RAID1 is DEGRADED, but which disk is faulty?

[jesse reynolds]

At 10:40 +0200 5/9/2003, Ruben de Groot wrote:
>On Fri, Sep 05, 2003 at 04:07:50PM +1000, jesse reynolds typed:
>>  At 18:26 +0200 3/9/2003, Ruben de Groot wrote:
>>  >On Wed, Sep 03, 2003 at 10:45:05PM +1000, jesse reynolds typed:
>>  >> Hi
>>  >>
>>  >> How do you tell which disk has problems in a hardware RAID1 array? My
>>  >> machine has been up for nearly six months, and only the last three
>>  >> weeks of messages logs are retained. There is nothing any in the last
>>  >> three weeks of system logs about any kind of hardware error, or
>>  >> warning, or anything mentioning ar0, ad4, or ad6.
>>  >>
>>  > > atacontrol is showing a DEGRADED status of the mirror:
>>  >>
>>  >> # atacontrol status ar0
>>  >> ar0: ATA RAID1 subdisks: ad4 ad6 status: DEGRADED
>>  >>
>>  >>
>>  >> How does one tell which disk is faulty? Or is it possible that
>>  >> they've become out of sync and need rebuilding, in which case which
>>  >> one do I detach and reattach?
>>  >>
>>  >> I'm scared to reboot as it's a running server. What can I do????
>>  >
>>  >You can try this:
>>  >
>>  >umount /dev/ar0
>>  >fsck -f -n /dev/ad4
>>  >fsck -f -n /dev/ad6
>>  >
>>  >This way (hopefully) you'll be able to establish which one of the disks
>>  >in the array has developed problems. Then reboot and rebuild the array
>>  >from the other disk using the controllers firmware.
>>
>>
>>  Given that all the filesystems, including /, are on this mirror ar0,
>>  I would have to boot it up off a live CD or some such in order to do
>>  an fsck on both the submirrors.
>
>Ah yes, in this case that's probably the safest way.
>
>>  Hmmm. I wouldn't be able to do an fsck on /dev/ad4 or /dev/ad6 though
>>  because they are not filesystems, they're the full disks which in
>>  turn contain freebsd slices.
>
>Replace ad4, ad6 with the names of your partitions, eg ad4s1a etc.

Right. ... do you think it is safe to reboot the server as it is? I'm 
wondering if dmesg might give more information about about which disk 
is still part of the mirror.

Why can't atacontrol tell you which subdisks are good and which have 
been offline'd?


>
>>  Is it safe to shut it down and see what the Promise firmware is
>>  saying? Or will i then lose the ability to tell which disk is good
>>  and which has problems?
>
>Not very probable. But since your filesystems are still usable, you could
>make a backup, just to be on the safe side.

Okay, thanks.

Cheers

Jesse

-- 

   ::: Jesse Reynolds +61 (0)414 669 790 ::: AIM - jessedreynolds :::
   ::: Virtual Artists Pty Ltd, Adelaide ::: http://www.va.com.au :::