Was it only LDP that dropped? Did you see any other control plane impact? Were the timers the same? If multiple protocols had issues I'd think about high CPU or a punt-path problem. If it's only LDP, I'd look at interface/forwarding engine/fabric congestion for where the failed peers are.
On Fri, May 24, 2013 at 9:46 AM, Grischa Stegemann <[email protected]> wrote: > Hi all > > Today we were facing a ddos attack with approx. 1.5 GBit/s and 3 > Mpackets/s. The traffic came in on a port of a 6704 and went out on > another port of the same linecard. > > We have two annoying phenomena to deal with: > > 1. The 6509 is equipped with a Sup720-3BXL and DCEF on all linecards. > During the attack it dropped all of its LDP neighbourships (actually > they were flapping) although the CPU usage remained below 40% for most > of the time and only went up over 50% at a very short peak not related > to the LDP drops. > > What might have caused the dropping of the LDP neighbourships and what > would one do to protect them? > > 2. On the outgoing interface we recorded a huge amount of output drops > up to 60kpackets/s. The receiving router here is another Sup720 with > 6704 but without DCEF. > From my understanding 1.5GBit/s and 3Mpackets/s should be no reason to > drop packets on a 10G link. > So what might have caused these packets to get dropped? > > Thanks for any advice or hint, > Grischa > _______________________________________________ > 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/
