Possible*, but tricky and potentially hazardous to your data. It would also require a bit of work at the CLI; you wouldn't be able to create the pool using the GUI. Vdev expansion is coming to ZFS, but you won't see it in FreeNAS for at least a year (and I'd plan on two).
* Edit: Well, I should probably explain, at least generally, how you'd do this. There are two basic methods you could use. The first is to create a degraded pool. In general, it would look like this:
- Create a 6 TB sparse file on your boot device (
dd if=/dev/zero of=/root/sparsefile bs=1m count=6m)
- Break your RAID1 mirror and put one of the 6 TB disks into your FreeNAS box along with the two 8 TB disks
- Manually partition each of the three disks in your box with a 2 GB swap partition as the first partition, and the remainder as freebsd-zfs
- Use
glabel status to get the gptids of the second partiton of each disk
zpool create tank raidz1 /root/sparsefile gptid/blah1 gptid/blah2 gptid/blah3 zpool offline tank /root/sparsefile. Note that this will remove any redundancy from your pool--you'll effectively have a three-disk RAID0. It will, however, have the same capacity as your final pool.
- Use the GUI to import your data (the "Import Disk" feature)
- Once the data's imported, use the "Replace Disk" feature in the GUI to replace /root/sparsefile in the pool with the remaining 6 TB disk. Once you start this, the copy of your data on what used to be the NTFS disk will be destroyed, and until the resilvering operation completes, you'll have no redundancy. A disk failure during that window of time will result in the loss of all your data.
The other option is to create a pool using partitions (hat tip to
@Stux for this method):
- Using the CLI, partition each of the 8 TB disks into two partitions of 4 TB each.
zpool create tank raidz1 da1p1 da1p2 da2p1 da2p1 (substituting the appropriate disk designations, of course).- Connect one of the 6 TB disks and use the Import Disk feature to copy the data onto your pool.
- Connect the other 6 TB disk and use the Replace Disk feature, repeatedly, to replace partitions with disks. Replace the first partition on the first 8 TB disk with the first 6 TB disk, the second partition on that 8 TB disks with the second disk, the first partition on the second 8 TB disk with the first 8 TB disk, and the second partition on the second 8 TB disk with the second 8 TB disk. This last step will require taking that partition offline first, which (as in the previous case) means you won't have any redundancy until the process is complete.
It's worth pointing out that we don't generally recommend RAIDZ1 when using disks larger than about 1 TB. A much simpler process, which doesn't require any CLI-fu and gives you more redundancy (though correspondingly less capacity), would be this:
- Using the GUI, create a mirrored pool with the two 8 TB disks
- Use the Import Disk feature in the GUI to import the data from the 6 TB disk
- Using the Volume Manager, extend your pool using the pair of 6 TB disks as a second mirror
- This gives you a pool of striped mirrors with 14 TB of capacity.
