I have a Server - SME9 / 4 Disks / Raid5 3 Active - 1 Spare
This is the older config, to get this build SME as a Raid5 on 3 disks then add the 4th
If you have been around servers for a while you can usually feel if something is not quite right, SMART produces an error or dmesg gives a hint - you wind up feeling one hard drive is going to fail - but it hasn't .. yet.
And it's active - not the spare
To move it out I use:
- mdadm --manage /dev/md1 --fail /dev/sdc2
- wait for the raid to build onto the spare
- mdadm --manage /dev/md1 --remove /dev/sdc2
- shutdown, replace hard disk
- partition etc. etc
- mdadm --manage /dev/md1 --add /dev/sdc2
- etc. etc
during step 2, your array is degraded and - if one of the good disks fails it's goodnight - and that happened - and I lost the machine
of course you can do due diligence - i.e.
smartctl -t long [all your disks], done that
so the questionif your config is:
md1 : active raid5 sdc2[3](S) sdd2[4] sdb2[5] sda2[0]is there a command to change the config to md1 : active raid5 sdc2[3] sdd2[4](S) sdb2[5] sda2[0]
which doesn't degrade the array during the process ?