[c-nsp] Purely Academic: Router swap and EIGRP doesn't work

Pavel Skovajsa pavel.skovajsa at gmail.com
Fri May 7 05:15:27 EDT 2010


strange things happen all the time. Same IOS version as the one on
3662? Maybe a bug on that version that affects only 3620, which would
be strange as 3620 and 3662 have same vendor of RISC processor, just
different clocking.

Basically, from the programmers viewpoint reinserting the network
statement causes a logical restart of some EIGRP subroutines on
specific interfaces (those that are matching the network statement),
in other words, the router forgets totally about the presence of EIGRP
on them and then learn them from scatch again.
Generaly speaking logic dictates that something must have changed,
that changed the EIGRP behavior on this interface. The problem is is
that it does not have to be a config change, it can be any state
change ranging from:
1) The interfaces (or int drivers) were originally not up while the
network statement was parsed
.
.
.
100) Bad position of Jupiter and Venus

-pavel

On Fri, May 7, 2010 at 1:08 AM, Rick Kunkel <kunkel at w-link.net> wrote:
> Ran in to an issue yesterday and today and got it resolved, but I'm
> wondering the why...
>
> We had a 3662 configured with EIGRP and everything working fine.  We wanted
> to put in a 3620 for a few days to use the 3662 in our lab, so we
> identically configured it (with a few interface references changed), and
> mounted it in the rack.  When ready, we quickly moved the WAN and LAN cables
> to the "newer" 3620.
>
> But stuff didn't work.  The router could access anywhere through it's
> default route.  It's upstream EIGRP neighbor was listed.  But equipment on
> the LAN side couldn't access the WAN side. I logged into the upstream
> router, and saw the same neighbor relationship.  However, the topology was
> missing.  The upstream wasn't hearing the routes from the 3620.
>
> We'd spent quite a bit of time getting to this point, but now that I knew
> that, I went into the 3620, removed a network statement and put it back, and
> -- viola! -- that block routed now.  I did the same for the rest, and all
> was fixed.
>
> My question is, what causes this?  The neighbor relationship was working,
> but the upstream didn't see the routes from the 3620.  Did it have to do
> with changing the cables quickly?  I'd chalk it up to a one-time fluke, but
> we did it over and over yesterday, and it never worked.  Furthermore,
> whenever we switched it BACK to the 3662 that we were trying to pull out, it
> worked instantly....
>
> Purely academic at this point...
>
> Thanks,
>
> Rick
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>



More information about the cisco-nsp mailing list