Yesterday I started a disk copy from an NTFS drive to unRAID.  This morning when I checked the unRAID server I see a disk with status of "DISK_DSBL".  This is a 1tb disk that was 88% full before the copy and now shows 91% full.  The array does have parity.  I tried checking the SMART status of this drive and it failed to display it.  I powered down the server and rebooted.  The array started but the disk status is the same.  This time SMART did run and indicated imminent failure - something like 1200 relocated sectors.  The drive is toast - I get that.

 

My question now is how do I recover and was the data being copied during the failure copied?  The copy was done to a user share and checking the telnet session shows no errors.  I've got a replacement drive in the server already.  It's not attached to the array and is currently formatted as NTFS (there is no data on it I need).  What is my best course of action from here?

syslog-2010-03-09.zip

Yesterday I started a disk copy from an NTFS drive to unRAID.  This morning when I checked the unRAID server I see a disk with status of "DISK_DSBL".  This is a 1tb disk that was 88% full before the copy and now shows 91% full.  The array does have parity.  I tried checking the SMART status of this drive and it failed to display it.  I powered down the server and rebooted.  The array started but the disk status is the same.  This time SMART did run and indicated imminent failure - something like 1200 relocated sectors.  The drive is toast - I get that.

 

My question now is how do I recover and was the data being copied during the failure copied?  The copy was done to a user share and checking the telnet session shows no errors.  I've got a replacement drive in the server already.  It's not attached to the array and is currently formatted as NTFS (there is no data on it I need).  What is my best course of action from here?

Your situation is easy to recover from.

 

1. Stop the array

2. Go to the devices page, assign the replacement drive to the slot where the disk has failed.

3. Go back to the main page.

It should say something like "upgrading disk" to the right of the "Start" button.  

4. Press the "Start" button to begin the process of reconstructing the old disk's contents onto the replacement.   Since parity was written to when the data disk failed it will have the correct data to be able to re-create what had been on the failed disk.     You will not see a "format" step since the old disk's formatting is automatically copied to the replacement disk as it re-constructs it.   Therefore, it does not matter what type of file-system was on the replacement disk previously, it will be overwritten.

 

Depending on the size of your array it could take from a few hours to 8 or more to reconstruct the disk.  (It will take just a bit more time than a parity check, since it has to read every sector on every other data disk and write every sector on the replacement disk)

Whatever you do, DO NOT press the button labeled as "restore" as it will immediately invalidate parity and set a new disk configuration without the failed disk's data.

 

Since you may want to RMA the defective drive I wanted to let you know why I did not have you remove it physically just yet. 

 

did not want you dislodging any other cables by accident until the replacement drive was re-constructed. (until it is, that would be a two disk failure, something parity cannot recover from)

 

Once the re-construction of the replacement drive is complete, you can stop the array, power down, remove the failed drive, then power back up.

 

Your situation is easy to recover from.

 

1. Stop the array

2. Go to the devices page, assign the replacement drive to the slot where the disk has failed.

3. Go back to the main page.

It should say something like "upgrading disk" to the right of the "Start" button.  

4. Press the "Start" button to begin the process of reconstructing the old disk's contents onto the replacement.   Since parity was written to when the data disk failed it will have the correct data to be able to re-create what had been on the failed disk.     You will not see a "format" step since the old disk's formatting is automatically copied to the replacement disk as it re-constructs it.   Therefore, it does not matter what type of file-system was on the replacement disk previously, it will be overwritten.

 

Depending on the size of your array it could take from a few hours to 8 or more to reconstruct the disk.  (It will take just a bit more time than a parity check, since it has to read every sector on every other data disk and write every sector on the replacement disk)

Whatever you do, DO NOT press the button labeled as "restore" as it will immediately invalidate parity and set a new disk configuration without the failed disk's data.

 

 

Thank you Joe.  I figued the procedure was something like this but since this is my first time recovering from parity I wanted to make sure.  One of my main concerns is that the data being copied WAS copied and it is intact after rebuilding.  That's rather cool that the data would be safe even when written to a failing/failed disk -essentially it just writes to the parity disk?  I also didn't realize that the format of the replacement disk didn't matter.  I figured I probably needed to use preclear to zero it out.  Thinking about it though it makes sense. 

 

I definitely agree with your other comments about pulling the disk and had no plans of doing that yet.  Besides the points you brought up, which are most important, it's a bit of a pain to remove the drives in my case so I figure I'll do an advance replacement so I can put the new drive in when I pull the bad one.