Baffling network conundrum

JakeJake Alec Baldwin's Chest Hair
edited August 2003 in Science & Tech
Alright, here's the situation, at one of the locations where I have a computer set up, dial-up is the only Internet option. There are two systems at the site; one dials in and shares Internet access via Windows XP's built-in Internet Sharing, and the other system receives that shared Internet access via a wireless connection.

The Internet host system is plugged into a D-Link DI-714 router via a CAT-5, and the client system receives the Internet via a D-Link DWL-AG520 wireless card. The router has DHCP serving disabled, and has been manually assigned the IP of 192.168.2.1 so as not to conflict with the 192.168.0.1 IP that Windows automatically assigns the host system due to the Internet connection sharing.

This setup seems to work perfectly fine for all of my needs at this site, save one: I cannot check the folding stats page at folding.extremeoverclocking.com from the client system, only the host system.

From what I have thus far experienced, this one website is the ONLY website for which this problem exists. Now, I have no idea if this will help anyone solve this mystery, but I have two screenshots for you to look at. This first screenshot shows the info returned from the IPCONFIG command on the host system, as well as the first hop in a TRACERT to www.extremeoverclocking.com and the first hop to folding.extremeoverclocking.com.
01.jpg 54.8K

Comments

  • JakeJake Alec Baldwin's Chest Hair
    edited August 2003
    And now, the second screenshot shows the info returned from the IPCONFIG command on the client system, as well as the first hop in a TRACERT to www.extremeoverclocking.com and the first hop to folding.extremeoverclocking.com. Note that the TRACERT to www.extremeoverclocking.com begins and passes the host system, but the TRACERT to folding.extremeoverclocking.com dies seemingly before even getting to the host system.
    02.jpg 47.9K
  • JakeJake Alec Baldwin's Chest Hair
    edited August 2003
    So, are there any brilliant network gurus hanging around who can give me any ideas as to what might be going on?

    Or, for that matter, are there any village idiots who can see the forest for the trees and recognize some obvious solution that has, thus far, eluded me?

    Any help is much appreciated.
  • TemplarTemplar You first.
    edited August 2003
    Try ipconfig /flushdns on both machines. Maybe you have a quirk somewhere.

    Clearing your cached files might also help, but I'm betting the dns flush will fix it, if anything. I don't think a tracert will actually "trace" if it can't resolve the target (at least that's been my experience).
  • JakeJake Alec Baldwin's Chest Hair
    edited August 2003
    Both systems, the host and the client, have had their DNS resolver caches flushed using IPCONFIG. Sadly, to no avail.
  • TemplarTemplar You first.
    edited August 2003
    Have you tried connecting the client machine directly to the internet connection instead of going through the host machine?

    I can ping that address, but I can't get a webpage. My guess is they're just having server trouble. You could email the admin of www.extremeoverclocking.com and ask him what's up.
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited August 2003
    I got both right now. This is a Verizon hookup, and both sites are masked with security redirects in end routing-- and end servers are not responsive to traceroutes.

    I can say this, the folding.extremeoverclocking.com page says it is updating stats, and has a semilocal route with an IP followed by a three part URL in its traceroutes.

    Idea, give it 24 hours, sounds like a strange route issue on their end, like someone is working on router close to that site. The www.extremeoverclocking.com site is not responsive to traceroute either right now, a router might be masking deliberately the route, and I suspect the end node site server might be being used for work in web recovery by owner also, so you might have hit a time when he was checking his router or when he was using it to check his sites as an admin.

    The other thing you can do if this persists, since one computer is hardwired, is to hardwire the other if feasible-- for a test. Your own router can set the DMZ inside of its wireless ports and if it was power-downed it might have come up partly at default and might not be feeding DNS outside itself quite right. The very oddball end address of the last two traceroute entries on hops 17 and 18 for folding.extremeoverclocking.com shows it was kludged and I suspect if you tell the admin these results he will know what is up and get it fixed, or Verizon will normalize routing later.

    traceroute hops 17 and 18 (tracing from Florida, ignore latency):

    17 unknown.Level3.net (63.210.132.78) 58.453 ms 59.274 ms 58.176 ms
    18 242.152.171.66.subscriber.vzavenue.net (66.171.152.242) 60.388 ms 57.942 m
    s 57.897 ms

    8 next hops no response to traceroute, aborted after hop 26.

    (note, I am not using Windows command of tracert, I am on a hardened Linux box here so I use traceroute as literal command call with site name after.)

    The other site is not on the same region of Verizon, and I think actually a hosted site in Western US but not West Coast.

    John, who thinks this might be routing hardening fluke, to ignore for 24-48 hours even if the problem has been that long, as Verizon is probably patching its own stuff and working around routing that normally woudl be used to let patches be applied offline on things serving normal routes, adn to block ports they can sometimes pull routers in this kind of attack to stop worms and IRCBOTS (like w32.randex.e 's bot, this is also an RPC DCOM worm (source Symantec and Microsoft in KB 826955 which is new MS KB for Blaster and includes links to w32.randex.e info on Symantec and the Blaster standalone fixer (FixBlast.exe) which is really only a ~140K dedicated Blaster, Blaster.b, and Blaster.c fixer), but affects all computers XP and Earlier other than IIS if not virus protected and for some if not firewalled) from propagating to some places at the routers.
  • dodododo Landisville, PA
    edited August 2003
    Perhaps the folding site is not running on port 80? Then maybe ICS won't forward it? Just a guess...

    ~dodo
  • JakeJake Alec Baldwin's Chest Hair
    edited August 2003
    Alright, problem fixed. Thanks for the ideas, all.

    Got some feedback on Icrontic that led me down the path to the solution. I changed the router's IP to 192.168.0.50, kept the router's DHCP server turned off, left XP's ICS turned on on the host system, and presto, the site is now available on both the host and the client.

    So, the client's requests for net access now go from 192.168.0.100 > 192.168.0.50 > 192.168.0.1 instead of 192.168.0.100 > 192.168.2.1 > 192.168.0.1. Don't you just love the occasional vagueries of subnets? :rolleyes:

    Once again, thanks for the ideas, folks.
  • TemplarTemplar You first.
    edited August 2003
    Subnetting make me cry. :)
Sign In or Register to comment.