7500 Switch - Routing issue with NLB multicast servers

[LionLantech]

Hi all,
in our network we have this situation: we are using a cluster of 2 virtual NLB servers hosted each on a ESX server and configured in multicast mode. To limit broadcast traffic, we would like to configure static multicast MAC address entry along the path toward servers.
In our case, servers are connected to a switch HP 7506 (switch "B"), which is directly connected to the core switch (another HP 7506, switch "A") using bridge-aggregation 2 of switch B to bridge-aggregation 1 of switch A.

We have configured static multicast MAC, fixing the VIP multicast mac-address both on the interfaces of switch B where two servers are connected (3+3 interfaces) and on the bridge-aggregation 1 (uplink of switch A to B):
mac-address multicast "mac-address" interface "interface-list" vlan "vlan-id"

We have also disabled the ARP entry check function on the switches (undo arp check enable).
After this configuration, we can reach servers from clients on the same vlan of the servers but not from clients on other vlans (routing is on the switch A); however we can ping the cluster virtual IP from switch A also using other vlans as ping source (so not only form server vlan).
We also tried to configure static ARP entry on the switch A (described as option on the reference guide) but nothing changed.

Whats' wrong with this configuration? Or are there some known issues witch L3/routing using multicast?
Thanks in advance to all.