• (unregistered)

    Actually, the congrats should go to BridgeCom International.  I'm not sure if Timmy will answer.  I hope Tim or someone overthere has their thinking cap on today.


    		 207.188.96.0 - 207.188.127.255
    BridgeCom International
    115 East Stevens Avenue
    Valhalla, NY
    US

    		 Shea, Tim
    <FONT color=#0000ff>[email protected]</FONT>
    +1-914-242-1100

    Or the problem could be due to BridgeCom's upstream provider.  Whatever...

  • (cs)

    Verizon - the network that decided that, to cut down spam, they'd dump all incoming email from abroad. Actually, they'd have had a better effect on the global spam problem if they'd stopped all outgoing email.

  • (cs) in reply to Bellinghman

    Bellinghman:
    Verizon - the network that decided that, to cut down spam, they'd dump all incoming email from abroad. Actually, they'd have had a better effect on the global spam problem if they'd stopped all outgoing email.

    Yeah, this has got to be one of the biggest WTF's ever. [:'(]

  • (unregistered)

    That kind of behaviour is called "hot potato routing".

  • (cs)

    I've seen this before when a router is down, and I don't really think its a WTF.

    Let's say that Router A and Router B are reduntant circuits that both connect to Rounter C. Router C is down. A message is sent that needs to go through Router C to get to its destination.

    Router A gets the message, and attempts the shortest path (directly to C) and fails. The next shortest path is though B, so it sends the message to B.

    B attempts its shortest path (directly to C) and fails. The next shortest path is through A. So it sends it back to A.

    I don't know if that is a misconfiguration of the routers (if it should know not to send it back to A), but we had this exact situation, and the technician knew that it meant that router C was down, and they sent a technician to take a look.

  • (unregistered)

    BradC - that particular behavior can be caused by routing protocols that use only local state; in that situation, it is called "counting to infinity." (A: I can get to C in N hops.  B: I can get to C in N+1 hops.  A: I can get to C in N+2 hops.  Lather, rinse, repeat.)  Minor tweaks to the protocol can make it harder to cause that sort of problem, but it can only really be fixed if each router has wider state about the network, which requires use of a different protocol.

    Inadvertent routing loops are not a big WTF in my book, but taking six hours to fix things is pretty incompetent.

  • (cs) in reply to
    :

    Actually, the congrats should go to BridgeCom International.  I'm not sure if Timmy will answer.  I hope Tim or someone overthere has their thinking cap on today.

    207.188.96.0 - 207.188.127.255
    BridgeCom International
    115 East Stevens Avenue
    Valhalla, NY
    US

    So, we're saying that ping packets go to Valhalla (*), and never return?

     

    (*) http://www.answers.com/main/ntquery?method=4&dsid=2057&dekey=valhalla&gwp=11&curtab=2057_1

     

  • (cs) in reply to BradC

    yes...bridgecom is our reseller, but in this case it was the upstream provider, Verizon.  And while I know that the symptoms themselves aren't exactly a WTF, the fact that it took not 6, but 24 hours total to fix it leaving us high and dry in the process is most certainly a WTF.  in my book anyway...but maybe I'm the only one that cares about uptime...

  • (cs)

    The real wtf is that it was a T1 - for SDSL this is rather par for course. A T1 comes with minimum service agreements, which means that the company will be compensated for every hour that the T1 is down. (Sometimes total, sometimes after 2-6 hours a month.) Verizon will probably be owing a lot of angry customers a lot of money/free months.

  • (unregistered) in reply to
    :
    That kind of behaviour is called "hot potato routing".


    Actually, "hot potato" usually refers to policy of dumping all outgoing packets from the closest possible exit point, not the best.

    When it comes to playing ping-pong with packets, single loops are not very spectacular. Skillful network engineers can create loops in many more colourful ways, such as 8s (a -> b -> c -> b -> a, etc) or clovers (a -> b - > c -> b -> d -> b -> a, etc), but it really requires that special something to achieve such wonders by accident.


  • (cs) in reply to

    <FONT face="Courier New" size=2>10    46 ms    38 ms    54 ms  unknown.Level3.net [209.246.126.10]</FONT>

    Heh, original name for a computer. Unless this traceroute is different from the ones in my experience.

    Heh. NSLookup-ed it.

    Name:    unknown.level3.net
    Address:  209.246.126.10

    Someone at level3 has a sense of humor [:D]

    Drak

  • (unregistered)

    [:'(]  Bwahahahahahahaha, Gasp, (wipe tears from eyes), Muahahahahahaha.  Had verizon for 35 T1's 3 T3's and it wasn't that bad although every timed it rained, some of the T1's went down...

  • (cs)

    Similiar thing happend  this past summer with Sprint.  None of our VPN traffic could get from upstate NY down to Jersey.  A Trace route showed obivously where the problem was, but it still took them a day to fix it.  

  • TwelveBaud (unregistered)

    ... wait... hold the fudge up...

    coxexpress.net? cox.net? level3.net?

    What are Cox Communications and Level3 Communications doing in a UUNet-turned-MCI-turned-MCIWorldCom-turned-WorldCom-turned-MCI-turned-Verizon-bashing?

Leave a comment on “Can you ping me now? Good!”

Log In or post as a guest

Replying to comment #:

Log Out

« Return to Article