Rubenerd

Michael asked in an online poll whether we create zpools:

• Artisanally by hand
• With scripts
• With a GUI
• After cursing and wasting another morning on ext4

For personal and family stuff, I use the (definitely not nasty) script below. I run ZFS on full drives which I’ve first configured with gpart(8) with a freebsd-zfs partition and a label (though I don’t think that’s strictly necessary anymore?). Normalisation is a big deal for us.

#!/bin/sh
	
zpool create -o ashift=12 \
	-O atime=off \
	-O exec=off \
	-O compression=lz4 \
	-O utf8only=on \
	-O normalization=formD \
	-O casesensitivity=insensitive \
	mirror "/dev/gpt/$1" "/dev/gpt/$2"

For work pools, I use orchestration scripts that the client can grok, such as Saltstack. Same goes for zfs datasets and, unfortunately, ext4.