Re: AD-review for draft-ietf-isis-*-interoperable

[Alex Zinin <[e-mail]>]

Radia,

Tuesday, April 1, 2003, 11:49:31 AM, Radia Perlman - Boston Center for Networking wrote:
>>From my distant memory...I believe I introduced the concept of purging other
> routers' LSPs. My reason was worry about a router R1 having a very
> fast clock relative to the other routers. In that case, R1 might purge
> R2's LSP long before anyone else does. And then R1 would be
> calculating paths on a different LSP database than the other
> routers. So it seemed like a good idea to have all routers purge
> LSPs at the clock of the fastest router, so their LSP databases would
> be the same.

> I have to admit I don't have a copy of 10589, so I'll ask a stupid
> question here...I'm assuming even if you don't synchronize the
> purge, that if R1's age on R2's LSP gets to 0 that R1 would
> throw away R2's LSP? Or has it changed to be that only R2 gets
> to purge its own LSP?

> Perhaps LSPs should never be purged, (memory's free these
> days, right?) or perhaps the rule should be "never purge R2's
> LSP if R2 is actually reachable" (seems a little dangerous...R2
> or actually a whole portion of the net might be temporarily
> unreachable to some other portion of the net, and so only some
> nodes might purge R2's LSP). On LANs (because of periodic CSNPs)
> LSP databases would get resynchronized. But maybe there might
> be some scenarios in which things could remain unsynchronized.

> If the rule is that only R2 can cause a purge of R2's LSP, then
> we have the router-with-the-fast-clock issue (assuming you
> throw away an LSP when the age expires).

Oh, purging expired LSPs is definitely fine and does ensure LSDB
synchronization.

The part of the document in question talks about purging _invalid_
received LSPs.

> And a really really stupid question. What do you mean by
> "the IP document" in the following:

>>>Section 6.2 of the IP document essentially deprecates TLV 133 and
>>>tells to use TLV 10 for authentication. This should be mentioned in
>>>this section.

> I assumed you meant RFC 1195, but there is no section 6.2 there.

this was regarding draft-ietf-isis-ip-interoperable-00.txt

Regards
Alex