Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: Cisco: NSP

Packet Loss During the Backbone Network Converge

 

 

Cisco nsp RSS feed   Index | Next | Previous | View Threaded


jstuxuhu0816 at gmail

Jul 26, 2012, 6:43 AM

Post #1 of 8 (1395 views)
Permalink
Packet Loss During the Backbone Network Converge

Hi All,

Below is the network topology, i ping from CE-2 to CE-1, during i shutdown
the PE-4, have some packet drops.
CE connect to the backbone network using HSRP, the hear beat is through the
CE switch. PE-1 and PE-3 is the active gateway, PE-2 and PE-4 is the
standby one.

PE equipment running OSPF as IGP, then on the top having MP-iBGP, MPLS VPN,
MPLS TE auto-tunnel.

Normally i send traffic from CE-2 to CE-1 will use the path
"CE-2>PE-3>PE-1>CE-1". The active tunnel from PE-3 to PE-1 is directly
path, the backup tunnel is path through PE-4 to PE-2 to PE-1.
So when i shutdown the router of PE-4, the active tunnel should be no
impact, but there is some packet drops. Don't understand why.

Firstly, i think it because of the HSRP, then i ping from PE-3 to CE-1
(Under vrf) when i shutdown PE-4, but the result is also have packet drops.
So i think is because of backbone network converge. But i am not sure which
part will need to converge.

<--CE-1-->
| |
| |
| |
PE-1----PE-2
| |
| |
| |
PE-3----PE-4
| |
| |
| |
<--CE-2-->

Will appreciate for any comments.

Best Regards,
Hu Xu
_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


jstuxuhu0816 at gmail

Jul 26, 2012, 8:54 PM

Post #2 of 8 (1332 views)
Permalink
Re: Packet Loss During the Backbone Network Converge [In reply to]

Anybody have any idea?

Thanks and regards,
Xu Hu

On 26 Jul, 2012, at 21:43, Xu Hu <jstuxuhu0816 [at] gmail> wrote:

> Hi All,
>
> Below is the network topology, i ping from CE-2 to CE-1, during i shutdown the PE-4, have some packet drops.
> CE connect to the backbone network using HSRP, the hear beat is through the CE switch. PE-1 and PE-3 is the active gateway, PE-2 and PE-4 is the standby one.
>
> PE equipment running OSPF as IGP, then on the top having MP-iBGP, MPLS VPN, MPLS TE auto-tunnel.
>
> Normally i send traffic from CE-2 to CE-1 will use the path "CE-2>PE-3>PE-1>CE-1". The active tunnel from PE-3 to PE-1 is directly path, the backup tunnel is path through PE-4 to PE-2 to PE-1.
> So when i shutdown the router of PE-4, the active tunnel should be no impact, but there is some packet drops. Don't understand why.
>
> Firstly, i think it because of the HSRP, then i ping from PE-3 to CE-1 (Under vrf) when i shutdown PE-4, but the result is also have packet drops.
> So i think is because of backbone network converge. But i am not sure which part will need to converge.
>
> <--CE-1-->
> | |
> | |
> | |
> PE-1----PE-2
> | |
> | |
> | |
> PE-3----PE-4
> | |
> | |
> | |
> <--CE-2-->
>
> Will appreciate for any comments.
>
> Best Regards,
> Hu Xu
>
>
>

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


nick at foobar

Jul 27, 2012, 2:43 AM

Post #3 of 8 (1315 views)
Permalink
Re: Packet Loss During the Backbone Network Converge [In reply to]

On 27/07/2012 04:54, Xu Hu wrote:
>> So i think is because of backbone network converge. But i am not sure
>> which part will need to converge.

Looks like convergence to me. What are your OSPF timers set to? You
should probably set them to hello:1, dead:4 on most modern kit.

Nick


_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


jstuxuhu0816 at gmail

Jul 27, 2012, 4:25 AM

Post #4 of 8 (1317 views)
Permalink
Re: Packet Loss During the Backbone Network Converge [In reply to]

Using the hello-interval 1, dead-interval 4.

7600 using fast-hello, but asr9000 cannot support, that's why using the 1 &4.

Thanks and regards,
Xu Hu

On 27 Jul, 2012, at 17:43, Nick Hilliard <nick [at] foobar> wrote:

> On 27/07/2012 04:54, Xu Hu wrote:
>>> So i think is because of backbone network converge. But i am not sure
>>> which part will need to converge.
>
> Looks like convergence to me. What are your OSPF timers set to? You
> should probably set them to hello:1, dead:4 on most modern kit.
>
> Nick
>
>
> _______________________________________________
> cisco-nsp mailing list cisco-nsp [at] puck
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


oboehmer at cisco

Jul 27, 2012, 6:06 AM

Post #5 of 8 (1314 views)
Permalink
Re: Packet Loss During the Backbone Network Converge [In reply to]

> Using the hello-interval 1, dead-interval 4.
>
> 7600 using fast-hello, but asr9000 cannot support, that's why using the 1
> &4.

you should use BFD instead..

but about the issue: You stated that PE4 is not in the traffic path, but PE4 failure is still causing packet loss, then I guess you want to check if PE4 is really not in the path? How long does it take until traffic continues? Is it msecs or seconds?
This requires close troubleshooting, "debug ip routing" on all routers would be a first start (with debug timestamps and sync'ed clocks).

oli

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


jstuxuhu0816 at gmail

Jul 27, 2012, 7:29 AM

Post #6 of 8 (1316 views)
Permalink
Re: Packet Loss During the Backbone Network Converge [In reply to]

Already double confirmed with Cisco tac, the routing should be no issue. Not sure is it possible caused by the physical hardware. The 7600 using ws-6700 line card.

Thanks and regards,
Xu Hu

On 27 Jul, 2012, at 21:06, "Oliver Boehmer (oboehmer)" <oboehmer [at] cisco> wrote:

>
>> Using the hello-interval 1, dead-interval 4.
>>
>> 7600 using fast-hello, but asr9000 cannot support, that's why using the 1
>> &4.
>
> you should use BFD instead..
>
> but about the issue: You stated that PE4 is not in the traffic path, but PE4 failure is still causing packet loss, then I guess you want to check if PE4 is really not in the path? How long does it take until traffic continues? Is it msecs or seconds?
> This requires close troubleshooting, "debug ip routing" on all routers would be a first start (with debug timestamps and sync'ed clocks).
>
> oli

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


alumbis at gmail

Jul 28, 2012, 6:27 AM

Post #7 of 8 (1297 views)
Permalink
Re: Packet Loss During the Backbone Network Converge [In reply to]

How many routes and tunnels are on the PEs?

Have you modified any of the TE timers from their defaults?

On Thu, Jul 26, 2012 at 9:43 AM, Xu Hu <jstuxuhu0816 [at] gmail> wrote:
> Hi All,
>
> Below is the network topology, i ping from CE-2 to CE-1, during i shutdown
> the PE-4, have some packet drops.
> CE connect to the backbone network using HSRP, the hear beat is through the
> CE switch. PE-1 and PE-3 is the active gateway, PE-2 and PE-4 is the
> standby one.
>
> PE equipment running OSPF as IGP, then on the top having MP-iBGP, MPLS VPN,
> MPLS TE auto-tunnel.
>
> Normally i send traffic from CE-2 to CE-1 will use the path
> "CE-2>PE-3>PE-1>CE-1". The active tunnel from PE-3 to PE-1 is directly
> path, the backup tunnel is path through PE-4 to PE-2 to PE-1.
> So when i shutdown the router of PE-4, the active tunnel should be no
> impact, but there is some packet drops. Don't understand why.
>
> Firstly, i think it because of the HSRP, then i ping from PE-3 to CE-1
> (Under vrf) when i shutdown PE-4, but the result is also have packet drops.
> So i think is because of backbone network converge. But i am not sure which
> part will need to converge.
>
> <--CE-1-->
> | |
> | |
> | |
> PE-1----PE-2
> | |
> | |
> | |
> PE-3----PE-4
> | |
> | |
> | |
> <--CE-2-->
>
> Will appreciate for any comments.
>
> Best Regards,
> Hu Xu
> _______________________________________________
> cisco-nsp mailing list cisco-nsp [at] puck
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


jstuxuhu0816 at gmail

Jul 28, 2012, 8:56 PM

Post #8 of 8 (1304 views)
Permalink
Re: Packet Loss During the Backbone Network Converge [In reply to]

Yes, i configure the reoptimize 25.
The PE just have 20 subnet routes, Active Tunnel is four, backup Tunnel is
also four, then about 4 mid-point Tunnel.

2012/7/28 Pete Lumbis <alumbis [at] gmail>

> How many routes and tunnels are on the PEs?
>
> Have you modified any of the TE timers from their defaults?
>
> On Thu, Jul 26, 2012 at 9:43 AM, Xu Hu <jstuxuhu0816 [at] gmail> wrote:
> > Hi All,
> >
> > Below is the network topology, i ping from CE-2 to CE-1, during i
> shutdown
> > the PE-4, have some packet drops.
> > CE connect to the backbone network using HSRP, the hear beat is through
> the
> > CE switch. PE-1 and PE-3 is the active gateway, PE-2 and PE-4 is the
> > standby one.
> >
> > PE equipment running OSPF as IGP, then on the top having MP-iBGP, MPLS
> VPN,
> > MPLS TE auto-tunnel.
> >
> > Normally i send traffic from CE-2 to CE-1 will use the path
> > "CE-2>PE-3>PE-1>CE-1". The active tunnel from PE-3 to PE-1 is directly
> > path, the backup tunnel is path through PE-4 to PE-2 to PE-1.
> > So when i shutdown the router of PE-4, the active tunnel should be no
> > impact, but there is some packet drops. Don't understand why.
> >
> > Firstly, i think it because of the HSRP, then i ping from PE-3 to CE-1
> > (Under vrf) when i shutdown PE-4, but the result is also have packet
> drops.
> > So i think is because of backbone network converge. But i am not sure
> which
> > part will need to converge.
> >
> > <--CE-1-->
> > | |
> > | |
> > | |
> > PE-1----PE-2
> > | |
> > | |
> > | |
> > PE-3----PE-4
> > | |
> > | |
> > | |
> > <--CE-2-->
> >
> > Will appreciate for any comments.
> >
> > Best Regards,
> > Hu Xu
> > _______________________________________________
> > cisco-nsp mailing list cisco-nsp [at] puck
> > https://puck.nether.net/mailman/listinfo/cisco-nsp
> > archive at http://puck.nether.net/pipermail/cisco-nsp/
>
_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

Cisco nsp RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.