From: Brian Hescock (bhescock@xxxxxxxxx)
Date: Fri Jan 12 2001 - 10:45:37 GMT-3
I concur, I would never suggest using 0.0.0.0 255.255.255.255 (it was just an
example to show the network statement has nothing to do with the network mask
that's advertised. Avoid using "redistribute connected" as well, there have
been some bugs with that and it also causes the routes to be external routes.
That can cause a problem on routers downstream where certain routes aren't
injected into the routing table. The reason for this is if you look at the
lsa for the route you'll see "forwarding address" (not next hop). The ip
address shown must be known via an intra or inter-area ospf route or the route
isn't injected into the routing table. The solution is to use the ospf
network statement instead of "redistribute connected".
Example:
int e 0
ip add 100.1.1.1 255.255.255.0
router ospf 1
redist connected subnets metric 100
(nothing else shown, not pertinent)
On a downstream router let's say I should have network 20.1.1.1 and it's not in
the routing table. I take a look at the ospf database and the lsa doesn't have
the routing bit set (if it's set you would see "routing bit for this lsa is
set" (or something very similiar) on the first line of the lsa. It's not
shown if it isn't set). Look further down into the lsa and you'll see
"forwarding address: 100.1.1.2" (this is not the same as the next-hop ip
address).
100.1.1.0 network is an external route because of redistribute connected (not
intra or inter area ospf route) so the routing bit isn't set. Change
"redistribute connected" to "network 100.1.1.1 0.0.0.0 area x" and it will then
work.
Don't worry if this doesn't make much sense, just avoid "redistribute
connected" whenever possible and you typically won't run into this problem.
Brian
David Ankers wrote:
> Becareful with "network 0.0.0.0 255.255.255.255" there are some IOS bugs
> known with that. Also IMHO before attempting the lab it would be a good idea
> to know subnetting and reserve masks back to front in your sleep.
>
> D.
>
> On Friday 12 January 2001 03:51, Brian Hescock wrote:
> > Harbir,
> > If you really want a timesaving way of doing ospf (if only a few
> > interfaces), and less chance of making an error, use the ip address from
> > the interfce and use wildcards bits of 0.0.0.0. No thinking
> > involved. Most people incorrectly think you need the wildcard bits to
> > represent what the network mask is that will be advertised (hey, I used to
> > think that... ;-) All the network statement does is tell ospf which
> > interface to turn ospf on, you're not giving it a mask at all. The mask
> > is taken from the interface itself. The only advantage of the wildcards
> > bits is you can turn ospf on more than one interface at a time with fewer
> > commands. Example: network 0.0.0.0 255.255.255.255 We're not
> > advertising network 0.0.0.0, we're turning ospf on each interface on the
> > router because of the wildcard bits.
> >
> > Brian
> >
> > On Thu, 11 Jan 2001, Harbir Kohli wrote:
> > > Hello
> > > I found a couple of time-savings shortcuts in IOS 12.0.
> > > If you wanted to quickly see the config of just an interface so
> > > show run int s0 (works in 12..1.5 but not in 11.2.24)
> > > OSPF inverse mask
> > > network 10.10.10.0 255.255.255.0 (works in 12.1.5 and 11.2.24)
> > > the router will automatically calculate the inverse mask. I actually
> > > use it even to write access-lists by creating an extra OSPF process and
> > > later deleting it, saves me trying to figure out invese masks.
> > >
> > > Enjoy
> > >
> > > Harbir
> > >
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:27:28 GMT-3