diskspotting (how to tell the dead one?)

diskspotting (how to tell the dead one?)

[Dexter Filmore]

Assuming one has 6 disks on two controllers and one dies, how does one tell 
the dead one? Which controller/port?

Dex

-- 
-----BEGIN GEEK CODE BLOCK-----
Version: 3.12
GCS d--(+)@ s-:+ a C++++ UL++ P+>++ L+++>++++ E-- W++ N o? K-
w--(---) !O M+ V- PS+ PE Y++ PGP t++(---)@ 5 X+(++) R+(++) tv--(+)@ 
b++(+++) DI+++ D- G++ e* h>++ r* y?
------END GEEK CODE BLOCK------

http://www.vorratsdatenspeicherung.de

Re: diskspotting (how to tell the dead one?)

[Richard Scobie]

Dexter Filmore wrote:
> Assuming one has 6 disks on two controllers and one dies, how does one tell 
> the dead one? Which controller/port?

There are probably more elegant ways, but I use smartctl -a on the 
failed drive to find it's serial number and use this to find the 
location based on where I installed the drives initially, which I note 
in a text file or a label attached to the exterior of the case on 
machines with removable drive trays.

Regards,

Richard

Re: diskspotting (how to tell the dead one?)

[Greg Freemyer]

On Thu, Oct 30, 2008 at 2:36 AM, Richard Scobie <richard@sauce.co.nz> wrote:
> Dexter Filmore wrote:
>>
>> Assuming one has 6 disks on two controllers and one dies, how does one
>> tell the dead one? Which controller/port?
>
> There are probably more elegant ways, but I use smartctl -a on the failed
> drive to find it's serial number and use this to find the location based on
> where I installed the drives initially, which I note in a text file or a
> label attached to the exterior of the case on machines with removable drive
> trays.
>

Agreed, the best way is to know in advance what is what.

At this point the safest thing to do is likely to create you
documentation for the 5 working drives and assume the other one is the
dead one.

Unfortunately the only truly safe way to do that I know involves
stopping the raid service.  (You can try with it running but data
destroying kernel bugs do occasionally show up when you read low-level
data at the same time as you access the raid via live filesystem
activity.)

So I would:

1) Stop the raid (via mdadm I assume)

2) Try to read the serial number directly from your first drive.

ie: hdparm -I /dev/sdb

repeat for all 6.  One should fail I assume.  Make note of it.

3) If this is sata, pull the sata power cable for one of the drives.
Repeat step 2, but when a new drive shows up failed, note which one it
is physically.

4) Repeat until done.

5) Power off your system, hook everything back up exactly like it was
and reboot.  Make sure everything is like it was when you started.

6) And you now know which drive is failed.  Sorry, but it is the only
way I know.

FYI: Disk racks that are made to hold raid arrays typically have an
LED for each position.  Then there is "find", "locate", or "blink"
command that will cause that LED to blink.  It is used for exactly
this purpose.

Greg
-- 
Greg Freemyer
Litigation Triage Solutions Specialist
http://www.linkedin.com/in/gregfreemyer
First 99 Days Litigation White Paper -
http://www.norcrossgroup.com/forms/whitepapers/99%20Days%20whitepaper.pdf

The Norcross Group
The Intersection of Evidence & Technology
http://www.norcrossgroup.com

Re: diskspotting (how to tell the dead one?)

[David Greaves]

Richard Scobie wrote:
> Dexter Filmore wrote:
>> Assuming one has 6 disks on two controllers and one dies, how does one
>> tell the dead one? Which controller/port?
> 
> There are probably more elegant ways, but I use smartctl -a on the
> failed drive to find it's serial number and use this to find the
> location based on where I installed the drives initially, which I note
> in a text file or a label attached to the exterior of the case on
> machines with removable drive trays.

Same here.

I write the last few digits of the serial number on a label and attach it to the
drive where it is visible when you open the case.

David


-- 
"Don't worry, you'll be fine; I saw it work in a cartoon once..."

RE: diskspotting (how to tell the dead one?)

[Chris Green]

cp /dev/md0 /dev/null

And then look at which drive's light isn't on.


-----Original Message-----
From: linux-raid-owner@vger.kernel.org [mailto:linux-raid-owner@vger.kernel.org] On Behalf Of Richard Scobie
Sent: Wednesday, October 29, 2008 11:36 PM
To: Dexter Filmore
Cc: Linux RAID Mailing List
Subject: Re: diskspotting (how to tell the dead one?)

Dexter Filmore wrote:
> Assuming one has 6 disks on two controllers and one dies, how does one tell
> the dead one? Which controller/port?

There are probably more elegant ways, but I use smartctl -a on the
failed drive to find it's serial number and use this to find the
location based on where I installed the drives initially, which I note
in a text file or a label attached to the exterior of the case on
machines with removable drive trays.

Regards,

Richard
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: diskspotting (how to tell the dead one?)

[Dexter Filmore]

Am Donnerstag, 30. Oktober 2008 17:47:31 schrieb Chris Green:
> cp /dev/md0 /dev/null
>
> And then look at which drive's light isn't on.
>
Thing is they don't have individual lights, they're just sitting in the tower, 
no backplane or such.


-- 
-----BEGIN GEEK CODE BLOCK-----
Version: 3.12
GCS d--(+)@ s-:+ a C++++ UL++ P+>++ L+++>++++ E-- W++ N o? K-
w--(---) !O M+ V- PS+ PE Y++ PGP t++(---)@ 5 X+(++) R+(++) tv--(+)@ 
b++(+++) DI+++ D- G++ e* h>++ r* y?
------END GEEK CODE BLOCK------

http://www.vorratsdatenspeicherung.de

Re: diskspotting (how to tell the dead one?)

[Billy Crook]

On Thu, Oct 30, 2008 at 12:18, Dexter Filmore <Dexter.Filmore@gmx.de> wrote:
> Am Donnerstag, 30. Oktober 2008 17:47:31 schrieb Chris Green:
>> cp /dev/md0 /dev/null
>>
>> And then look at which drive's light isn't on.
>>
> Thing is they don't have individual lights, they're just sitting in the tower,
> no backplane or such.

tar md0's contents, pipe into /dev/null, and use a stethoscope to find
the one drive that's not clicking wildly?

Seriously.  Getting the serial number with smartctl or hdparm, and
searching through the disks for that SN while the machine is offline
is probably the safest route if you aren't absolutely certain which
drives are where.  With anything less certain than the serial number,
there's noteable chance you could pick wrong.  Depending on what raid
level is used, and how failed it is, you might not have the liberty of
a second try if you pull the wrong drive with the array in use.