I replaced the parity drive in my UnRAID with a new, larger drive.  When I started the array, I began rebuilding the parity drive.  About 10% through the process I started getting read errors one of the data drives -- thousands of them.  I ran the SMART tool on the drive with the errors and it showed over 600 remapped sectors so it appears the drive is going bad although it still shows as 'PASSED' and none of the individual tests failed.  Unsure of what to do, I cancelled the parity rebuild and started copying the data from the failing drive to a separate external drive to salvage my data.  That copy has now failed due to read errors.  I still have the original parity drive and had completed a parity check just last week.  I've had no errors until I started rebuilding parity with the new parity drive.  I also made no changes to any data on the array during the first 10% of the parity rebuild. 

 

Here is my question:  Can I put the old parity drive back in and get the array back to the state it was before I made the parity drive switch?  I then want replace the failing data drive and do a rebuild of the data.  Once that is done, I will proceed with the parity drive replacement.  If all of this is the correct approach, how do I proceed?

 

Thanks!

 

Jeff

I replaced the parity drive in my UnRAID with a new, larger drive.  When I started the array, I began rebuilding the parity drive.  About 10% through the process I started getting read errors one of the data drives -- thousands of them.  I ran the SMART tool on the drive with the errors and it showed over 600 remapped sectors so it appears the drive is going bad although it still shows as 'PASSED' and none of the individual tests failed.  Unsure of what to do, I cancelled the parity rebuild and started copying the data from the failing drive to a separate external drive to salvage my data.  That copy has now failed due to read errors.  I still have the original parity drive and had completed a parity check just last week.  I've had no errors until I started rebuilding parity with the new parity drive.  I also made no changes to any data on the array during the first 10% of the parity rebuild. 

 

Here is my question:  Can I put the old parity drive back in and get the array back to the state it was before I made the parity drive switch?

Yes, you will need to use a modified version of "trust my parity" as described in the wiki.  Instead of issuing a command stating the the invalid drive is drive 99, you will want to issue one with the drive number of the actual failing drive.

  I then want replace the failing data drive and do a rebuild of the data.

It should work.

Once that is done, I will proceed with the parity drive replacement.  If all of this is the correct approach, how do I proceed?

Thanks!

 

Jeff

Read everything here several times before you do anything.  If you have questions, ask before you do anything.

 

Basically,

Stop the parity calc onto the new parity disk.  It is worthless, since you data cannot be read from the data disk.

Power down, put back your original parity drive.  Install an appropriately sized data drive in place of the one failing.  That drive MUST be as big or bigger than the failing data drive, but not bigger than the parity drive.

Power up.  The array will probably not start (because it will see the new disks)  You might need to assign the old parity drive as parity and the new data as the replacement for the one that has failed on the devices page.

Back on the main page, Press the button labeled "restore" but DO NOT START the array just yet.  Not until after you enter the next command and get an OK response.

Log on via telnet or on the system console as "root" and Type:

 

  cd

  mdcmd set invalidslot N

 

where N = the number of the failing disk.  If disk4 was failing, type "mdcmd set invalidslot 4", if disk9 was failing, type: "mdcmd set invalidslot 9"

Then, you can press "Start" and the array will begin the process of rebuilding from parity and the other data disks to the new replacement for the disk that failed.

 

The button labeled "restore" has nothing to do with restoring data.  It is actually a "Set Disk Configuration" button.  It sets the configuration to that of the currently assigned and working drives. (Which you will have already assigned on the "devices" page)  normally it immediately invalidates parity.  You can override that and force a different slot to be considered invalid using the command I listed above.  By doing so, the array will then begin the process of reconstructing the new replacement data disk.

 

Joe L.

Thanks.  That did it.  I started the process on Friday night and it finished last night.  Everything is working fine.