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

List:       ipng
Subject:    RE: ndproxy loop prevention
From:       "Dave Thaler" <dthaler () windows ! microsoft ! com>
Date:       2005-05-21 2:07:03
Message-ID: 2E33960095B58E40A4D3345AB9F65EC111A348EA () win-msg-01 ! wingroup ! windeploy ! ntdev ! microsoft ! com
[Download RAW message or body]

Bib Hinden writes:
> >This means that the STP-running sub-cloud need to be P-bit
transparent,
> right?
> 
> Yes, that is my understanding.  I think the draft is consistent with
> this.  That is, a ND proxy implements either the P-bit (choice a) or
the
> STP (choice b) modes, but not both.  In STP mode the P-bit (or any of
the
> other bits in the RA) isn't touched.
> 
> >Thus if there is an upstream which sets the P-bit, then any
downstreams
> >from the sub-cloud should receive the P-bit.
> >To do that the ndproxy nodes which run STP need to behave quite
> >differently than then ones that don't run STP. What are the exact
rules
> >for them?
> >Are they
> >  - ignore any P-bit
> >  - make sure any RAs that are received with the P-bit set are
proxied
> > (transmitted) with the P-bit
> 
> Right, and I think that is the defined behavior of a proxy
implementing
> STP (i.e., they pass the P-bit setting through unchanged).
> 
> I think it would be good if another paragraph was added to Section
4.1.4.3
> "ICMPv6 Router Advertisements" that clarifies this behavior.  Also,
> something added to Section 6. "Loop Prevention" that explains the case
of
> a site with a mixture of proxies.

I just submitted an update which should appear shortly, which
incorporates the above suggestion, and changes "STP" to "RSTP"
throughout in accordance with the 2004 bridge std from IEEE which
updates STP and renames it to RSTP.

Added to end of 4.1.4.3:
> If, on the other hand, RSTP is implemented as described in section
> 6, no special processing is done for RAs.  That is, the P bit is
> ignored, and the P bit is unmodified in RAs proxied to other
> interfaces.

Added to 6 (language very close to that in the thread on this list):
> Note that it is possible that a site has a mixture of P bit
> proxies and RSTP-capable proxies.  To understand how this works,
> consider the case where only P bit proxies are present.  However,
> within each link, there may be classic IEEE 802 bridges.  Those
> classic bridges are responsible for ensuring that that link is
> loop-free.  To P bit proxies, proxies which run RSTP are the same
> in this respect as classic IEEE 802 bridges.

-Dave


--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

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