• LACP failover did not happen to leaf2 when leaf1 lost BGP connection to spines.
• Due to core-isolation feature leaf1 LAG bundle brought down when it lost connectivity to the spines.

During the failover scenario complete loss of connectivity might be seen for  the host connected to the respective LAG bundle.

• External devices such as servers almost always require an LACP state that matches the ESI-enabled interface on the leaf switches (PE) to ensure the server can actively send traffic across all member interfaces in the LAG bundle. Specifically, the same ESI and LACP system identifier should be configured on all links that make up a LAG bundle. To ensure the suggested system ID consistency, it’s a best-practice to base the system ID on the ESI you assign to each LAG.
• For example, if you assign ESI 00:03:03:03:03:03:03:03:03:01 to a LAG, then it is suggested that you derive the corresponding system ID from the last 6 bytes of that ESI, i.e., “03:03:03:03:03:01”. The suggested best practice of deriving the system ID from the LAG’s ESI is demonstrated in the configurations presented later in this document.
• Configure LACP system identifier with below command if ESI is manually configured
• Interfaces <ae*> aggregated-ether-options lacp system-id <system-id>.

Initial version.

Minor modification in Title, Description and solution.