If you use the 4 hosts for productive my advise is to wait some weeks.
I have some trouble with Resource Allocation in HA after 3.5.
With 3.02 there was no problem with 80 VMs an HA.
It was also a surprise for me, that 2 NICs are suggested for the management port. Otherwise you receive a warning (or error) with HA.
When I updated VC, I did had some errors / problem with HA. I had to disable and re-enable HA to fix it. Since then, all is fine. Are you sure your HA problem is related to 3.5?
We're using IBM HS21 blade. They only have 2 NICs. So I do have a warning message in VC saying: "Host XX currently has no management network redundancy" even if my hosts are still in 3.0.2. Since I cannot have more than 2 NICs, I won't be able to address this issue. I guest there must be a way to remove this warning.
Thanks for your help.
Is there a particular reason why you don't assign both NIC's to the SC?
Unless you somehow actually need the full bandwidth for VM's or VMotion, theres really nothing wrong with putting them both into one vSwitch.
Of course, if the two switches in the BladeCenter are in totally different VLAN's back to the Cores...then you obviously can't.
I'm using HS21's with two NICs and I'm getting the same warning. I have one vSwitch with the two pNICs attached and both are set to active.