Hi list, I'm seeing blackholing on my inbound traffic from XO and their downstreams (notably CNN) via Cogent. Prepending towards Cogent changes my inbound path from 2828 174 39029 to 2828 3549 3292 39029 and it works fine, even though the outbound path is still via Cogent: tore@cr3> traceroute 2610:18::3050 wait 2 no-resolve traceroute6 to 2610:18::3050 (2610:18::3050) from 2a02:c0:1000:1::2, 64 hops max, 12 byte packets 1 2a02:c0:1000:1::1 0.822 ms 1.290 ms 0.355 ms 2 2001:978:2:30::1 0.465 ms 0.477 ms 0.395 ms 3 * * * 4 * * * 5 2001:7f8:4::ae:1 582.740 ms * 82.538 ms 6 * * * 7 2001:504:f::c 106.967 ms 109.421 ms 107.129 ms 8 2001:550:3::11e 121.969 ms 103.842 ms 112.119 ms 9 2610:18::3050 143.462 ms 142.717 ms 145.179 ms 2610:18::3050 is the first hop in XO's network on the way to ipv6.cnn.com. If I stop prepending towards Cogent, 2001:550:3::11e is the last hop that gives me any replies. So it appears my announcements reach XO, but the packets gets lost somewhere along the line. Cogent told me they found nothing wrong with their end of the peering. Anyone in or behind XO with the same problem? 2001:978:1:326::2 is an example of a router in Cogent's network (in Stockholm) that should respond to pings if everything works fine. It seems like the problem happens with Cogent's own routes as well, as neither 2610:18::3050 nor 2620:0:2200:8:8888:8888:8888:8901 (ipv6.cnn.com) get any replies when pinging from Cogent's looking glass. Best regards, -- Tore Anderson Redpill Linpro AS - http://www.redpill-linpro.com Tel: +47 21 54 41 27