How to confirm this is a TTL decrement issue/BGP losing route

news2010a · ‎06-15-2008

Folks, I connected a Cisco WAAS appliance to a remote site, inline, router--shaper---Cisco WAAS--internal switch.

Initially communications worked fine.

Then users on remoteSite reported sporadic application issues.

I've seen a behavior before in which the Cisco WAAS box by default decrements the TTL once traffic goes through, thus causing the BGP routes to go missing.

So here is my question:

If I want to validate that such TTL decrement issue is what is causing this project, which bgp command should I use? I do not have access to the Remote branch (RouterB and RouterC) configuration at this time.

On SwitchA I attempted to monitor with 'debug ip bgp events' but I have not seen anything relevant.

Please advise on how to approach this. Once I power off the WAAS (and the circuit closes) all communication goes back to normal.

If I need to solve this with the ebgp multihop command, in which router exactly I should enter the ebgp multihop command?

Please see attached diagram.

n.nandrekar · ‎06-15-2008

Hi!

I dont think that ebgp multihop would solve the problem. If multi-hop was the problem, your bgp session wouldnot have come up / would have gone down. Since you are saying that nothing relevent is seen in debug ip bgp, I assume that the bgp neighborships are up and just the routes are withdrawn.

I would suggest you use "debug ip bgp updates" This could probably point in the direction of why the routes are withdrawn.

Even with WAAS on, you are not able to ping switch B. But are you able to ping router B and router C?

Which routes go missing? does the network of switch B just go missing, or does router C and Router B networks also go missing?

Regards,

Niranjan

news2010a · ‎06-15-2008

Just note that the BGP session was up and running before the WAAS was introduced. So once I put the WAAS, it kept going up/down sporadically.

Unfortunately I need to wait for next maintenance window in order to test your suggestion.

Assuming this is a multi-hop problem, as long as this can be fixed by introducing the ebgp multihop command from my end that's fine. I mean, I just hope that I do not need to mess with the service provider cloud at all to get this fixed (I don't think that is the case).

michaelchoo · ‎06-16-2008

From your original post, you suggested that when you powered down the WAAS, all comms came back as normal. I may be pointing out the obvious here, but, sounds like the WAAS is the source of the issue, hence you need to look closely into that. I don't know, maybe log a fault with TAC? What's the point of mucking around with your BGP config and all that when you've isolated the issue being WAAS? I've never used WAAS so I don't know what it can or can't do. However, one thing that popped in my head: is there a way to configure WAAS to simply "pass through" all TCP 179 traffic untouched? If there is a way to do that, it might solve your BGP flapping issue.

news2010a · ‎06-17-2008

In reality I confirmed that this version of WAAS IOS no longer decrements TTL. Moreover, I got the run config in the router and the router is running OSPF, not BGP.

I've got TAC involved. I will wait for maintenance window day now.

Thanks.

ajchobany · ‎06-25-2008

Yes, the WAAS box will count as a hop between your bgp peers. You will need to configure ebgp multihop on your neighbor statements on both sides.