Upping the block size to more than whatever SMR overlaps on the drives works for workflows that don't care about consistent random rewrite performance. Can get

Upping the block size to more than whatever SMR overlaps on the drives works for workflows that don't care about consistent random rewrite performance. Can get 120 MB/s sequential writes on the cheap seagate SMR drives in my big slow pool which is enough for clients on gigabit or less networks. Although frankly at this point if you care about performance you're on NVME anyway.