Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

New Member

Policy Based routing issues with EIGRP

I have 2 sites linked by an 64kB ISDN-leased line and standard eth 100MB.

My hardware are Cisco 3620 and Cisco 1750 for the leased-line and a Cat 3550 with EMI Image.

Use a Cisco 3620 for the home office and Cisco 1750 with Cat 3550 for remote.

I need to implement PBR to move Citrix across the leased-line and all other via eth with backup functionality.

I have configured EIGRP which works fine.

I use IOS 12.1(16f) on cisco 3620 with fast-switched pbr and

IOS 12.2(11)T9 on Cisco 1750 also fast-switched.

The Catalyst 3550 with EMI IOS 12.1(14)EA1a works cef based by default and can`t disabled.

After placing the route-map to my routed interfaces on Cisco 3620 and Cat 3550 the packets will forwarded

to the next-hop address. But the backup funktion via EIGRP does not work !

The failover from pbr match criteria to normal routing EIGRP does not work.

I also tested multiple next-hops on Cisco 3620 or interfaces and sub sequences in the route-map

for example, but nothing.

Does anybody know which IOS better supports pbr features for my hardware ?

Thanks for help

Best regards



Re: Policy Based routing issues with EIGRP

Could you post your policy config?




New Member

Re: Policy Based routing issues with EIGRP

My config`s :

Building configuration...

Current configuration : 1670 bytes

[snipped...] NOTE from Moderator: Config removed due to size. Please use the attachment feature to post large config info.


Re: Policy Based routing issues with EIGRP

Is the route to removed from the 3620's routing table when you initiate an ISDN failure? If so, standard destination-based routing should take over and you probably have an EIGRP issue. I'm not sure when BRI interfaces do and don't change state to 'down' -- if the interface is staying up for some reason, the route won't be removed.

However, it looks like you might have a bigger problem in the other direction (remote -> home) if I'm interpreting the configs correctly. It doesn't look like the 3550 has any way to know when the ISDN line goes down; hence, it will always policy route the Cytrix traffic to the 1750. The 1750 will then route the traffic back to the 3550, causing a routing loop.

This is a common problem: the routing protocol detects the failure but PBR doesn't, so they get out of synch and the result is usually routing problems. A fix was just introducted in 12.3(4)T: It allows you to track PBR next-hops via ICMP. In your case, if the BRI interface goes down PBR on the 3550 can detect this when the pings start timing out.

So you may have different issues in both directions, or it may just be the remote -> home direction that's experiencing problems.

New Member

Re: Policy Based routing issues with EIGRP

Hi tbaranski

You are right, the route won`t be removed because the int bri state is spoofing and not

down !

When I shutdown the int bri by hand, route will be removed and the traffic is routed recording EIGRP via

fastethernet 0/1.

The problem to be resolved is indeed a case for rtr which enables the router to detect the dead (spoofing) bri link.

Reading the related document about rtr in 12.3(4)T

I tried to configure rtr on Cat3550 12.1(14)EA1a in the same manner. There I discovered some differences in configuring rtr on Catalyst.

Do have any information regarding rtr configuration on it, because I wasn`t able to find some advice.

Looking forward for your advice.




Re: Policy Based routing issues with EIGRP

Ah, it didn't occur to me that it was the 3550 you'd be configuring PBR tracking on. As this feature just came out in the most recent IOS T-Train release, it's not going to be supported on the 3550 yet.

So, this complicates things substantially. The only solution I can think of offhand is for you to connect another Ethernet interface on the 1750 (if you have a spare) to an isolated VLAN on the 3550. You can put a floating static route on the 1750 so that it routes traffic to the 3550's isolated VLAN when the ISDN link goes down (the only traffic the 1750 should ever see is the policy routed traffic). The 3550 will then send the traffic over the link (assuming the destination is in EIGRP) because you're not policy routing on the 3550's isolated VLAN interface.

But this is somewhat ugly, and doesn't fix the issue of BRI interfaces not going down. Perhaps someone else can comment on how to work around this (there has to be a way), as I'm not much of an ISDN person.

New Member

Re: Policy Based routing issues with EIGRP

Dear tbaranski

Thanks for your advice.

Your tips with floating static routes are very helpful, but on the 1750 I have no spare interfaces

in the moment, so I can`t configure it !

This Lab was for future time where ISDN is not present for the connect between the Cisco 3620 and the 1750. So my only problems are the routing loops

which occures when I set ISDN to down state.

So what can I do ?

My new (idea) config are attached.



Re: Policy Based routing issues with EIGRP

Can you describe your idea? It's difficult to tell just by looking at configs.

This may not be a very helpful suggestion, but your best bet may end up being to use the Ethernet link as the primary path for everything (including Cytrix).