We have built a 3 node S2D cluster using SuperMicro 6029U-TRT SuperServer. Each node has 2 Intel P4600 NVMe drives for cache and 8 10TB SATA drives (Seagate 3.5" 10TB,7.2K RPM,SATA 6Gb/s,256MB,512E,Helium) for capacity.
Everything ran through cluster validation and built fine. When testing, the storage pool stays online if one node fails, but if you fail a second node the storage pool goes offline.
Doing get-StorageEnclosure shows 3 enclosures.
Using failover cluster manager, I can see the disks in each node.
This morning I rebuilt it trying to change the Cluster Fault Domain settings. Then when I re-enabled the cluster I did get this message:
racks' on Target 'HVCL3'.
I thought that since I set each node to be in a different chassis and different rack, that maybe the fault tolerance would have been fixed, but it still only withstood a single node failure.
The only other issue I have found close to mine is:
https://social.technet.microsoft.com/Forums/windowsserver/en-US/4fc1fb86-61fa-4976-8b3f-9e314586fef8/storage-spaces-direct-cluster-virtual-disk-goes-offline-when-rebooting-a-node?forum=winserverClustering
James - Right Size Solutions