smartd - Offline uncorrectable sectors

Issues related to hardware problems

smartd - Offline uncorrectable sectors

Postby kilko » 2011/02/14 16:11:07

I recently installed two 160Gb hdds, and use these with striped LVM technics.

..now I came across /var/log/messages these lines... :-/
---
Feb 14 16:50:44 centos55 smartd[4826]: Device: /dev/sdf, 34 Offline uncorrectable sectors
Feb 14 16:50:44 centos55 smartd[4826]: Device: /dev/sdg, 14 Offline uncorrectable sectors
---

To my understanding, this means that the disk is beginning to fail, and should be replaced..

Short story..
I have about 35.000 images on these disks... (shared on network, as a samba share with PCs using Picasa)
I've doing backup to 2 separate USB disk, and sometimes I connect a third usb disk and transfer to this aswell.


...I'm worried now that "some" of my image/data might have been lost - due to the 2 lines above...
(checking if something is missing, of the 35.000 images is not easy.. and I have not made any txt file with all files/names/size og the images -...I see now that I also should create such a list, and not just take backup)


How critical is this "Offline uncorrectable sectors" and how can I check/do something about it ?
kilko
 
Posts: 34
Joined: 2009/06/03 08:25:28
Location: Norway

smartd - Offline uncorrectable sectors

Postby pschaff » 2011/02/14 16:41:41

This HowTo may help.

If you really care about the contents you may want to consider a RAID with some redundancy when you replace the disks. Striping can increase performance, but decreases the reliability.
pschaff
Retired Moderator
 
Posts: 18277
Joined: 2006/12/13 20:15:34
Location: Tidewater, Virginia, North America

Re: smartd - Offline uncorrectable sectors

Postby kilko » 2011/02/14 20:31:04

Been reading on that link HowTo and found that this is good to know howto repair if stuff goes wrong..

I have been using Raid before.. and it been great using it,.. when it works that is.. i.e. until one disk fails. (I used striped that time too..)

So I found the safes and easiest is to have several USB disk that I mount, using bash script, copy the content and unmount again..
(i mount by using /dev/disk/by-id/ - to make sure I mount the correct drive)

I think I'm just going to take out those 160Gb disks, and replace them..
Since their seems to be starting to fail.. I do not want to have any problems related to disk-IO etc..

Thank you pschaff for helping and commenting on the issue.
kilko
 
Posts: 34
Joined: 2009/06/03 08:25:28
Location: Norway

Re: smartd - Offline uncorrectable sectors

Postby rnichols » 2011/02/14 20:50:12

If in the course of copying your data the OS tries to read one of the uncorrectable sectors, you will get an I/O error. You can be reasonably confident that any files that were copied without an I/O error are OK.

Before you invest in RAID, get a proper backup for your data. Write the following on the blackboard 100 times: "RAID is not a backup." RAID does not protect against accidental deletion or overwriting, the OS scribbling on the disk and corrupting the file system, or physical destruction or theft of the machine. Most home users have little need for RAID. RAID is primarily targeted for commercial environments where it is important to keep the system up continuously while a failed disk can be replaced. For the average home user, having the system down for a few hours or even a day or two while a new disk is installed and restored from backup is not a big deal.
rnichols
 
Posts: 38
Joined: 2010/05/30 17:47:47

Re: smartd - Offline uncorrectable sectors

Postby TrevorH » 2011/02/15 01:19:23

kilko wrote:

How critical is this "Offline uncorrectable sectors" and how can I check/do something about it ?


It's nasty enough so that you want to think about replacing the disks, especially if those counts carry on increasing. As a temporary measure you could try running `badblocks -n` against the drive whilst it's not mounted and see if that either increases the number (bad news!) or leaves it much the same. Also probably worthwhile running smartctl to initiate a long self-test and leave it to complete (or error out).
User avatar
TrevorH
Forum Moderator
 
Posts: 9132
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: smartd - Offline uncorrectable sectors

Postby pschaff » 2011/02/15 02:57:51

kilko wrote:
,,,. and it been great using it,.. when it works that is.. i.e. until one disk fails. (I used striped that time too..)

That's why I said "with some redundancy". RAID0 has none.

Thank you pschaff for helping and commenting on the issue.
pschaff
Retired Moderator
 
Posts: 18277
Joined: 2006/12/13 20:15:34
Location: Tidewater, Virginia, North America


Return to CentOS 5 - Hardware Support

Who is online

Users browsing this forum: No registered users and 1 guest