That is what it defaults to when I rescan the HBA's and it bounces as well.

The latency is probably due to the bouncing.

Change the multipathing to "Most Recently Used" and see if that helps.

Could it be a LUN number issue? Currently both virtual disks are lun 0 and when I change the mapping to the cluster group it requires that I change the lun for the virtual disk.

Each virtual disk needs a unique Host LUN number (on a host/hostgroup). I'd start with 0 and work up from there (no gaps).

Think I may have figured out my issue but won't know until this weekend whether or not it is actually the problem. I think what I was doing was giving preferred paths to the virtual disk over separate controllers. So R710-1 had a preferred path to virtual disk 1 over controller 0 and R710-2 had a preferred path to virtual disk 1 over controller 1, and since only one controller can access a virtual disk at the same time, this may explain why it was flopping. Each server was accessing the virtual disk simultaneously and therefore the controller ownership had to switch back and forth. Sound about right?

Ok, I was able to play with it this morning and got it going. The issue was that I was preferring paths to the same virtual disk over 2 different controllers. Once I drew it all out and took my time I was able to get the multipath settings correct in vm.

The only remaining issue I have is this...

Since both R710's now access the same virtual disk over the same controller, what happens when I unplug one of the HBA's from one of the R710's? What I assume will happen is the R710 that I unplugged the HBA from will use the secondary HBA to access the virttual disk, and that path is on the other controller. So that HBA will be accessing the virtual disk over one controller and the other R710 will be accessing the virtual disk over the other controller. Won't this once again cause the controllers to start flapping?

Hello. I am interested in knowing if you were able to test this controller flapping issue any further. I want to use the MD3220 DAS in a VMWare HA configuration, but am worried about your last scenario. Dell/VMWare are not coming right out and saying it is not supported, but are also not saying it will not work. Thanks, Jason

Oh, I just noticed I hadn't updated this with my most recent experience, I had been running Fixed which was working fine, until I vmotion'd a few vm's over to another host. As long as I moved all the vm's from one host to another it was ok. But it seemed as if I only moved a few or only moved a few back, the controller thrashing started and everything grinded to a hault. This is when I discovered I shouldn't be using the fixed policy at all, and plan to change to MRU.

Jason, I had been hoping to get a maintenance window in order to change my multipath configurations to MRU, but have been unable get the time. I hope by next weekend I can try it out. According to the vmware hardware compatability list, MRU is the preferred multipath method, which I have not tried yet. I was using fixed. I sure hope that does the trick, otherwise the SAN is useless and won't do any good for vmotion, HA, fault tolerance etc. I will update this and let you know how it goes.

MRU or ROUND ROBIN, but not fixed.