Hi:
See below. I was just focusing on the DHCPv6 protocol issues
in my
analysis.
- Bernie
-----Original Message-----
From: Templin, Fred L [mailto:Fred.L.Templin boeing.com]
Sent: Monday, January 22, 2007 11:34 AM
To: Bernie Volz (volz); jinmeiisl.rdc.toshiba.co.jp
Cc: dhcwgietf.org; stig.venaasuninett.no
Subject: RE: [dhcwg] dhc wg last call on agentopt-delegate
and
srsn-optiondrafts
Bernie,
> 1. If the leases are not reused, there is no serious
issue.
> The Relay Agent may continue to advertise the routes
and/or
> allow the traffic to pass. But, as the client isn't
using the
> prefix/address, no significant harm occurs. [This is
similar
> to the case where the client disconnects from the
network
> without having released the leases.]
Wouldn't it be preferable if the relay/router
could send destination unreachables back to
the source instead of just black-holing in
that case?
BV> Certainly the routers should do what they normally do
if the
destination is unreachable and they know it isn't reachable.
But in this
case it may not always know or it may take some time before
this is
detected (ND timers).
> So, looks like any way we slice this there's always
going to
> be a loophole?
There was the idea of the relay/router also
configuring a client to ensure reliability in
case the true lease-holding client has departed
or terminated the Release exchange w/o waiting
for a reply. Any problems with that?
Another idea is for the relay/router to
periodically perform NUD and then purge stale
state if the client fails to respond to its
delegations. That is what is being recommended
in NETLMM at least.
BV> Yes, the router should certainly use other techniques
if possible to
determine whether the destination is unreachable or not. For
example,
the router can always use a routing protocol to assure that
the link is
up and operational between itself and the client router.
Fred
fred.l.templinboeing.com
_______________________________________________
dhcwg mailing list
dhcwgietf.org
https://
www1.ietf.org/mailman/listinfo/dhcwg
|