High CPU seen on backup RE with NSR configured

Numerous EBGP peer groups can lead to memory exhaustion on back-up Routing Engine (RE) which also causes performance issue on the main RE.

Too many protocols: bgp:groups can lead to RPD on backup RE restarts at RLIMIT with BGP NSR

Numerous EBGP peer groups can lead to memory exhaustion on back-up Routing Engine (RE) which also causes performance issue on the main RE. This problem is more likely to happen on 32-bit Junos software since RPD size limit is 2Gbytes. This issue is indicated by routing protocol daemon restarts with RLIMIT.

While it is convenience to use protocols:bgp:groups to manage BGP peering; considerations has to be made to ensure that bgp:groups is not overused – especially in case of External BGP (EBGP) peering. The problem is multiplied when using non-stop-routing.

Each BGP group has its own copy of AS Path information of the entire BGP routing table.
It is best to optimize memory usage by minimizing the number of BGP groups in your configuration.

About the author

Prasanna

Leave a Comment