[prev in list] [next in list] [prev in thread] [next in thread] 

List:       ietf-vrrp
Subject:    RE: [VRRP] MAC address of packets sourced by a router
From:       "Don Provan" <dprovan () bivio ! net>
Date:       2005-02-14 18:23:33
Message-ID: 006001c512c2$4b977660$9618a8c0 () corp ! networkrobots ! com
[Download RAW message or body]

[Attachment #2 (text/plain)]

Danny's response is, as always, the exactly correct
factual answer: the spec does *not* require using
the VR as the source of forwarded packets because
it says *nothing* about the source of forwarded
packets. But since you ask to leverage experience, I
would go on to say that, from my experience it is
my opinion that the source MAC layer address in *all*
packets sent should reflect the hardware on which
they are sent, never the VR MAC address (except,
as Danny mentions, on VRRP packets themselves where
the spec says specifically, and for good reason,
that the VR MAC address *must* be used as the source).

I have two basic justifications for my opinion.
One is that it easily resolves all the questions
that were puzzling you: there is essentially no
logical answer in general for the question of which
MAC address to pick for the source once you
recognize that the transmission cannot always be
associated with a specific VR. What this shows me
is that the VR *really isn't* the entity transmitting
the packets -- the *actual* router is -- so there's
really no reason to begin with to think that the VR
should be identified in the MAC layer of those packets.

My second justification is that it is the only way
to tell from the packet who transmitted the packet
on the physical network. Without this information,
it is hard to work out problems involving packets
being transmitted unexpectedly by a router that
you don't think should be master of the VR. This
again leads me around to the logical conclusion that
the VR doesn't actually operate on packets inbound
to the network and should be consider irrelevant
for traffic headed in that direction.

-don

> -----Original Message-----
> From: vrrp-bounces@ietf.org [mailto:vrrp-bounces@ietf.org]On Behalf Of
> Danny J. Mitzel
> Sent: Saturday, February 12, 2005 11:34 AM
> To: Biju Mammen; vrrp@ietf.org
> Subject: Re: [VRRP] MAC address of packets sourced by a router
> 
> 
> --- Biju Mammen <bijum@tasmannetworks.com> wrote:
> > if my understanding is correct, for egress
> > packets coming out of the said interface, 
> > the source MAC address has to be  set to the Virtual
> > MAC (as it is in MASTER mode)
> 
> this is incorrect.  the only requirement on source MAC
> selection in the protocol specification applies to
> VRRP protocol messages (e.g. Advertisements).  for 
> that case the requirement is to use the VMAC.  since
> the Advertisement applies to a single VR the selection
> is unambiguous.  for all other IP,ARP,etc packets 
> forwarded or sourced by the VRRP router there is no
> restriction on source MAC selection.
> 
> 
> _______________________________________________
> vrrp mailing list
> vrrp@ietf.org
> https://www1.ietf.org/mailman/listinfo/vrrp

_______________________________________________
vrrp mailing list
vrrp@ietf.org
https://www1.ietf.org/mailman/listinfo/vrrp


[prev in list] [next in list] [prev in thread] [next in thread]