 & 6 to keep the TX/RX pairs properly twisted. See http://www.nerdlabs.org/documents/tp_cable.php for the correct pinouts. In fact, before you invest in a longer cable, try the configuration with just the downstairs machine attached directly to the BEFSR41. Move the computer closer if you have to. If that doesn't solve the problem, then you know the problem is either the cable, the ethernet card or the drivers in the downstairs machine. (You already know it can't be the BEFSR41 because you have tried different ports, right? And you have tried a different cable, NIC and drivers, do it can't be that, but try anyway.) The most likely cause given what you have said if that the Gigafast and the BEFSR41 are not making nice with each other. In the old days, we saw this problem with unswitched routers using longish cables and more than three segments, for example: hub ------ hub ----- hub is usually OK, but as soon as some non-technical individual segmented the network further, we'd start to have problems. These were mostly due to the latency involved in a multi-segmented non-switched collision based LAN. In big offices, I sometimes had to put a router in the middle just to divide the traffic. With a switched 10/100 system, the problems are different, but the symptoms may be the same. Switches don't send packets everywhere like unswitched hubs used to do. That means that they have to figure out where things go. ARP caches must be maintained, broadcasts much be carried out, packets must be examined and things can go wrong leading to increased latency. The Gigafast is a... (ahem) value product! It may have some... (ahem) challenges figuring out that the packets destined for the gaming software have to be switched through another hub rather than directly to a NIC. This is pure speculation on my part, but I think it is the first place to look. Anyway, the obvious question is: What is the Gigafast doing there? If there is a good answer to this, let me know and I'll have another think. Alan Kali |
Subject:
Re: Unresolved network packet loss/ ping issue
From: xeno555-ga on 06 Apr 2004 23:24 PDT |
Here a few thing you should try and post the results 1: On all your computers (DOS PROMPT) type "ping 127.0.0.1" (ping should be less that <10ms) 2: Now ping your other computers, from each other. (Should be <10ms) Suggestions Use static IP's i.e. 192.168.0.1,192.168.0.2 ect... with subnet 255.255.255.0 Type in your gateway (your broad band router/modem) X |
Subject:
Re: Unresolved network packet loss/ ping issue
From: alkali-ga on 07 Apr 2004 21:40 PDT |
Shortywop, You have covered everything obvious, so it is time to start thinking laterally. There are three obvious sources of problems: (1) The downstairs computer has a hardware problem. You've changed the drivers, so it is probably not software, and you've changed the NIC so it cannot be that. This leaves some kind of PCI or chipset problem - very rare indeed producing the symptoms you describe, but not unheard of. You haven't told us what hardware and NIC(s) you are using downstairs. It could be important, so you should tell us. On balance though, this probably aint it. (2) A problem with ASE. I have been assuming that your ping times to game servers were high, but that is a big mistake. ASE implements ICMP differently from Windows 2000. You should note down the IP address of the gamer servers not responding in ASE (assuming you can get that far) and ping them manually from the Win2000 command prompt. See if they are also high from the prompt. If they are, then you truly have a high ping loss/latency problem. If not, then ASE is your problem. Since the computer upstairs is working fine, and assuming you are running the same version of ASE, then you have ASE misconfigured on the downstairs machine. Perhaps you have forgotten to check the "Broken Router" box in the Options - Network page of ASE on the downstairs box. This is necessary with the BEFSR41. If you don't have a "Broken Router" checkbox, make sure you are running v2.3.1 or greater of ASE (3) A protocol problem. This is less likely, since you should not have to configure the BEFSR41 specially unless you are serving under ASE, but there is a remote possibility of some wierdness. For example, you could have a wacky packet filtering rule in place that allows ICMP only to the upstairs machine. This is way out there, but that's the territory we're moving toward. If this doesn't fix it, can you tell us: Hardware (downstairs & upstairs) NIC models ASE version "Broken Router" checked? Command prompt ping gives same results as ASE ping times? Thanks, Alan Kali |
Subject:
Re: Unresolved network packet loss/ ping issue
From: alkali-ga on 07 Apr 2004 21:51 PDT |
Some weirdness crept into that post, all right. I do know when to break the "i before e rule" and I am not a gamer but a lamer when I speak of "gamer servers", so please delete the "r". Alan Kali |
Subject:
Re: Unresolved network packet loss/ ping issue
From: alkali-ga on 08 Apr 2004 01:09 PDT |
Oh, yes - on the network packet testing program. There are many great, free network analyzers out there. Ethereal is the one I use, and I used to use the Microsoft one that came with SMS. Unfortunately, it is a lot of work setting up the filters for an unknown problem, and it may not tell you anything, anyway. If you have very high latency or packet loss, even a properly constructed filter will probably just show you what you already know: pings are going out and coming back late or not at all. Diagnosing ASE that way is going to be even more challenging than looking for regular packet loss, since the documentation for ASE is a little on the sparse side. Creating your filters will be fun. If the problem is somewhere in your computer or the BEFSR41, you won't see that with a sniffer; you only get what's on the segment. You won't see problems ascending or descending the layers at either end, and, of course, you won't see the Internet traffic. Also, if stuff is getting switched wrong (or slow) you won't see that, either, unless you run the analyzer on each segment. In the case of a switched network, this means every port. It used to be a lot easier in the old days. In case you want to try it, here are some packet analyzer links: http://www.ethereal.com/ http://www.analogx.com/contents/download/network/pmon.htm http://netsecurity.about.com/cs/hackertools/a/aafreepacsniff.htm (links page) Alan Kali |
Subject:
Re: Unresolved network packet loss/ ping issue
From: alkali-ga on 08 Apr 2004 14:04 PDT |
The fact that you are getting high ping times to the game servers from the command prompt is great. It eliminates a problem with ASE specifically, and it gives you a reproducible error - an excellent diagnostic tool. I cannot suggest strongly enough that each time you test an alternate configuration, you standardize your testing conditions. Here are some suggestions: (1) Use a set of IP addresses that give consistently high ping times downstairs and ping all of them from the command prompt. (2) Make sure that there are no programs running when you do this - especially ASE. Shut down anything in the system tray, and try to stop unnecessary services. (3) Check your autostarts and remove anything you do not need. Use something like StartupRun v1.20 from http://nirsoft.mirrorz.com/ that allows you to disable, rather than delete autostarts if you are not sure about how to do this. (4) Temporarily disable any firewall and virus scanner on the downstairs computer. Make sure there is no built-in packet filtering going on in the Win2000 network configuration settings. (5) It will mean a lot of running upstairs, but you should compare each result of a ping session from downstairs with exactly the same procedure upstairs. You have changed NICs and drivers, so it cannot be a hardware or driver problem with the downstairs NIC. You are running the same OS and service pack level as upstairs, so it cannot be the OS. You have low ping times to servers on the LAN, so that all but eliminates the possibility of a hardware problem on the Epox board. It is possible, but very unlikely. Also, any settings for speed and duplex on the downstairs NIC are likely to be unrelated, since an inappropriate setting here would affect local as well as game server ping times. Cables are out; you've changed them enough to be sure. Cable path interference is out. You've tried that one. The only logical conclusion is that pings are actually being lost or delayed; that is what the downstairs machine is telling you, and at this point I am inclined to believe it. One of the chief methods of troubleshooting is to simplify the system as much as possible while you test a reproducible error under standard conditions as described above. You should take the network down to just the downstairs machine --- BEFSR41 --- Internet and perform the ping test from the command prompt. No other devices should be installed: no Xbox, no Gigafast, no upstairs computer, nothing. Confirm that pings are high in that configuration, then plug in the upstairs computer and do the ping test there for a control. Then go back downstairs and repeat the ping test to confirm that the results have not changed. If you get high ping times from the command prompt in this configuration, then pings are most likely being delayed somewhere beyond the NIC of the BEFSR41. Your upstairs machine gets through it fine, so the only reasonable explanation is that you have a wacky filtering rule on the BEFSR41 that is preferentially returning traffic to the upstairs machine, or blocking traffic to the downstairs machine. I have to admit that I think this is an outside possibility. I suspect that the problem is on the LAN, and that it has been difficult for you to diagnose due to nonstandard testing. Whenever I have gotten to the stage you have with a network problem, namely that it seems to defy all common sense, I have found that the problem was me assuming something without testing in a rigorous, standardized fashion. Sometimes I have been so convinced of something that the problem seemed to be supernatural; it never was - I just had to reexamine my conclusions. By the way, while I don't think it matters at this point because you have probably eliminated these things, You are a little vague about the hardware. When you say Realtek RTL1839, you must mean RTL8139. Realtek makes chipsets. The card is probably a generic with a Realtek chipset. Can you identify it further as a RTL8139C(L), RTL8139C(L)+ or RTL8139D(L) chipset? When you say you swapped out your 530TX, I assume you mean DFE-530TX+ (http://www.dlink.com/products/?pid=122). And could you please tell us what is the model number of the Eox board in the downstairs machine? Best Regards, Alan Kali |
Subject:
Re: Unresolved network packet loss/ ping issue
From: alkali-ga on 08 Apr 2004 14:30 PDT |
Shortywop, One other thing: in testing just now, we have been able to reproduce a problem where ASE ran, timed out, and then prevented the system from pinging any hosts (both local and Internet) from the command prompt. This occurred both during and after ASE was in operation, and resolved by itself about a minute after ASE was terminated. It is very important therefore that you make a list of high-ping time servers in ASE, then ensure that ASE is not running and you have low ping times to local machines - possibly even reboot before doing the tests. So far as I can tell, ASE is using up all network resources on the local machine. This is only affecting one machine on our LAN: an Emachines laptop that normally does not run ASE. ASE works fine elsewhere, so this may mean that ASE has problems with certain configurations. Alan Kali |
Subject:
Re: Unresolved network packet loss/ ping issue
From: alkali-ga on 09 Apr 2004 10:01 PDT |
OK, so it is not just high ping times and ping packet loss. From your latest post, it is a global failure affecting multiple protocols (TCP, UDP and ICMP by the sounds of it). This is more likely a broken stack. A broken stack can manifest in ways that are hard to pin down. Sometimes small or single packets will get through, and large or frequent packets won't. I suggest that you try moving a large volume of data between the downstairs computer and the upstairs computer. See if it gets corrupted or if packets get lost. There are several ways to do this, and you should try as many as you can. Set up a web server on the upstairs machine. AnalogX (at http://www.analogx.com/contents/download/network/sswww.htm) have a great one for this purpose. Make sure you set it up to serve a variety of test data, including text as well as large and small images. If you get the same errors as you have reported above, then you probably have a broken TCP/IP stack on the downstairs machine. You should also try FTP (http://smallftpd.sourceforge.net/) and SMB. I would share a temporary directory on the upstairs machine and map it from the downstairs machine, then transfer .zip or .rar files back and forth. If the transfer hangs or fails, or if you test the archive after transfer and find errors, once again it points to a broken stack downstairs. If these tests are negative, and the problem is confined to Internet traffic, then it is more difficult to explain, given what you have told us. In this case, you may want to make absolutely sure that you have a standardized bare network testing setup as I described earlier and confirm that the problem still exists. Alan Kali |
If you feel that you have found inappropriate content, please let us know by emailing us at answers-support@google.com with the question ID listed above. Thank you. |
Search Google Answers for |
Google Home - Answers FAQ - Terms of Service - Privacy Policy |