RAID-6 – Replace Two Dead Drives at the Same Time or One at a Time?

3ware provides 3dm2 monitoring / management program. just look onm their webpage, there is a binary version of it for linux, and it even works fine [under debian at least].

thing is - i never trusted those fancy tools... so i do as follows. so i use own scripts based on tw_cli - also downloadable from 3ware website.

once per week i run patrol read:

./tw_cli /c0/u0 start verify

and all the time, every 15 minutes i dump current state of raid:

./tw_cli /c0 show > current.txt

i use very simple nagios plugin to check if current status is identical to expected [ i just compare content of a file to well known status dump that was taken at the beginning ].

#!/bin/bash

if [ `diff current.txt expected.txt|wc -l` -ne 0 ] ; then
        echo "CRITICAL - current state of raid does not match expected pattern "
        exit 2
fi

if [ `find . -name current.txt -mmin -16|wc -l` -ne 1 ] ; then
        echo "CRITICAL - state file is old "
        exit 2
fi
echo "OK"
exit 0

you would probably put mailing instead of exit 2... or even more probably use 3dm2.

You de-split it into separate disks, then convert into degraded RAID1 and then add the second disk. That's how we migrated. Stupid, but it works.