Mark Smith a écrit :
On Fri, 02 Oct 2009 16:49:27 +0200 Alexandru Petrescu
<[email protected]> wrote:
Vijayrajan ranganathan a écrit :
Hi, Perhaps a naive question, but can somebody mention some
practical use cases for advertising multiple prefixes in a Router
Advertisement?
Let me give some more reasons which are non-standard. One could
qualify them as research.
I put several prefixes in a single RA. RAs are exchanged by
routers as they'd exchange routes with a dynamic routing protocol.
This is done in order to propagate routes. It's not the main
objective of RAs, but works ok. Well, partially.
What OS/environment are you using?
Linux and radvd modified.
The reason I ask is because I've got a slightly odd scenario where
having a router listen to RAs (in my specific case, it's own RAs,
containing a 6to4 address derived from a dynamic IPv4 address
configured via PPP) and configure it's interface addresses from it
would be useful, but as far as I know if an interface is designated
as a forwarding interface, RAs are ignored.
I think (not sure) radvd may have an option in the conf file allowing
the router to auto-configure an address based on the on-link prefix
present in the RA, even if it's a forwarding interface.
I've forgotten the exact operational details, however Appletalk
supported routers announcing network numbers to each other
("cable-ranges"), and having the receiving router automatically
configure it's addresses. I seem to remember that the receiving
router might even automatically announce that learned prefix into the
the IGP.
A-ha, good to know in a certain env the receiving router receiving an RA
may propagate that route in the IGP.
The routers I'm using don't auto-configure an address based on the
received RA; they install routes in the rt tables, towards the prefixes
present in the RA. These prefixes are not the prefixes "on-link" on
which the RAs are sent, but are prefixes of the 'other' interfaces of
the routers.
This is for a star-like topology, where by definition the link-layer
doesnt'form loops... otherwise the technique may not survive loop forming.
Alex
Regards, Mark.
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
