From: Peter van Oene (pvo@usermail.com)
Date: Wed Sep 04 2002 - 10:24:34 GMT-3
Hi Nigel,
comments below
At 07:40 PM 9/3/2002 -0400, Nigel Taylor wrote:
>Hey Pete,
> See comments inline...
>
>
>----- Original Message -----
>From: "Peter van Oene" <pvo@usermail.com>
>To: "Nigel Taylor" <nigel_taylor@hotmail.com>
>Sent: Tuesday, September 03, 2002 6:54 PM
>Subject: Re: NSSA and type 7 LSA's -same problem. Configs incl.
>
>
> > Nigel,
> >
> > I expect the code here is behaving according to 1587 as you properly
> > reference. In case you haven't seen them, Pat Murphy had a bunch of nssa
> > drafts out of which some have been incorporated in 3101 which obsoletes
> > 1587. In this new draft, Pat standardizes P-Bit settings which were
> > apparently never nailed down industry wide (ie how to set them and what
> > setting means what) He also has some nifty drafts out for leaking type
>5's
> > into nssa areas selectively using the P-bit like the ISIS up/down bit but
>I
> > haven't seen anyone implement them.
>
>NT: I went in search of rfc3101 with no luck. Maybe I trying the wrong
>number?
>I was also able to find #11 of the OSPF NSSA Option draft which I will be
>reading. The leaking of type 5's into the NSSA areas selectively using the
>P-bit as you noted brings a few questions to mind.. namely what Alex Zinin
>notes in his book - Cisco IP Routing. I mean that with the leaking of any
>type 5's there would also have to be mechnisms in place to allow the
>recieveing NSSA device not to drop the LSA as it will do in normal
>processing of all/NSSA lsa's. (I'll be doing some reading tonight..:-) )
PV: Here is the link to the rfc. It's brand spanking new and really hard
to find. I hit it with my first search yesterday and figured it was well
listed, but the IETF doesn't have it posted yet. They tend not to be the
quickest :)
http://www.isi.edu/in-notes/authors/rfc3101.txt
If you hit ietf.org and then go to the draft index, you can hit the ospf wg
and see all (reasonably current) drafts. I don't find it to ever be fully
inclusive, but most stuff is usually there.
The type 5's leaked into an NSSA would be leaked with P-bit clear and as
type 7 LSA's such that the stubbinesss rule would not be violated. Pats
7/5 draft covers it
http://www.ietf.org/internet-drafts/draft-ietf-ospf-5to7-01.txt
> >
> > In this case, I think R2 would only become an ASBR if it were to actually
> > originate externals itself. I don't believe translating them makes it an
> > ASBR. Being an ASBR really only helps ABR's provide connectivity to you
>in
> > the event that you are a non backbone router which in not relevant in the
> > case of an ABR so I think whether you call it an ASBR or not might be
> > mostly a case of semantics.
>
>NT: I figured as much...
>
> >
> > I think, though I am confused now due to a few threads on NSSA topics,
>that
> > the original poster was talking about a non backbone internal ASBR (r3
> > maybe) and not the case where the ASBR is also an ABR.
>
>At one point in reading through the thread I did jump to the conclusion that
>the original poster was refering to a non-internal backbone connected
>device( for which I had to make a drawing to keep me in-check)
>
> >
> > In the notes you describe (and I didn't peak at 1587) I think you are
> > dealing with the same case. P-Bit settings are vaguely documented most
> > places and 1587 isn't an exception. My take is that many vendors may not
> > even process it and instead constrain the 7 to 5 process manually with
> > route filters on the ABR. However, in the case you describe where an NSSA
> > ASBR happens to be an ABR which it is in both cases, this router MUST
> > transmit a type 5 into the backbone and MAY transmit an type 7. Hence,
>the
> > 5 will always exist and thus reachability is achieved for backbone
> > routers. If the vendor choses to also originate a 7, it needs to clear to
> > P bit to ensure a routing loop doesn't occur as you point out.
>
>Do you have any links to any of the drafts you refered to here. Please post
>them to the list also so someone else may benifit.
I think the two above should do the trick. Here is the link to the ietf
side of the ospf draft index.
http://www.ietf.org/ids.by.wg/ospf.html
Pete
<snip>
This archive was generated by hypermail 2.1.4 : Mon Oct 07 2002 - 07:43:43 GMT-3