and this process creates a new non-redundant fully-degraded RAIDZ2 pool, and before restoring the redundancy with resilvers, the original pool is destroyed.
With my process both the old and new pools have single redundancy during the migration, and there is never any point at which there is zero redundancy.
I ordered a PCI SATA card with 4 slots and a PCIe power extender. I’m currently in the middle of doing a ZFS replication to the new pool. Fingers crossed that the replication goes smoothly, and then it’ll be time to point all my apps to the new pool.
Be sure that your PCI(e?) SATA card meets the recommendations for such cards made on both this forum and the old community forum when used with ZFS as many PCI SATA cards create problems.