LACP Active/Active with VC connected directly to NetApp filer

[stueng]

Hi, twice now (for two seperate customers) I have experienced the same issue

We have two, stacked, VC modules in the back of a C7000 enclosure

We have two, failover, NetApp filers in a cluster

We connect, using a crossed configuration - 4 X Fibre Ethernet cables between the VC modules and the filers

Expected outcome:

Active/Active/Active/Active - in two seperate LACP trunks; one LACP trunk per filer

Actual outcome:

Active/passive/passive/passive

This means that we can only communicate with one filer at a time with this configuration

In both instances I have had to work around this issue by creating two seperate networks, one for each filer in seperate broadcast domains. This eliminates any possibility for a redundant configuration

I suspect that if I was to install a 10GbE switch between the VC and the NetApp filers that the negotiation would work properly and I could have a fully active and redundant configuration

My question is, is it a supported configuration to have a VC module connected directly into a NetApp filer and expect two full active LACP trunks within the same broadbast domain? If so, how is this acheived?

I have attached two very simple diagrams to show what I want and what I get. Only a single switch is drawn because the switch is supposed to be stacked

[Psychonaut]

As far as I've heard that is not a support configuration, as the VC is technically a Layer-2 device and not a switch. To get LACP you need a swtich to manage/create that connection; keep in mind you wouldn't need an external switch, it could be another interconnect.  

The only supported direct attached FC storage is 3Par.