Fail Safe Network (FSN) or Trunking (Etherchannel or LACP) are the High-availability options for a given data mover. So to configure a FSN for server_2 - you need to use the Network devices on server_2 only - you can not use a device from server_3.

To add, server_3 is typically configured as a standby data mover to server_2. In other words, the entire data mover is a standby to the active data mover. So, for active connection you can not mix the network interface of two data movers - rather you need to configure the network infrastructure and connectivity for both server_2 and server_3 exactly in the same way, so that in face of data mover failover or failback - everything works properly.

I 'll suggest - you please go through the following documents available on Powerlink -

1)  "Configuring and Managing EMC Celerra Networking" and

2)

Configuring and Managing EMC Celerra Network High Availability"

Thanks,
Sandip

Server_2 only

Thanks. That explains.

No – like with any other computer all parts of an interface needs to be on the same data mover