Why are BGP connections re-established after the peer connect-interface command is configured

1

Currently, the switch does not support BGP dynamic capability negotiation. Therefore, if certain capabilities of the BGP peer are changed, the BGP connection is automatically disconnected and then the capabilities of the neighbor are renegotiated.

If the peer connect-interface command is configured, the BGP session needs to be set up by designating the egress. Therefore, the source address of the TCP connection may be changed and the TCP connection needs to be re-established by using the new source address.

Other related questions:
Why can BGP not immediately attempt to establish connections after the peer is configured
To prevent the Border Gateway Protocol (BGP) from frequently tearing down the neighbor relationship for renegotiation, the device waits for a proper period before establishing connections to ensure that related configurations are complete. The recommended period in RFC4271 is 120 seconds. The implementation in AR series routers is 32 seconds.

Why are BGP connections reestablished after the peer connect-interface command is executed on S series switches
Currently, S series switches do not support BGP dynamic capability negotiation. Therefore, when capabilities of the BGP neighbor are changed, the BGP connection is automatically cut off and then the capabilities of the neighbor are renegotiated. If the peer connect-interface command is configured, the BGP session needs to be set up by designating the egress. Therefore, the source address of the TCP connection may be changed and the TCP connection needs to be re-established by using the new source address.

Why does BGP attempt to establish connections over 30 seconds after the peer is configured
Compared with IGP configuration, BGP configuration is more complex. In addition to the peer and AS, the egress, multi-hop, timer, and various capabilities need to be specified. Currently, BGP does not support dynamic capability negotiation. Therefore, these parameters, after being modified, need to be renegotiated. To avoid frequent interruptions during the renegotiation, a proper time parameter is required to ensure that the relevant configurations are complete before the link establishment attempt. RFC4271 recommends 120s, whereas 32s is adopted by switch.

Why is the BGP connection not interrupted immediately after the interfaces connecting two peers are shut down
When External Border Gateway Protocol (EBGP) peers are directly connected and the ebgp-interface-sensitive command is run in the Border Gateway Protocol (BGP) view, the BGP peer relationship is interrupted immediately after the interfaces connecting the two peers are shut down. By default, the ebgp-interface-sensitive command is run in the BGP view. In other cases, the BGP peer relationship is not interrupted until the Hold timer times out.

Why does an S series switch try to establish a BGP connection over 30s after peer configuration
For S series switches that support BGP, it is more complex to configure BGP than IGP. Besides peers and ASs, outbound interfaces, multiple hops, timers, and various capabilities need to be configured. At present, BGP does not support dynamic capability negotiation. Therefore, the changes of these parameters need to be negotiated again. To prevent the BGP peer relationship from being interrupted frequently, a proper timer is required. Ensure that related configurations are complete before the attempt of setting up a BGP connection. In RFC 4271, you are advised to set the value of the timer to 120s. On Huawei devices, the value of the timer is set to 32s.

If you have more questions, you can seek help from following ways:
To iKnow To Live Chat
Scroll to top