Re: Routing Questions: 1 Radius, 1 OSPF

Krzysztof Adamski (kadamski@netsurf.net)
Fri, 8 Aug 1997 08:12:29 -0400 (EDT)

I have it working on both PM3 and PM2 none of them running 3.5.1b20, the
PM2 is at 3.5.1b17 and PM3 is at 3.5.1b11
The trick is to set the assign address to start at a subnet boundry and
setting the pool size to match the subnet. So for 32 addresses you have to
start at one of the following
.0
.32
.64
.96
.128
.160
.192
.224

For smaller pools you have twice as many start points.

Hope this helps.

Krzysztof

On Fri, 8 Aug 1997, Randy Moore wrote:

> Hmm, Krzysztof says it should work and MZ says it should work.
>
> I'm configured just like both suggest and it doesn't seem to work. Jon
> Lewis showed a bit of his routing table that shows exactly the problem I'm
> having.
>
> Could this be a difference in the OSPF implementations on the PM2's and
> PM3's ??? Maybe something that was fixed on COMOS 3.5.1b20, which I can't
> run on a PM2 ???
>
> - Randy Moore
>
> At 08:40 AM 8/7/97 -0400, you wrote:
> >It's very simple to get the PM to aggragate the dialin routes, for example
> >on my pm3 I have Assigned Address: xxx.xxx.249.224 (Pool Size 32)
> >which gives me:
> >xxx.xxx.249.224 27 xxx.xx.236.54 ospf/E2 ND 2 ether0
> >Yes I know I'm wasting 9 IP (32 - 23) but at a cost of having less routes.
> >And Yes all IP's work. I'm doing the same on other PM's with smaller pools
> >(one is 16) The trick is to start on a subnet boundary and to make the
> >pool just the right size.
> >
> >Krzysztof
> >
> >On Thu, 7 Aug 1997, Jon Lewis wrote:
> >
> >> On Thu, 7 Aug 1997, Randy Moore wrote:
> >>
> >> > 2) I have OSPF working fairly well, but I've noticed that each address in
> >> > my dialup pools gets advertised as a seperate route. Pretty much half of
> >> > my routing table is taken up by these "host routes". Since the pool is
> >> > always a contiguous block, it seems to me it should be possible to
> >> > advertise this entire pool as a single block, which should save some
> memory
> >> > and CPU time in each of my routers. I've read the tech notes & manuals,
> >> > but haven't seen any way to do this.
> >>
> >> Like these:
> >>
> >> 205.229.60.30 205.229.48.10 255.255.255.255 UGH 0 0 10
> eth0
> >> 205.229.60.1 205.229.48.10 255.255.255.255 UGH 0 0 9
> eth0
> >> 205.229.60.28 205.229.48.10 255.255.255.254 UG 0 0 6
> eth0
> >> 205.229.60.2 205.229.48.10 255.255.255.254 UG 0 0 14
> eth0
> >> 205.229.60.24 205.229.48.10 255.255.255.252 UG 0 0 6
> eth0
> >> 205.229.60.4 205.229.48.10 255.255.255.252 UG 0 0 27
> eth0
> >> 205..229.60.16 205.229.48.10 255.255.255.248 UG 0 0 39
> eth0
> >> 205.229.60.8 205.229.48.10 255.255.255.248 UG 0 0 37
> eth0
> >>
> >> The aggregator code in ComOS is Drain Bamaged. This pool should be
> >> advertised as 205.229.60.0/27. Perhaps the PM needs some Cisco features
> >> like being able to nail a route to null0, and then use redistribute
> >> commands and route-maps to tell the router what you want to export into
> >> OSPF.
> >>
> >> With just one PM, this isn't a big deal. If I had a stack of PM's
> >> instead of 160 ports in 3 linux boxes running gated, I'd probably be a bit
> >> more concerned about the routing table bloat.
> >>
> >> ------------------------------------------------------------------
> >> Jon Lewis <jlewis@fdt.net> | Unsolicited commercial e-mail will
> >> Network Administrator | be proof-read for $199/message.
> >> Florida Digital Turnpike |
> >> ______http://inorganic5.fdt.net/~jlewis/pgp for PGP public key____
> >>
> >
> >
> >
>
> - Randy Moore
> Atlantech Online, Inc.
> (301) 593-2794
> http://www.atlantech.net/
>