[c-nsp] ASR9k LC fib programming problem

[Šturmankin Miroslav]

Hi James,

yes, it's an old code but we were running it without a single problem or bug for 2y25w now and this is the first problem we've encountered (if it ain't broke, don't fix it). An upgrade will reload the chassis so yes, it will fix the issue for sure.

I did not try any of the disruptive method, yet, as clearing the CEF or restarting RIB as this is a core box doing 40G+ traffic. But if I don't find a way to clear the single CEF entry from LC, I will have to schedule a window. Was the "process restart ipv4_rib" really without a dropped packet? Thanks.

Regards/S pozdravom,

Miro Sturmankin


-----Original Message-----
From: cisco-nsp [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of James Bensley
Sent: Friday, July 14, 2017 11:25 AM
To: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] ASR9k LC fib programming problem

On 13 July 2017 at 16:02, Šturmankin Miroslav <miroslav.sturmankin at swan.sk> wrote:
> router specifications:
> ASR9006 IOS XR 4.3.4 SP4
> A9K-RSP-4G
> A9K-MOD160-TR + A9K-MPA-8X10GE & A9K-MPA-4X10GE

That is a an old IOS-XR version which I urge you to upgrade from, it's full of bugs, as you can see, I suggest you upgrade and the problem with probably go away.

Have you tried to restart the IPV4 RIB process? "process restart ipv4_rib"

That should cause the route to be re-downloaded to the card when the process restarts without a forwarding interruption. It won't clear routes from the card but when the process starts up again it will check what is on the card vs. what is in the central FIB and reprogram any miss-matches. We had a race condition on 4.3.4 [default through to SP4] with Inter-AS L3VPN routes not being programmed into the line card when there was a next-hop change, restarting that process sometimes fixed the problem.

If you route is also a labelled entriy you can try "process restart mpls_lsd" which will have same affect as above, if the label is wrong in a route when the process starts up again it should re-program the route into the line card.

You said you have cleared the route from the line card but that didn't worked, can you arrange a maintenance window and clear the entire CEF table on the card? "clear cef linecard location 0/1/CPU0" that will cause a "blip" outage as all routes are dropped from the card and then re-programmed.

Really though, in my opinion, an XR upgrade is needed here as the long term fix.

Cheers,
James.
_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/