[AuckLUG] Re: Thanks, weather!

[Justin Cook]

To answer my own question, I fixed it. I'm so proud of myself.

The partition table for the drive that fell out of the array was 
damaged. I used fdisk to remove the partitions and re-added them. Then I 
figured out how to readd the device to the array with mdadm.

mdadm's documentation is nasty! So for the other newbs, here's how you 
remove and add a device to the array:

# mdadm /dev/md1 -r /dev/hdb1
mdadm: hot removed /dev/hdb1
# mdadm /dev/md1 -a /dev/hdb2
mdadm: re-added /dev/hdb2

>From there # cat /proc/mdstat will tell you how it's doing re-mirroring the array.


Now that that's sorted, does anyone know how to check a drive's integrity?



Justin Cook wrote:
> We just had a brownout and as a result it appears one of the drives in 
> my mirror array has died. Sort of. Here's what I mean:
>
> [root at neko ~]# cat /proc/mdstat
> Personalities : [raid1]
> md1 : active raid1 hda1[0]
>      79007552 blocks [2/1] [U_]
>
> md0 : active raid1 hdb2[1] hda2[0]
>      1020032 blocks [2/2] [UU]
>
>
> Damn. How does one work with mdadm? Is the best strategy just to bang 
> a replaccement drive in there? Why would one partition fail, but not 
> the second one? Should I remove the semi-functional drive immediately?
>
> Anyway, the mirror did its job. Hooray!
>