#7 Linux LVM, RAID, failed drive recovery 2021

In the previous SSD cached LVM/RAID notes some time ago (in a galaxy far, far away) we setup a moder, mdadm-less Linux software RAID setup.

However, in the meantime maybe a drive failed, and as I found the error recovery a bit tricky, let's document exactly what commands are needed to recover the LVM RAID from an error:

RAID5 failed / missing drive

pvcreate /dev/sd*new
vgextend vg0 /dev/sd*new
lvconvert --repair vg0/lv

As this will still leave the old UUID missing, you might want to clean that up with:

vgreduce --removemissing vg0

Easier replaceing an online driver

If you notice an aging and degrading, or already partially failed drive, you should be able to replace that a bit easier when still online:

pvcreate /dev/sd*new
vgextend vg0 /dev/sd*new
lvconvert --replace /dev/sd*old vg0/lv /dev/sd*new
vgreduce vg0 /dev/sd*old

Added bonus: periodically srub the whole thing to prevent bit rot!!1!

As magnetically or otherwise stored data bits might decay and thus "bit rot" after some time, some consider it a good idea to periodically scrub and refresh it. This also can help to catch drives going bad earlier:

lvchange --syncaction check vg0/lv

External links

The Author