From: Bhaskar Sivanesan (bas_bharath@yahoo.com)
Date: Sat Jun 23 2007 - 05:48:33 ART
Hi Victor
Thanks for the detailed info... But think I should have posed by question still clear...
Considering your scenario where dampening is configured in R2. Suppose we have router R1 which is eBGP peered with R2, and has its loopback advertised to R2 in its BGP through network statement like below.
R1;
in loopback 0
ip address 1.1.1.1 255.255.255.0
router bgp 100
neighbor 2.2.2.2 rmeote-as 200
neighbor 3.3.3.3 remote-as 100
network 1.1.1.0 mask 255.255.255.0
R2:
router bgp 200
bgp dampening
neighbor 2.2.2.1 remote-as 100
In this scenario, my actual question is
1. When I shut the loopback interface in R1 it takes much longer time to advertise
to eBGP neighbor(2.2.2.2) than to a iBGP(3.3.3.3). It advertises to iBGP immediately but to eBGP it
takes around 10 to 15 seconds. - WHY ?
2. If I remove the network statement to cause the flap, it is even more worse.
it taking around 35 to 45 seconds for the change to get advertised. - WHY?
thanks
Bhaskar
----- Original Message ----
From: Victor Cappuccio <victor@ccbootcamp.com>
To: Bhaskar Sivanesan <bas_bharath@yahoo.com>; ccie forum <ccielab@groupstudy.com>
Sent: Saturday, June 23, 2007 8:28:54 AM
Subject: RE: BGP dampening
Hi,
if dampening is configured, any flapping networks gets assigns a penalty to
it.
BGP dampening allows you to assign penalties to flapping routes. After a
route flaps enough, the route will be suppressed, and the route flapping will
not cause repeated BGP updates. BGP dampening is configured under the BGP
process, and can be set for all BGP routes, or can be configured with a
route-map if you only want to dampen specific routes.
The default values for dampening are:
Half life 15 minutes
Suppress Limit 2000
Reuse Limit 1000
Max Suppress 60 minutes
For example
Configure R2 such that if any network it learns of is flapping, it assigns a
penalty to it. The routes should get a Suppress Limit of 3000, a half Life
value of 20 minutes, and a Reuse value of 1000. Use the default Maximum
Suppress Time.
R2(config)#router bgp 2
R2(config-router)#bgp dampening 20 1000 3000 60
we are told to keep the max suppress time the same, and we are given specific
values for the other parameters.
Lab23R2(config)#router bgp 2
Lab23R2(config-router)#bgp dampening 20 1000 3000 60
We can verify dampening by flapping one of the BGP networks. For this
example, we can use the loopback on the router announcing this network in BGP.
After we shutdown and reenable the interface, we can see that R2 has dampening
information for the route.
R2#show ip bgp 9.9.9.0
BGP routing table entry for 9.9.9.0/24, version 14
Paths: (1 available, no best path)
Flag: 0x208
Not advertised to any peer
(14) (history entry)
145.20.9.9 (metric 65) from 145.20.9.9 (145.20.9.9)
Origin IGP, metric 0, localpref 100, confed-external
** Dampinfo: penalty 994, flapped 1 times in 00:00:14
another example
if you want "10 800 1500 30" this could be read as
Half life is the "lifetime" to decay 50%. If one flap incurs a penalty of 1000
and you want the penalty to decay to 500 in 10 minutes, that's a 50% decay so
half-life would be 10 minutes. If they said they wanted penalty to be 250
after ten minutes, that's a 50% decay and a 50% decay (because 250 is 25% of
1000, 50% of 50% is 25% so you could detect that they're asking for two
half-lives to be 20 minutes, and determine that one half life would be 10
minutes).
my 2 cents
thanks,
Victor Cappuccio.-
- CCSI# 31452
Network Learning Inc - A Cisco Sponsored Organization (SO) YES! We take
Cisco Learning credits!
victor@ccbootcamp.com
http://www.ccbootcamp.com (Cisco Training and Rental Racks)
http://www.ccbootcamp.com/groupstudy.html (groupstudy member discounts!)
Voice: 702-968-5100
FAX: 702-446-8012
-----Original Message-----
From: nobody@groupstudy.com on behalf of Bhaskar Sivanesan
Sent: Sat 6/23/2007 0:11
To: ccie forum
Subject: BGP dampening
Hi Group
when testing teh bgp dampening feature, I was flapping a route by
sshutting/no shutting a loopback interface, which is advertised in BGP through
network commands.
1. When I shut the loopback interface it takes much longer time to advertise
to eBGP neighbor than a iBGP. It advertises to iBGP immediately but to eBGP it
takes around 10 to 15 seconds.
2. If I remove the network statement to cause the flap, it is even more worse.
it taking around 35 to 45 seconds for the change to get advertised.
is this a right behavior and why?
Thanks
Bhaskar
This archive was generated by hypermail 2.1.4 : Sun Jul 01 2007 - 17:24:51 ART