I may be missing something obvious, but shouldn't your prefixes have hit the Internet with a hop count of 1 and the bogus routes a hop count of at least three? If that's the case wouldn't your prefixes be the best path? Assuming I've missed something than the no-export community and as prepend suggestions proposed by others should work.
Sent from my iPhone On Apr 18, 2011, at 2:36 AM, Artyom Viklenko <[email protected]> wrote: > Hi, list! > > I had some problem last weekend. Lets say we - ISP-A, announce > some prefixes to Customer. Due to some bug or misconfiguration > these prefixes reached ISP-B (who provides another uplink to > Customer). I'm was surprised that ISP-B received our prefixes > from Customer (wrong filters?). And then, these prefixes was > announced to Internet and some Exchange Points. > This lead to incorrect routing of incoming thrafic towars our > prefixes via slow Customer's links. This lead to near 100% > packet loss. > > After several phone calls problem was fixed. But now, I'm > trying to find some solution to prevent such problems in future. > > One solution I thinking of is to mark all announces to such > non-transit Customers with no-export community. > > What do you guys think about this? Is it acceptable or not? > Is it any other possible solutions to prevent such cases > already in place? > > Thanks in advance! > > -- > Sincerely yours, > Artyom Viklenko. > ------------------------------------------------------- > [email protected] | http://www.aws-net.org.ua/~artem > [email protected] | JID: [email protected] > FreeBSD: The Power to Serve - http://www.freebsd.org > _______________________________________________ > cisco-nsp mailing list [email protected] > https://puck.nether.net/mailman/listinfo/cisco-nsp > archive at http://puck.nether.net/pipermail/cisco-nsp/ > _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
