[SATLUG] 1st question - RAID5 quit - Short tests completed

David Labens adlabens at swbell.net
Mon Aug 17 15:52:41 CDT 2009

Don,  I'll reseat all the connectors on sdc this evening.  THANK you for the detailed explanation.  And, I'll post the results of the "mdadm --examine /dev/sda1 /dev/sdb1 /dev/sde1" test this evening, also.
The data is important, but the most of it is pictures, and I've got a lot of them copied, just nothing from the past year.  The rest of it is recipes, old data from a real estate venture that's been completed & probably won't be needed again, and music that can be replaced.  I don't want to lose the pictures, but we still have the kids ;+) and are still taking pictures, so the pictures - tho major compared to the other data - are really minor.  I'd have to discuss with April the costs of a recovery service, but I've heard of SpinRite, and heard good things, so we'll probably go that route.
Someone mentioned having extra drives, and the warrantee.  I'll order a new replacement drive and pay for it, then will pursue replacement - under warrantee - of the defective drive after this is resolved.  That way, we'll ultimately end up with an extra drive.
And, I will start a regular backup routine, AND, I will install the APC UPS that we got for this thing - including the auto-shut-down connections & software.

David Labens
San Antonio, TX

--- On Mon, 8/17/09, Don Wright <satlug at sbcglobal.net> wrote:

From: Don Wright <satlug at sbcglobal.net>
Subject: Re: [SATLUG] 1st question - RAID5 quit - Short tests completed
To: "The San Antonio Linux User's Group Mailing List" <satlug at satlug.org>
Date: Monday, August 17, 2009, 1:40 PM

On Mon, 17 Aug 2009 11:42:40 -0500, Jeremy Mann <jeremymann at gmail.com>

>Since only 1 drive is bad, why can't you just mark that drive as
>offline, shutdown the computer and put a new drive in, and hotadd the
>new drive to the array? This should start the rebuilding process

Jeremy & Sam: Right now mdadm thinks two out of four drives on the
RAID-5 are bad. There's a fair chance that the third drive only had a
small RAID hiccup and can be reinserted, but that hasn't been tested

David: Regarding /dev/sdc that failed the short test. First, power down
the system and disconnect the AC power cord at both ends.[1] Remove and
reattach the drive power cable and both ends of the data cable. This
will wipe the electrical contacts and should improve any intermittent
connection. If there is a SATA power adapter (short cable that connects
the old 4-pin power plugs to the new flat SATA power) then disconnect
and reconnect the 4-pin side of that, too. 

After reconnecting the drive power and data cables, semi-gently tug on
the individual wires of any 4-pin power connections. These can get loose
and deliver bad power. If you find any loose ones, disconnect them and
locate the side with the open barrel contact. You should see a gap along
one side so the metal looks like a nearly-closed 'c'. I generally use a
small screwdriver to reduce the diameter of the 'c' by gently wedging
the screwdriver between the metal and the plastic body. I go all around
the outside of the metal contact to attempt a uniform reduction in the
diameter. Reattach the 4-pin connector and try the tug test again. If
you still find loose contacts, repeat the procedure. If all is well, you
may reconnect the AC power cord and repeat the short test on /dev/sdc.

>From the first post I conclude the data on the array is important to
you. To recover it I'd try SpinRite from www.grc.com. It's been saving
data for a long time, and the current release works with non-Windows
partitions. You need a Windows machine briefly to create a bootable
image on floppy, CD, or USB flash drive, but after that it's
self-contained. If this data is super-important and extremely difficult
to rebuild manually, then STOP NOW and start shopping for a drive
recovery service. You may not need to go this route if the other three
drives are OK, but we haven't gotten that far yet.

I have to run, so the others on this list can have at you for a while.
Please post the results of 
"mdadm --examine /dev/sda1 /dev/sdb1 /dev/sde1" for us to chew on.

[1] These instructions are (overly?) detailed because you (an
experienced system builder) are not the only one reading them.

A computer is a hole in your desk you pour time into.
SATLUG mailing list
SATLUG at satlug.org
http://alamo.satlug.org/mailman/listinfo/satlug to manage/unsubscribe
Powered by Rackspace (www.rackspace.com)

More information about the SATLUG mailing list