I setup ‘lxd init’ to use an existing ZFS dataset, zfspool/lxd. After running init I checked ‘zfs list’ and saw that all the lxd datasets were using ‘legacy’ as the mountpoint. Did a bit of searching and found that ‘legacy’ mountpoint enables legacy tools to manage those file systems rather than ZFS.
Just want to make sure that this is all setup properly. I had expected to see ‘none’ for the lxd mountpoints rather than ‘legacy.’
Output of ‘zfs list’ is below:
NAME USED AVAIL REFER MOUNTPOINT
zfspool 562K 1.11T 24K none
zfspool/lxd 240K 1.11T 24K legacy
zfspool/lxd/containers 24K 1.11T 24K legacy
zfspool/lxd/custom 24K 1.11T 24K legacy
zfspool/lxd/deleted 120K 1.11T 24K legacy
zfspool/lxd/deleted/containers 24K 1.11T 24K legacy
zfspool/lxd/deleted/custom 24K 1.11T 24K legacy
zfspool/lxd/deleted/images 24K 1.11T 24K legacy
zfspool/lxd/deleted/virtual-machines 24K 1.11T 24K legacy
zfspool/lxd/images 24K 1.11T 24K legacy
zfspool/lxd/virtual-machines 24K 1.11T 24K legacy
zfspool/zfs 24K 1.11T 24K /zfs
Is this proper? Does ‘lxd init’ always setup lxd datasets to use ‘legacy’ mountpoints?
Edited to add: In case this makes any difference, when I setup the zfspool/lxd dataset I did so via “sudo zfs create zfspool/lxd”, then in ‘lxd init’ I pointed lxd to zfspool/lxd.