network port failover with IPMP on cluster environment

Hello, community user.

If each zfs cluster head have two 10g network card(4 port/2port each HBA) and IPMP group was configured as below.

All network cable was estabilished to proper ethernet switch.(total 8 lines)

1. ZFS01 (IPMP Active-StandBy configuration)

    IPMP-A(ixgbe0/ixgbe2) - 192.168.10.101

     ixgbe0 -> Active / ixgbe2 -> StandBy

2. ZFS02 (IPMP Active-StandBy configuration)

    IPMP-B(ixgbe1/ixgbe3) - 192.168.10.102

    ixgbe1 -> Active / ixgbe3 -> StandBy

The Question is, If ixgbe0 interface failed(cable problem), service network ip trans to ixgbe2 on same ZFS head(ZFS01)

and then, If ZFS01 system downed(not manually, abnormally occured), which network interface failover to ZFS02? (ixgbe0 or ixgbe2)