[Gluster-users] Network redundance

[Alessandro Briosi]

Hi all,
a feature that would really be nice is network redundance.

In a current installation I have 3 servers with 2 switches (staked with
10gb uplink). The servers contain VM.
 
I have configured the hosts to comunicate through the first switch, but
what happens is the switch goes down? Probably the VM would shutdown as
the quorum is not available at any host.

Is there a way to make it to switch automatically on a the backup link
on the second switch ? (The hosts right now have bonding interfaces with
LACP on the switches, bond0 -> switch1, bond1 -> switch2)
I tried using balance-alb across the 2 switches but as the hosts are 3
there's no balance at all...

A bond of a bond is not possible AFAIK, but I'd like to have something
similar.

It would be nice to have something like iSCSI multipath or at least an
active/backup network for the storage.

Anybody has a suggestion on how to solve this right now?

Thanks,
Alessandro