On Fri, Dec 24, 2021, at 6:38 AM, cen wrote:
I recently had to replace a bad disk in raid1 array and finding
proper
docs was not a good experience.
Looking at my notes, I can add some of my notes. These may or may not apply to you...
(Should not be needed since you are not changing the UUID of the Raid)
Add any missing raid UUIDs to /etc/default/grub, there should be a
"rd.md.uuid=UUID" for each one of your RAID (not partition)
UUIDs. Rerun grub2-mkconfig to update your grub.cfg.
(Should not be needed since you are not changing the UUID of the Raid)
Remember to check /etc/mdadm.conf, just in case.
The above is mostly a reminder to review all raid related stuff after any such change.
If you happen to use a mix of ssd and rotating rust then remember this bit:
Enable it if you add an SSD for sda:
echo writemostly > /sys/block/md1/md/dev-sdb1/state
echo writemostly > /sys/block/md1/md/dev-sdc1/state
Make 100% sure that then new disk has at least as many sectors (I think it is the sectors)
as the old drive. If not then something needs to be shrunk down a bit to allow that
`sfdisk` copy to work properly. That is why it is best to leave some unused space when
first building these.
Not 100% sure that the "grow" is correct there since you are not changing the
size but it will let you know one way or the other. My notes have "mdadm --manage
/dev/md127 --add /dev/sdd1"