BGP Over VPn reset Exactly every three mins

wrobbin · ‎07-07-2011

From my Main Router

Code

c2800nm-advsecurityk9-mz.124-15.T3.bin

Jul 7 09:53:53: %BGP-5-ADJCHANGE: neighbor 10.10.242.130 Down BGP Notification sent

Jul 7 09:53:53: %BGP-3-NOTIFICATION: sent to neighbor 10.10.242.130 4/0 (hold time expired) 0 bytes

Jul 7 10:03:30: %BGP-5-ADJCHANGE: neighbor 10.10.242.130 Up

Jul 7 10:07:36: BGP: 10.10.242.130 open failed: Connection timed out; remote host not responding, open active delayed 34483ms (35000ms max, 28% jitter)

Branch router

Code

c2800nm-advsecurityk9-mz.124-24.T5.bin

From Debug

Jul 7 08:03:23: BGP: 10.128.7.30 open active, local address 10.10.242.130

Jul 7 08:03:30: BGP: 10.128.7.30 read request no-op

Jul 7 08:03:30: BGP: 10.128.7.30 went from Active to OpenSent

Jul 7 08:03:30: BGP: 10.128.7.30 sending OPEN, version 4, my as: 1, holdtime 180 seconds

Jul 7 08:03:30: BGP: 10.128.7.30 send message type 1, length (incl. header) 53

Jul 7 08:03:30: BGP: 10.128.7.30 rcv message type 1, length (excl. header) 26

Jul 7 08:03:30: BGP: 10.128.7.30 rcv OPEN, version 4, holdtime 180 seconds

Jul 7 08:03:30: BGP: 10.128.7.30 rcv OPEN w/ OPTION parameter len: 16

Jul 7 08:03:30: BGP: 10.128.7.30 rcvd OPEN w/ optional parameter type 2 (Capability) len 6

Jul 7 08:03:30: BGP: 10.128.7.30 OPEN has CAPABILITY code: 1, length 4

Jul 7 08:03:30: BGP: 10.128.7.30 OPEN has MP_EXT CAP for afi/safi: 1/1

Jul 7 08:03:30: BGP: 10.128.7.30 rcvd OPEN w/ optional parameter type 2 (Capability) len 2

Jul 7 08:03:30: BGP: 10.128.7.30 OPEN has CAPABILITY code: 128, length 0

Jul 7 08:03:30: BGP: 10.128.7.30 OPEN has ROUTE-REFRESH capability(old) for all address-families

Jul 7 08:03:30: BGP: 10.128.7.30 rcvd OPEN w/ optional parameter type 2 (Capability) len 2

Jul 7 08:03:30: BGP: 10.128.7.30 OPEN has CAPABILITY code: 2, length 0

Jul 7 08:03:30: BGP: 10.128.7.30 OPEN has ROUTE-REFRESH capability(new) for all address-families

BGP: 10.128.7.30 rcvd OPEN w/ remote AS 4

Jul 7 08:03:30: BGP: 10.128.7.30 went from OpenSent to OpenConfirm

Jul 7 08:03:30: BGP: 10.128.7.30 went from OpenConfirm to Established

Jul 7 08:03:30: %BGP-5-ADJCHANGE: neighbor 10.128.7.30 Up

0109_Aurora_CO_2811#

Jul 7 08:03:30: BGP: 10.128.7.30 read request no-op

0109_Aurora_CO_2811#

Than After 3 mins exactly the session will go to idle than stay stuck in Active....

Jul 7 08:06:32: BGP: 10.128.7.30 connection timed out 180884ms (last update) 180000ms (hold time)

Jul 7 08:06:32: BGP: 10.128.7.30 went from Established to Closing

Jul 7 08:06:32: %BGP-5-ADJCHANGE: neighbor 10.128.7.30 Down BGP Notification sent

Jul 7 08:06:32: %BGP-3-NOTIFICATION: sent to neighbor 10.128.7.30 4/0 (hold time expired) 0 bytes

0109_Aurora_CO_2811#

Jul 7 08:06:32: BGP: 10.128.7.30 send message type 3, length (incl. header) 21

Jul 7 08:06:32: BGP: 10.128.7.30 local error close after sending NOTIFICATION

Jul 7 08:06:32: BGPNSF state: 10.128.7.30 went from nsf_not_active to nsf_not_active

Jul 7 08:06:32: BGP: 10.128.7.30 went from Closing to Idle

Jul 7 08:06:32: BGP: 10.128.7.30 closing

0109_Aurora_CO_2811#

Jul 7 08:06:50: BGP: 10.128.7.30 went from Idle to Active

Jul 7 08:06:50: BGP: 10.128.7.30 open active, local address 10.10.242.130

0109_Aurora_CO_2811#

Jul 7 08:07:20: BGP: 10.128.7.30 read request no-op

Jul 7 08:07:20: BGP: 10.128.7.30 open failed: Connection timed out; remote host not responding, open active delayed 3642ms (5000ms max, 90% jitter)

0109_Aurora_CO_2811#

Jul 7 08:07:24: BGP: 10.128.7.30 open active, local address 10.10.242.130

0109_Aurora_CO_2811#

Jul 7 08:07:54: BGP: 10.128.7.30 read request no-op

Jul 7 08:07:54: BGP: 10.128.7.30 open failed: Connection timed out; remote host not responding, open active delayed 5336ms (10000ms max, 60% jitter)

0109_Aurora_CO_2811#

Jul 7 08:07:59: BGP: 10.128.7.30 open active, local address 10.10.242.130

Giuseppe Larosa · ‎07-07-2011

Hello Wrobbin,

this looks like a problem of recursive routing: once the BGP session up one router starts to think that it can reach the external IP address in another way.

3 minutes is exactly the default BGP hold time.

once the BGP session is up connectivity is broken and so after three minutes the side that didn't receive any BGP message for 180 seconds sends out a BGP notification message.

Be sure that the external networks are not advertised over the BGP session

Hope to help

Giuseppe