[j-nsp] ISIS Routing Problem

[Dan Evans]

Eric,

Check out section 3.10.2 of RFC 1195 as I believe it answers your question
on why L1 routes are always preferred to L2 routes. There's also additional
information in RFC 5302 Section 3.



On Wed, Jun 16, 2010 at 9:56 PM, Eric Van Tol <eric at atlantech.net> wrote:

> I do need L2 enabled, as they are the two core routers, and as such, are
> the backbone of the network.  It's a good suggestion, though.  I'll probably
> end up just going the static route way, at least until I can swap these out
> with two almost-out-of-service Sup720s that can run some decent code.
>
> As for the L1 route preference, that's what I don't understand.  If R1/R2
> are getting each other's loopbacks through L2 with a preference of 18, but
> then I swap the L1/L2 preferences so that L2 now has a pref of 15, why would
> the L1 route always get preferred?
>
> -evt
>
> From: Dan Evans [mailto:pzdevans at gmail.com]
> Sent: Wednesday, June 16, 2010 9:05 PM
> To: Eric Van Tol
> Cc: juniper-nsp
> Subject: Re: [j-nsp] ISIS Routing Problem
>
> Eric,
>
> Since R1 and R2 are L1/L2 routers they'll each always prefer the L1 route
> over the L2 route due to default route preference. It's an interesting
> situation for sure. Removing the loopback from L1 isolates R1 and R2 from
> advertising their loopbacks to R3 and R4, but with the loopback enabled for
> Level 1 you'll always get the long path for R1 to R2 traffic.
>
>  Do you require R1 and R2 to have L2 enabled on the R1--R2 link? You could
> turn off L2 on that link and it should give you the result I think you're
> expecting. R1 and R2 and still be L1/L2 or L2 only with any other upstream
> routers to keep L2 continuity with any routers not in your original
> topology.
>
>
> On Wed, Jun 16, 2010 at 7:28 PM, Eric Van Tol <eric at atlantech.net<mailto:
> eric at atlantech.net>> wrote:
> Hi all,
> I'm scratching my head over this one and I'm sure the answer is very
> simple.  I have four routers:
>
> R1 --- R2
> |      |
> |      |
> R3 --- R4
>
> R1 and R2 are L1/L2 routers.  R3 and R4 are L1-only routers.  Due to IOS's
> inability to do MD5 authentication at level 2, I cannot make R3 and R4 L1/L2
> routers.  R1 and R2 are running 9.6R3.8.  Now the problem...
>
> I have 'interface lo0.0 passive' configured on both R1 and R2.  The
> loopback is being injected into the L1 level, then being re-injected back
> into the L2 level when it's seen from R3 and R4.  I've tried messing with
> the preference values for level 2 on R1 and R2, but the loopbacks are always
> being preferred through the L1 level.  If I configure 'interface lo0.0 level
> 1 disable' on R1 and R2, the loopbacks disappear, but then I end up with BGP
> recursive route lookups from R3 and R4.  I'd prefer not to configure static
> routes for the R1 and R2 loopbacks on R3 and R4, but that seems to be my
> only recourse at this point.
>
> Am I missing something simple here or am I just going to have to do the
> static routes on R3 and R4, redistribute those between R3 and R4, but deny
> their redistribution up to level 2?
>
> Thanks,
> evt
>
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net<mailto:
> juniper-nsp at puck.nether.net>
> https://puck.nether.net/mailman/listinfo/juniper-nsp
>
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp
>