Troubleshooting:EBGP Neighbor Flapping

Nahid
Nahid  Diamond  (1)
7 years 10 months ago  View: 1506  Reply: 6
1F
There are three possible causes for neighbor flapping:
1.Improper processing during the forwarding of the route whose AS-Path attribute has more than 255 AS numbers to the EBGP neighbor
2.Sending of a wrong packet to Huawei device by a device of another vendor due to the failure to process the route with more than 255 AS numbers
3.Failure to parse the packet with more than 255 AS numbers by a device of another vendor although Huawei device can properly process the route


The corresponding checking methods are as follows:
1.Run the display bgp peer peer-ip-address log-info command to check whether neighbor flapping is caused by 3/11 (3 indicates an Update packet error; 11 indicates an AS-Path attribute error in the Update packet).


2.Run the display bgp routing-table command to check whether the preferential routes with no less than 255 AS numbers are available for the neighbor. These routes will be forwarded to the neighbor.


Case1: The logs on a device of another vendor that connects to Huawei device show that an error packet sent by Huawei device is received.
The AS-Path attribute with more than 255 AS numbers is not processe22d. Therefore, the AS number count field (one byte) is 0, and an error packet is generated. In this case, Huawei device is faulty.


Case2: The EBGP neighbor between Huawei device and a device of another vendor flaps and then automatically recovers to the normal state. The logs on the device of another vendor show that the device receives the Update packet from Huawei device.The packet carries more than 255 AS numbers. The cause of neighbor disconnection is also that the AS-Path attribute of the route advertised by Huawei device to the device of another vendor carries more than 255 AS numbers. According to the protocol, each AS-Seq can store a maximum of 255 AS numbers. When the number of AS numbers exceeds 255, another AS-Seq should be used to store the additional AS numbers. However, the peer device of another vendor deems it as a fault (which is allowed by the protocol) and shuts down the neighbor. Thus, 5/0 BGP error occurs on Huawei device, which results in neighbor interruption. In this case, Huawei device is normal.

akther
akther  Gold 
7 years 10 months ago
2F
good job done!
jfjina
jfjina  Gold 
7 years 10 months ago
3F
nice troubleshooting. hope to see more post like this
nasir007
nasir007  Gold 
7 years 10 months ago
4F
good job. thanks
nazmun021
nazmun021  Gold 
7 years 10 months ago
5F
nice doc
ashrafulhakim
ashrafulhakim  Silver 
7 years 10 months ago
6F
good
user_2837311
user_2837311  Diamond 
3 years 9 months ago
7F
useful document, thanks