The solution, if you continue to get un-correctable media failures while reading the drive, is to replace the drive.

 

I see however, that there are no re-allocated sectors... This might indicate that it is simply all "soft" errors and they are being corrected as the "read" failures re-write the sectors based on the the disks in the array.

 

As an alternative to replacing the drive, you might see if the failures eventually slow down... (as it would not need to re-write the same sector twice)

Or, you could speed things up by stopping the array, un-assigning the drive that is failing, starting the array without it, then stopping it once more, re-assigning it, and re-starting it, having it completely write the contents back onto the failing drive. (basically it now thinks you replaced the drive with a new one)   Since all the errors seem to be soft errors, it might fix itself.   

 

That will cause you to be without parity protection during the time the rebuild takes... So if there are any critical files on your server, make copies of them on multiple disks... just in case a different disk fails too.

 

Thinking about it a bit more, you could perform almost the same re-write by simply moving the files on the failing disk to a different disk, and them moving them back... This would re-write everything while keeping parity protection.

 

Joe L.

 

I happen to have 3 new drives of the same brand (Seagate).  I just finished preclearing them and need to get my Pro license so I can add at least one to the array ASAP.  I was thinking though that maybe I should use one of the disks as a replacement and just swap it out and let parity rebuild the contents as if the drive failed?  I could then do more reading and writing to the suspect disk (would another preclear be worthwhile?) to see if it improves or gets worse.  What do you think?

The solution, if you continue to get un-correctable media failures while reading the drive, is to replace the drive.

 

I see however, that there are no re-allocated sectors... This might indicate that it is simply all "soft" errors and they are being corrected as the "read" failures re-write the sectors based on the the disks in the array.

 

As an alternative to replacing the drive, you might see if the failures eventually slow down... (as it would not need to re-write the same sector twice)

Or, you could speed things up by stopping the array, un-assigning the drive that is failing, starting the array without it, then stopping it once more, re-assigning it, and re-starting it, having it completely write the contents back onto the failing drive. (basically it now thinks you replaced the drive with a new one)   Since all the errors seem to be soft errors, it might fix itself.   

 

That will cause you to be without parity protection during the time the rebuild takes... So if there are any critical files on your server, make copies of them on multiple disks... just in case a different disk fails too.

 

Thinking about it a bit more, you could perform almost the same re-write by simply moving the files on the failing disk to a different disk, and them moving them back... This would re-write everything while keeping parity protection.

 

Joe L.

 

I happen to have 3 new drives of the same brand (Seagate).  I just finished preclearing them and need to get my Pro license so I can add at least one to the array ASAP.  I was thinking though that maybe I should use one of the disks as a replacement and just swap it out and let parity rebuild the contents as if the drive failed?  I could then do more reading and writing to the suspect disk (would another preclear be worthwhile?) to see if it improves or gets worse.  What do you think?

That seems t be a safe way of testing.  If the errors continue with the spare drive once it is swapped in, it would point to something other than the old disk you removed

 

Once the spare drive is installed, and the data re-constructed onto it, you can run a pre-clear cycle on the one with the read errors..  Don't clear it until after you rebuild the contents onto the new.

That seems t be a safe way of testing.  If the errors continue with the spare drive once it is swapped in, it would point to something other than the old disk you removed

 

Once the spare drive is installed, and the data re-constructed onto it, you can run a pre-clear cycle on the one with the read errors..  Don't clear it until after you rebuild the contents onto the new.

 

Just want to confirm that I should choose this option:

 

  Start will bring the array on-line, start Data-Rebuild, and then expand the file system.
I'm sure I want to do this

 

And not this one:

 

  Restore will initialize the stored array configuration; all drives will appear as New, but data disk contents are not affected.
I'm sure I want to do this

 

I assume the second option would be if I were just adding the drive to the array?  Also, I didn't see any reference to formatting the drive.  Will that come after I confirm option 1 above?  Thanks again!

That seems t be a safe way of testing.  If the errors continue with the spare drive once it is swapped in, it would point to something other than the old disk you removed

 

Once the spare drive is installed, and the data re-constructed onto it, you can run a pre-clear cycle on the one with the read errors..  Don't clear it until after you rebuild the contents onto the new.

 

Just want to confirm that I should choose this option:

 

  Start will bring the array on-line, start Data-Rebuild, and then expand the file system.
I'm sure I want to do this

YES, that is the correct option to rebuild an existing drive's contents onto a replacement.      It is the one you want to use when you have a failed drive and are replace it.  It is ALSO the one to use when adding an additional drive to the array.  Using this button ensures you have parity protection throughout the add of a new drive,

And not this one:

 

  Restore will initialize the stored array configuration; all drives will appear as New, but data disk contents are not affected.
I'm sure I want to do this

 

I assume the second option would be if I were just adding the drive to the array?  Also, I didn't see any reference to formatting the drive.  Will that come after I confirm option 1 above?  Thanks again!

Do NOT use this button to rebuild existing contents onto a replacement drive.  In fact, if you were to use the "restore" button (poorly labeled.  It should say "Initialize Array configuration") it will initialized the array configuration based on the currently assigned and working drives, forget ALL parity information, and then proceed with a new parity calculation as if you started the array for the very first time.  Pressing this button after swapping out a defective drive would cause it to forget everything on the old drive.  You would eliminate any possibility of reconstruction.  There are very few occasions where this button is used, mostly when hardware is changed, or a disk is REMOVED from the array and will NOT be replaced with another. 

 

Now, if you use the "restore" button when adding a disk to the array, it will initialize a new configuration, and forget the old parity as it will no longer be valid, and you will lose parity protection for the time needed for a full calculation... for that reason, the "Start" button is a better choice when adding an additional drive too.

 

As far as formatting, when replacing an existing drive with a new drive, and reconstructing contents onto it, the formatting is copied from the old drive to the new.  There is no separate formatting step.

 

Joe L.

Thanks for the detailed response Joe.  As it turns out, I'm glad I asked as I wasn't entirely clear on the meaning for the second option!

Well, I successfully installed the replacement disk, restored data from parity drive and got my array back up and running.  Attached is the syslog from the preclear of the "bad" drive.  It says that the disk precleared successfully but the ATA error count increased significantly and I've got a Current_Pending_Sector value of 1 as well as Offline_Uncorrectable value of 1.  Seems to indicate the drive is getting progressively worse.  I called Seagate customer service and they issued an RMA based on what I told them and without having to format the drive with a partition to run SeaTools.  The kicker of all of this is that I either have to pay $20 USD to Seagate to have them ship me a replacement (refurbished no less) drive and use their packaging to ship it back OR I have to find some packaging other than what the drive was shipped to me in from Newegg (wrapped in bubble wrap!) to keep from voiding the warranty!  It's unbelievable that a reseller can send me the drive in a package that would void the warranty if I used it to send it back knowing the drive is bad already!!  Oh well, guess I learned my lesson...

 

Anyway, I've attached a syslog from the preclear as well as SMART reports before and after just for confirmation of what I think I already know.  Thanks again for the support in troubleshooting this drive.