All things related to multicast

[Mark Fullmer <>]

On Thu, Nov 04, 1999 at 01:10:44PM +0100, 

 wrote:

> I suppose you see these problems with both 12.0(6)S and 12.0(7)S,
> and that the spurious memory accesses only appeared with 12.0(7)S?

It was happening in 12.0(5)S (7505) and 12.0(6.6)S1 (7206VXR)

The 12.0(6.6) code also logged the same low memory access (0x18).

> This particular peer is 96ms away (across the Atlantic, so not
> much we can do about that), so TCP will take a while to ramp up
> it's data rate after connection reset (or after a quiescent
> period).

I'm seeing this on directly connected routers.

> I wonder if this isn't exposing a weakness of running MSDP (and in
> particular sending the SA messages) over TCP.  Doesn't the current
> internet-draft for MSDP specify UDP as an option for encapsulating
> SA messages, an option which Cisco currently doesn't implement?  If
> I recall, Cisco has also made sounds to the effect that UDP
> encapsulation of SA messages isn't on their TODO list either. :-(

The TCP resets are just how the protocol works and isn't necessarly a
bad thing.  If one of the peers loses its place in the bytestream
they have to reset to sync up.

--
mark