Equallogic network alchemy ??

I haven't used EQL PS arrays in awhile but recently I encountered one. In case you aren't aware these arrays have multiple network interfaces (for ISCSI) but all of the NICs have to be in the same subnet. So right now all of the hosts and all of the array NICs are connected to the same switch. Which works fine and dandy even multipathd works fine.

I wanted to move one of the controller NICs (and its failover on the secondary controller) as well as the secondary ISCSI NICs on the hosts over to another switch for minimal redundancy...

So I connected the ports up and if I ping the secondary NIC on the PS controller it fails. I am assuming that either the Linuc ISCSI initiator is trying to send the ping requests out via the other NIC (same subnet, required remember) or the array is sending the response back on the wrong NIC.

I have these commands on my linux host to try and make Linux not care which NIC it sees the replies on. Which is probably why multipathd works with the single switch setup. I just cannot for the life of me remember how you make it work across two switches that are not interconnected. Also since the PS group uses a load balancing front-end IP I don't think routing would help.

sysctl commands

net.ipv4.conf.p2p1.arp_ignore=1 net.ipv4.conf.p2p1.arp_announce=2 net.ipv4.conf.p2p1.rp_filter=2 net.ipv4.conf.p2p2.arp_ignore=1 net.ipv4.conf.p2p2.arp_announce=2 net.ipv4.conf.p2p2.rp_filter=2

I used to know how to fix this but its been... a while. If anyone remembers please let me know. :)