I think it's worth mentioning that SCSI/SAS devices would show up as da devices (not to be confused to the old ata(4) driver's ad devices).

It's probably worth mentioning _why_ one may or may not want to create partition for ZFS.

One potentially useful scenario is that the administrator is not sure if they would be able to get replacement drives at a later time that is at least the same size of the current hard drive, this happens for some consumer grade hard drives (for example, different manufactures may have different opinion on how big 4TB is), so it used to be helpful to create a slightly smaller partition, let's say use a few gigabytes as spare space or even as a swap partition for each drive, to give the administrator a peace of mind that when they purchased a replacement drive they are guaranteed to have the partition properly sized (and when one replaces drive, it's likely because the current drive is failing or already failed, and getting something that can't be used would create more stress in such situations). This, however, may be seen as a waste of space by some other people, because one can always purchase a larger drive to replace the smaller, older drive.

Another potential use case is to make it easier to utilize GPT's label feature, for example one can name the partition in a more meaningful way, like '/dev/gpt/disk0' or '/dev/gpt/<SERIAL>-data', etc. when the administrator deliberately wanted to avoid using /dev/diskid/ devices.

That's said, there are indeed some legitimate reasons to create partition, but it's not really required. I personally create partitions on my own systems, usually because I want to do full disk encryption for the data storage and also have redundant boot devices at the same time (so I must use two or even three partitions per drive), but for single drive pools, creating partitions is probably not particularly helpful.

maybe zpool list / zpool status here instead of df?