EIGRP distribute-lists with route-maps

Good evening all,

I was doing some practise labs tonight and playing around with EIGRP.
I'm trying to grasp a better understanding of the limits of route-maps
when used with distribute-lists in EIGRP.

For the sake of simplicity, let's imagine a topology with three routers
all connected over a shared, broadcast media (Ethernet).

R2 & R3 are the upstream routers both advertising a network 200.2.2.0/24

R4 is the downstream router with EIGRP neighbour adjacencies to both
these upstream routers. As a result, R4 is doing equal cost
load-balancing between both upstream routers for 200.2.2.0/24.

sh ip ro | s 200.2.2.0
D 200.2.2.0/24 [90/1024640] via 10.1.1.3, 00:00:16, Ethernet0/0
                    [90/1024640] via 10.1.1.2, 00:00:16, Ethernet0/0

The requirement was to have R4 prefer the route via R2, unless R2 was
unavailable, in which case it would route through R3.

- Configuration can only be done on R4.
- No PBR (Policy Based Routing).
- No manipulation of the Administrative Distance (AD) allowed (This was
actually the easy answer, but I wanted to try and find an alternate method).

So we can use route-maps in conjunction with distribute lists and I
guess this is where my confusion currently is. Here's what I tried to do:

1. Access list to match the route:

access-l 44 permit 200.2.2.0 0.0.0.255

2. Access list to match one of the route-sources (R3):

access-l 3 permit host 10.1.1.3

3. Route-map configuration. I'm trying to bump up the metric of a route
by 1000 so it's no longer the preferred route:

route-map PREFER-R2 permit 10
  match ip addr 44
  match ip route-source 3
  set metric +1000
route-map PREFER-R2 permit 20

4. Apply it to the EIGRP routing instance:

router eigrp AS100
  addr ipv4 uni auto 100
   topo base
    distrib route-map PREFER-R2 in

And it didn't end up working. So clearly my intended use of the
route-map in the above scenario is invalid. Probably fine for when
redistributing, but not when used in conjunction with distribute lists.
If I add a different "set" parameter (one which seems to be commonly
referred to in the configuration guide; a route tag), then I see that it
gets applied:

sh ip ro 200.2.2.0
Routing entry for 200.2.2.0/24
   Known via "eigrp 100", distance 90, metric 1024640
   Tag 200, type internal
   Redistributing via eigrp 100
   Last update from 10.1.1.3 on Ethernet0/0, 00:00:13 ago
   Routing Descriptor Blocks:
   * 10.1.1.3, from 10.1.1.3, 00:00:13 ago, via Ethernet0/0
       Route metric is 1024640, traffic share count is 1
       Total delay is 1002 microseconds, minimum bandwidth is 10000 Kbit
       Reliability 255/255, minimum MTU 1500 bytes
       Loading 1/255, Hops 1
       Route tag 200
     10.1.1.2, from 10.1.1.2, 00:00:13 ago, via Ethernet0/0
       Route metric is 1024640, traffic share count is 1
       Total delay is 1002 microseconds, minimum bandwidth is 10000 Kbit
       Reliability 255/255, minimum MTU 1500 bytes
       Loading 1/255, Hops 1

I'm looking through the configuration guide and the CLI reference to try
and work out what match criteria and set commands are available when
route-maps are used under different contexts, but I'm not having much
luck finding anything thorough or concrete. When you think about it, it
does kind of make sense, distribute-lists are for filtering not routing
metric manipulation but since route-maps allow for so much manipulation
of routes an understanding of what parameters you can use under what
scenarios would be great.

Ideas or resources?

Thanks.

Billy

Blogs and organic groups at http://www.ccie.net
Received on Sun Jul 27 2014 - 00:18:45 ART