Warning message

  • Cisco Support Forums is in Read Only mode while the site is being migrated.
  • Cisco Support Forums is in Read Only mode while the site is being migrated.


Unanswered Question
Jun 18th, 2003
User Badges:

dear all,

refer diagram below:

PE A -- P A -----CORE CLOUD ----- P B ---- PE B




PE A and P A located remotely. P B and PE B are at inside the core and close to RR C and RR D

Few places, we used c3640 as a PE and c72xx as P. RR is c75xx

PE A and PE B are client for RR C. (ipv4 and vpn4)

P A and P B are client for RR D (normal ipv4)

Problem is PE A always lost bgp conenction with RR C. show logg shown BGP-5-ADJCHANGE many times and show ip bgp sum, up/down session is very short time. If backbone or WAN contribute to lost bgp session, P A up/down is very stable and not down for long long time.

is this because RR C?. I don't think so PE B did not show bgp lost logg.

Maybe because ios, mp-bgp config or else?

Can someone assist me. And is there anyone that use c3640 as PE. can you let me know which IOS that u are using.



  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
ipotts Thu, 06/19/2003 - 00:33
User Badges:

Could it be that the keepalive and holdtime are lower between PEA and RRC than PA and RRC. Please try sh ip bgp nei {RRC_routerid_ipaddress} on PEA and PA and check these values,



nshahrin Thu, 06/19/2003 - 00:51
User Badges:

Hi Ian,

Thanks for reply.

PE A and PE A are provider edge and ingress router. Both of them use RR C as a reflector.

Meanwhile, P A and P B are LSR and they use RR D.

the keepalive and holddown value should be standard and default value. I never change that value.

my next plan is try to change the IOS. what do u reckon


ipotts Thu, 06/19/2003 - 02:30
User Badges:

The only other thing is whether PE A is directly connected to the circuit that is failing. Therefore, its neighbor will drop due to factors other than the BGP keepalive (i.e. losing carrier where BGP will drop immediately, or the interface keepalive),



nshahrin Thu, 06/19/2003 - 17:33
User Badges:

Hi Ian,

Thanks for your feedback on this. Really appreciate it :)

PE A is directly connected to P A using FE. No error shown for this circuit.

That's why I believe our WAN is not the issue because P A has very stable bgp session with RR D.

I already informed TAC and they suggested to upgrade the IOS. let me update you the result.


melkomy Sun, 07/13/2003 - 07:15
User Badges:

That same problem i was suffering from in our network but the difference was that we're using MGX 8850 with RPM-PR acting as the ATM Edge LSR.....while BPX 8600 with Router 7200 (External LSC) acting as the ATM LSR.

That MP-BGP flapping behavior between the PEs was solved by IOS upgrade as there was a bug in the IOS we were using which was affecting the TCP session upon which the BGP session established.



nshahrin Sun, 07/13/2003 - 16:58
User Badges:

Hi Komy,

Yes, and just to share with others, after upgraded the IOS (12.2(11) to latest 12.2(15T5), it solved the problem ... cannot afford to face same problem in production network again in future ...

thanks everyone ...

melkomy Sun, 07/13/2003 - 23:19
User Badges:

Hi all,

and also just to share with others the bug ID that was causing IBGP flapping is CSCdy57048 and as you said it's fixed in 12.2 (15)T5


This Discussion