After reading up a little more, I
think I am talking at cross purposes - or more likely I need to continue reading.....
For my setup, I have a Dell server with a PERC RAID card that utilises a bunch of 2.5" SAS SSDs and HDDs in several native volumes. ESXi then sees this hardware directly I use VMware datastores directly on these RAID volumes. At this point, TrueNAS has nothing to do with anything, so this is a relatively typical ESXi host deployment
In addition to the PERC RAID card, the server uses a more standard SATA controller to which are attached several 3.5" WD Red HDDs. I have placed the SATA controller into the passthrough mode and then run TrueNAS as a guest VM on ESXi and use this passthrough controller. The HDDs on this specific controller are then accessed directly by the TrueNAS guest VM. Currently, the vdev created within TrueNAS on this HDD is empty, but the intention is to backup the VM on the ESXI data site outside of TrueNAS to the vdev volume. I wasn't considering using TrueNAS as ESXi datastore unless, of course, that is essentially what happens if I am using the VMware Snapshots option (as per
https://www.truenas.com/docs/core/coretutorials/storage/vmware-snapshots/).
When I attempt to set up the VMware snapshot and connect the TrueNAS pool with and ESXi data store, I see the following warning, which then make me worried I am doing something wrong:
Then there is the issue of a SLOG cache. Currently, the only drives exposed directly to TrueNAS are SATA HDDs, and the are allocated to a single vdev. The True guest VM exists as in a VMDK on the SAS RAID array, and whilst I could crete more VMDKs for TrueNAS on this RAID array for a log vdev, it dosn't seem like that is a good idea.