I have a 2Tb drive in my array that seems to be failing.

 

Whilst doing a parity check the drive started to show many read errors, so stopped the check and I took the drive out and put it into my Windows PC and copied off the data that I could onto another drive (there was only a few files that could not be copied and they are easily replaced).

 

However I then put the drive back into the array and continued the restarted the parity check which took a long time as the drive kept showing errors that seem to be read errors according to the syslog. The PC has now finished and the drive shows with a green ball but a smart check shows an increasing number or sectors about to be re allocated so I think I definately need to change the drive. I have a 2Tb drive that has just finished pre clearing and I want to replace the dieing drive.

 

Should I let UNRAID rebuild the drive from the parity or replace the drive and transfer the data to the new drive from my backup?

 

Ridley

Did you do a non-correcting Parity Check or a Correcting PArity Check?

 

If you did a non-correcting parity check, I would allow unRAID see if it could rebuild the disk.  If the disk failure occurred after the disk started to go flakey, you might be able to recover the entire contents of the disk.  If a rebuild doesn't work, then you have the option of installing the new disk as a "New" disk and setting up a new configuration.   

 

Keep that backup of the failing disk on your Windows computeruntil you are satisfied that everything is OK with the new one.   

I have had loads of problems recently and it was while upgrading the parity drive that the other drive started to fail so I had two invalid drives so the parity was not valid when the drive started to fail.

I will give my thoughts on what I would do-- using the following assumptions. 

 

First, your parity drive does not have valid parity since the other/second drive failed when you were rebuilding parity on the parity drive.

 

Second, you have copied all of the readable files off of that second drive.

 

Third, you have precleared the replacement drive and it passed the preclear tests.

 

I would now boot the server with the defective drive in it.  Now, get all of the drive assignments--- Disk numbers and serial numbers for every drive. 

 

Setup a new configuration without that drive using the 'New Config' utility under the Utils Tab of the GUI.  (I believe this is much easier on recent version rc versions.) See the following links for procedure:

 

      http://lime-technology.com/forum/index.php?topic=23575.msg207851#msg207851

 

      http://lime-technology.com/wiki/index.php/FAQ#How_do_I_remove_a_hard_disk_that_I_do_not_plan_on_replacing.3F

 

You will have to rebuild parity on the new configuration.  When that is done, you will have a new working array with correct parity for the drives that you have installed.  (You should be able to remove the defective drive at this point.  Since I am somewhat anal    , I would next perform a non-correcting parity check to verify that everything is working properly.  If I got zero errors on the non-correcting check, I would install the new precleared drive as a 'new' drive.  After it was installed, I would copy the files back over to the new drive and run a another non-correcting parity check.  ( I said I was anal!!!) 

 

In the interest of full disclosure, I must say that I have never done this.  But since no one has put forth any suggestions on how to proceed, I decide to outline what I would do.  If I have made any mistakes in how to do it, I would hope someone who is more knowledgeable than me would jump in ASAP!

As Frank outlined above, you do NOT have good parity -- so you can NOT rebuild a drive.    Don't even bother to try.

 

What Frank outlined is what you need to do ==> essentially simply start over with your configuration.    Attach your GOOD drives; do a New Config ... assigning ONLY your good drives and NO parity drive.    Start the array and confirm everything looks okay ... no drives show "unformatted", and you can access the array on your network.    Then Stop the array; assign a parity drive; and restart the array => this will initiate a parity calculation (several hours).    When that's done, run a parity check (another long process) ... and when that completes, assuming it has no sync errors, than all is okay.