If it turns into a real mystery, look at the OSPF router ID's for
each router. If there are duplicates or the id picked belongs to
an interface that's down, routes don't propagate although everything
looks normal. If this happens to be it, just create a loopback on
each router with an IP that's likelty to be higher than any that
you actually use on an interface.
I don't know *why* this matters, but several of our "OSPF is broken
again, and we can't figure out why" hair-pulling episodes have turned
out to be router-id related. Why cisco can't put in an explict
router-id directive in is beyond me.
George
> Date: Thu, 9 Dec 1999 16:27:27 +0000
> From: Dave Wilson <dave.wilson@heanet.ie>
> To: cisco-nsp@puck.nether.net
> Subject: [nsp] OSPF funny in 12.0(5)S
> Resent-From: cisco-nsp@puck.nether.net
>
> Hi there,
>
> Any suggestions on this?
>
> Four Cisco 7507's, running 12.0.5(S), and four 7206's running the same.
> One of each per site. Each 7507 has:
>
> * E3 connections to two of the others, so the sites are
> arranged in a ring.
>
> Three of the 7507s have:
>
> * A local Packet Over Sonet connection to a 7206 at each site
>
> * A fast ethernet connection to that same 7206, a 3640, and a 2610
>
> Therefore each 7206 is reachable from the rest of the network only
> through its local 7507.
>
> We're running OSPF on all the interfaces, area 0. Every router shows
> all its neighbours - two via E3, one via POS, and one (or more) via Fast
> Ethernet.
>
> Every 7206 sees the entire routing table, including all three ethernets.
> However, *one* of the 7507's - not the one missing a 7206 - is not showing
> any routes learned over the E3 lines in a "sh ip route". The routes learned
> over POS are shown (only one at the moment, and that's a backup route to
> the local ethernet, but it's working nevertheless). The routes learned
> via the E3s are passed on to its local 7206. But they do not appear in
> the output of a "sh ip route".
>
> I did have incorrect masks on my loopbacks, which were fixed, but didn't
> solve the problem. We've rebooted every router in the network; didn't
> solve. We pulled out one of the RSP's in the 7507, and rebooted every
> router again; it showed the same problem. It affects both E3 lines,
> but there seems to be no physical problem with either of them. So it
> doesn't look like hardware.
>
> ip cef came turned on by default on the 12.0 config. We tried turning
> it off. No effect.
>
> FWIW, this is *very* similar to what happens in BGP if you forget to type
> "no synchronisation" into your BGP config - but there the similarity ends.
> As far as I know, OSPF just shouldn't ever behave like that, or even fail
> like that.
>
> Confused,
> Dave
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:08 EDT