On 10/09/14 13:09, Dean Pemberton wrote:
http://blog.bimajority.org/2014/09/05/the-network-nightmare-that-ate-my-week...
TL;DR version seems to be: driver bug (Intel NIC, in Dell), plus many IPv6 addresses (eg, older IPv6 privacy addresses), plus many IPv6 Neighbor Discovery multicast groups, plus what seems to some sort of MLD-handling slow path (Juniper) seems to have led to priority inversion that caused STP processes not to run often enough which led to flooding. All were sad. On the driver bug front, this (from the comments to the post Dean linked to): http://packetpushers.net/good-nics-bad-things-blast-ipv6-multicast-listener-... seems to be the same driver bug issue. May be worth investigating if you have Intel NICs on certain Dell hardware and some IPv6 things enabled (that second event seems to have been with Cisco switches). Ewen