[Novalug] smart and bad sectors

Mark A. Metz mametz@aol.com
Tue Jul 9 06:15:59 EDT 2013


By following a 3-2-1 backup plan for all your important data, a failed hard drive shouldn't be too much of a concern.
But to address your initial question, I have an external drive that has had an imminent failure warning for a good three years now.

Maxwell Spangler <maxlists@maxwellspangler.com> wrote:

>On Mon, 2013-07-08 at 21:10 -0400, Jon LaBadie wrote:
>
>> Based on sectors reallocated to spare sectors,
>> when should one consider replacing a drive?
>> 
>> I've got a 1.5TB drive that "smart" still rates as
>> good.  A few days ago I noticed it had over 400
>> sectors reallocated.  Not a huge number 400 out of
>> 3 billion but more than I'm used to seeing.
>> 
>> I set up a cronjob to tell me if the number was
>> increasing or stable.  After 3 days the count
>> was up by 2.
>> 
>> What guidelines do you use to decide if a drive
>> should be replaced?
>
>
>In my opinion, you should replace that drive with something more
>healthy
>and trustworthy (especially if it has important data on it.)
>
>When I see reallocated sectors, I go from trusting the drive with blind
>faith to worrying about it.
>
>I put the worrisome drive on a bench system running Linux and run
>badblocks scrubbing for as long as possible.  The longer it runs
>without
>killing the drive or increasing the reallocated sector count, the more
>I'm willing to trust it.  And then, I'd like to see that drive as part
>of a RAID 10 type setup where it can fail abut its buddy will still
>keep
>going.
>
>I had a few anecdotal things happen to me in the last few years along
>these lines:
>
>* I had a six drive RAID 10 array fail.  It's likely that somewhere in
>that RAID 10 setup drive A failed and then buddy drive B failed and the
>array was lost.  When I pulled all drives, two were completely DOA and
>three had reallocated sectors.
>
>I put those three drives on a Linux workbench system and ran badblocks
>for.... 30 days.  I just left it running at a remote site hoping all
>the
>activity would kill the drives.  Then, I could use the warranty on them
>to get replacements.  But they didn't die.  So at this point, I
>actually
>trust them for the stress they've been through, but I still feel they
>could fail at any time.
>
>* Just a few days ago I found myself bench marking a NAS raid array to
>try to figure out why the unit could never provide good performance
>despite good CPU, memory and gigabit Ethernet.  A close inspection
>using
>'atop' showed that three of four drives had equal I/O loads but the
>fourth drive had twice as much outstanding I/O and delay.  Upon
>investigating the 4th drive's SMART data I found reallocated sectors.
>This led me to believe that the drive was encountering them while part
>of the RAID array and re-try efforts were slowing it all down.  I was
>using old drives for experimentation purposes and after going through 3
>more swaps, I finally got to a truly healthy drive and got the
>performance I wanted.
>
>The bottom line for me is that when you buy a drive new, you have blind
>faith that it's reliable.  It's probably reliable but there is really
>no
>way of knowing how long that drive will last.  But you can enjoy the
>blind faith.
>
>Once the drive starts to encounter reallocated sectors, reality hits
>that all drives fail and yours might be failing soon.  So you should
>(as
>you are) be responsible with moving critical data off it and ensuring
>you have backups.  Stress the drive to regain confidence but you'll
>never get that "it's a new drive!" blind faith back.
>
>Finally, when you think a drive is failing, try to get it to fail and
>call in for a warranty replacement.
>
>If you can't.. Don't trust the drive.  Get rid of it and avoid going
>through the frustrating process of dealing with an unscheduled,
>unexpected dead drive.
>
>I hope this helps!
>
>
>
>
>-- 
>Maxwell Spangler
>========================================================================
>Linux System Administration / Virtualization / Development / Computing
>Services
>Photography / Graphics Design / Writing
>Fort Collins, Colorado
>http://www.maxwellspangler.com
>
>
>------------------------------------------------------------------------
>
>_______________________________________________
>Novalug mailing list
>Novalug@calypso.tux.org
>http://calypso.tux.org/mailman/listinfo/novalug
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.firemountain.net/pipermail/novalug/attachments/20130709/37ebe80c/attachment.htm>


More information about the Novalug mailing list