Hello !
We recently experienced the following in our DSSV6/ESX environment:
Primary/active had a raid controller failure,
the controller didn't die completely,
but couldn't write to his volumes anymore.
So logically the iscsi connections were still up,
but all our VMs on our ESX hosts were frozen.
If a volume becomes physically unwritable on primary/active,
we suggested the system to failover to secondary.
Are we wrong ?
Regards
Ralph