Change BGP policy take affect immediately

[napster1987]

Hi everyone,

I just update my cisco 7609 to Version 12.2(33)SRD6. I encounter a strange problem with this version, everytime i change BGP policy ( input or output ) this will take affect immediately without "clear ip bgp neighbor <address> soft". Are there anyway not to take BGP policy affect unless command "clear ip bgp neighbor <> soft" ?

Thanks

[ameya_oke]

Hi Phuong,

The route refresh capability presents solution to ur problem by allowing the local router to request a copy of the  database of its peer.

Any peer that is willing to participate in this operation should advertise its route refresh capability during the session establishment with its peer using BGP Capabilities advertisement.

When you change your inbound policy configuration the local speaker will send a Refresh-Request message with a specific pair asking the peer to re-advertise updates.

http://www.networkers-online.com/blog/2010/03/bgp-route-refresh-capability/

http://www.ietf.org/rfc/rfc2918.txt

http://www.cisco.com/en/US/docs/ios/12_2/ip/configuration/guide/1cfbgp.html

Please rate helpful posts

Cheers!!

Ameya

[napster1987]

Hi Ameya,

Unfortunately i only control router at my side ( beacause i'm peering with another AS). And My customer router don't support Route refresh Capability yet.

Maybe you don't understand my problem clearly. My problem is when i change BGP policy it will affect immediately to my routing table. Before that version i use version 122-33.SRD2a and it don't behave like this, BGP policy only affect when i type "clear ip bgp neighbor <...> soft".

Is it a bug or just a small change of version 12.2-33-SRD6? If this is a change how to fix it

Thanks for your fast answer

[ameya_oke]

Hi ,

Can you post just the bgp config?

Ameya

[napster1987]

Hi Ameya,

Here is my configuretion

HCM-EPZ#show ip bgp summary

BGP router identifier 118.69.255.1, local AS number 18403

BGP table version is 25940184, main routing table version 25940184

377188 network entries using 46016936 bytes of memory

1740212 path entries using 90491024 bytes of memory

118716 multipath network entries and 237432 multipath paths

135278/63511 BGP path/bestpath attribute entries using 10281128 bytes of memory

69 BGP rrinfo entries using 1656 bytes of memory

119275 BGP AS-PATH entries using 3372198 bytes of memory

1785 BGP community entries using 98416 bytes of memory

0 BGP route-map cache entries using 0 bytes of memory

1557 BGP filter-list cache entries using 18684 bytes of memory

BGP using 150280042 total bytes of memory

Dampening enabled. 366 history paths, 1257 dampened paths

873568 received paths for inbound soft reconfiguration

BGP activity 519945/142753 prefixes, 23074476/21334260 paths, scan interval 60 secs

Neighbor        V    AS MsgRcvd MsgSent   TblVer  InQ OutQ Up/Down  State/PfxRcd

121.189.2.73    4  4766 1702256   24141 25939676    0    0 2w0d       359636

router bgp 18403

neighbor 121.189.2.73 remote-as 4766

neighbor 121.189.2.73 description *** KT AS4766 ***

neighbor 121.189.2.73 password 7 002520474E1F4245

neighbor 121.189.2.73 version 4

address-family ipv4

  neighbor 121.189.2.73 activate

  neighbor 121.189.2.73 remove-private-as

  neighbor 121.189.2.73 soft-reconfiguration inbound

  neighbor 121.189.2.73 prefix-list rcv-filter-in in

  neighbor 121.189.2.73 route-map KT-121.189.2.73-in in

  neighbor 121.189.2.73 route-map AS4766-Routes-Export out

  neighbor 121.189.2.73 dmzlink-bw

Thanks

Phuong Luong

[Marwan ALshawi]

Are You aware of this:

Soft reconfiguration allows policies to be configured and activated without clearing the BGP session.

Performing inbound reconfiguration enables the new inbound policy to take effect

[Peter Paluch]

Marwan,

Good catch! I also believe that the Soft Reconfiguration might be at the heart of this. However, personally I do not recommend configuring the Soft Reconfiguration anymore - it was an ugly and naive (then again, the only available) solution to the BGP's lack of ROUTE-REFRESH message, but that message is universally supported now.

Best regards,

Peter

[Giuseppe Larosa]

Hello Phuong,

what you see is normal in Cisco BGP implementation I have seen the same on a C7609-S with RSP-720 and 12.2(33)SRD

I cannot say more as I'm not a Cisco employee and I don't know all the internals of IOS BGP code.

but my results are the same as yours if this makes you more confident you may expose your results to your boss, otherwise wait for other colleagues' comments on this (recommended)

Hope to help

Giuseppe