Customer has two services:

1. L2vpn primary backup

2. EVPN-VPWS

They had tested the l2vpn and the EVPN-VPWS from remote PE over MPLS network

But they facing issue only l2vpn traffic, during fall-back & primary bring-up scenario.

When disable the interface from primary R1 toward the switch, the l2vpn service moved to R2

But when they enable the interface on R1, the l2vpn revert to the first R1, but the ping doesn't work between the two CE's.

When they disable and enable the interface in R2, the ping return to work through the l2vpn on R1.

The ESI lag is not qualified for l2vpn connections & hence engineering recommended not use it.

The hot-standby knob resolve the issue in setup, but its not supported or qualified knob for l2vpn either.

Below is the update from engineering on internal PR1853869-

ESI lag is not supported for L2VPN. Instead of ESI lag, please separate IFDs from CE.

And also hot-standby is not supported for L2VPN, only active-standby is supported.

Further analysis on current situation-

With hot-standby, it looks working fine like all-active even if CE1 does load balancing and forwarding path on both PEs (PE1 and PE2) are programmed correctly and working fine.

Without hot-standby, when primary link comes UP back on PE1, CE1 still sending traffic to PE2 (old primary), traffic getting dropped on PE2 as the l2vpn connection is down. In this case, need to configure separate ae bundle on CE connecting to each PE and add both interfaces as part of vlan.