Once you have installed the main networking, you can go ahead and install the secondary master. Because this involves copying the entire contents of the primary master onto another machine, this step is a little involved to ensure that this is done safely.
First, install the secondary master into its slot, and connect the networking connections to both switch 3 and the primary master.
Next, configure the database from the Panel PC to include the MAC address of the secondary master in the 'secondary master' slot. This is usually in the documentation for the delivered device, but if you are unable to get this, the ipxe USB image will print all interfaces onto any attached displays. You will need to enter the MAC address of the second port - net0 or enp2s0, the one connected to Switch 3 - into this field. The MAC address should consist of a number with 12 positions and start with 4C38D5xxxxxx.
Now, boot the secondary master with a PXE boot USB attached to it - you can temporarily steal one of these from the compute node in the same cabinet if necessary. If done correctly, the boot sequence will appear to hang part-way through, having connected to the primary master for commands.
From this point, you will want to execute the 'Clone Master' command. This will actively allow this booted node access to the root filesystem of the current master, so that it can copy it to itself. This may take several minutes.
After this has completed, the new secondary master will automatically reboot, and will proceed to boot as normal from its internal storage. You can remove the PXE boot USB at this point. After booting, it will automatically connect to the primary master, and begin to clone the redundant filesystem.
Please be advised that this recloning process can take some time - an hour or two if the filesystem is particularly full. If the harvester is reset during this time, then this process will stop, and will resume only when power is restored. Full filesystem redundancy can only be assured once this cloning process completes.