medin@NSIPO.NASA.GOV ("Milo S. Medin", NASA ARC NSI Project Office) (12/15/90)
Well, actually, at the last router requirements meeting I suggested that RIP and all routing protocols which broadcast updates should require the TTL of the outgoing packets to be set to 1. This will result in a TTL expiration if a router attempts to forward it. In OSPF, we multicast instead of broadcast, so normally, routers not running the protocol will never see the traffic. However, in the case that someone has a bug in their multicast support, they might still try and forward it, which is why we also specified all multicasts should have a IP TTL of 1. Sure enough, we ran into someone who had such a bug about a month ago, and the only thing that prevented it from trying to forward the datagram was the TTL being set to 1. In general, routing packets should never be forwarded, but because of problems with certain implementations not being able to tell if a particular packet came in via a media level multicast, you can still run into trouble. This is why the setting of the TTL to 1 is a good idea. Thanks, Milo