On Mon, Apr 15, 2019 at 03:45:56PM +0200, Andreas Klauer wrote: > # mdadm --grow /dev/md42 --level=5 > mdadm: level of /dev/md42 changed to raid5 > # mdadm --manage /dev/md42 --add-spare /dev/loop2 --replace /dev/loop0 --with /dev/loop2 needs another --wait between these two steps I'm not sure if the data relocation steps 0->5->0 can be skipped. (by growing to a layout with parity on missing disk 3?) Oh, maybe that's why they use raid4 instead of raid5 on SE. You figure it out... luck Regards Andreas Klauer
