Fedora's RAID-1: how to recover from a crash?
 

Hello everyone,

I have installed Fedora Core 1 and Core 3 on my servers and, in all installations, I'm using software RAID-1. I usually configure four partitions:

/dev/md0 ==> /boot
/dev/md1 ==> /
/dev/md2 ==> /var
/dev/md3 ==> swap

Well, after a long time feeling secure, one of the hard disks of one of the servers crashed. So I shut the server down, removed the crashed HD, turned the server on with the remaining HD, and, to my dismay, it didn't work. The remaining HD was corrupted.

Did I do something wrong? What should I do in the case of a crash? By the way, the HDs are not hot-pluggable.

Thanks,

Juliano Pastorino

Partition the new drive to match the old drive, then start each array (for example):

/sbin/mdadm /dev/md0 --fail /dev/hda1 --remove /dev/hda1
/sbin/mdadm /dev/md0 --add /dev/hda1

You can check the status of the array with:

cat /proc/mdstat

or

/sbin/mdadm -D /dev/md0

Hi macemoneta,

Thanks a lot for your help. Just to be sure, I would like to know if I should follow the steps below:

1) When I notice that a HD has failed, I first execute:

/sbin/mdadm /dev/md0 --fail /dev/hda1 --remove /dev/hda1

for each partition.

2) Then I turn off the server, remove the old HD, install de new HD, and turn on the server.

3) Then I partition the new HD, and execute:

/sbin/mdadm /dev/md0 --add /dev/hda1

for each partition.

Is that correct?

Thanks,

Pastorino

That'll do it. Make sure you have the mdadm command installed on your distribution. If not, the old (pre-mdadm) way was with:

/sbin/raidhotremove /dev/md0 /dev/hda1
...
/sbin/raidhotadd /dev/md0 /dev/hda1

Either way will work, but using the mdadm command is preferred.