I have a 2TB drive, factory sealed, in case I lose one of the drives in my array. Should I preclear this drive and then put it back on the shelf? Or should it stay factory fresh? I am looking for a quick / safe recovery of the array.

 

thanks

I would suggest you preclear the drive, I have two drives that are precleared but not connect to the array

I have a 2TB drive, factory sealed, in case I lose one of the drives in my array. Should I preclear this drive and then put it back on the shelf? Or should it stay factory fresh? I am looking for a quick / safe recovery of the array.

 

thanks

Absolutely. pre-clear it.

 

Lately about 1 in 5 brand new drives shows some issues when first tested/exercised with the preclear script.

 

Are you looking for a 4-out-of-5 chance for a "quick / safe recovery of the array" or something closer to 5 out of 5?

Since you don't need it immediately in your array, many would advise you run it through several cycles.

 

One more thing... unRAID is NOT hotswap.  You must power down to swap out a drive.

Joe L.

Thanks, will get it installed tonight and start the preclear

One more thing... unRAID is NOT hotswap.  You must power down to swap out a drive.

 

Wellllll.... you can do this:

 

1) keep pre-cleared hot spare in the server.

2) on failure, add hot spare to array config.

3) copy from (simulated) failed drive to hot spare.

4) zero out bad drive.

5) stop array and remove zeroed-out bad drive from array config.

6) restart array, and trust parity.

One more thing... unRAID is NOT hotswap.  You must power down to swap out a drive.

 

Wellllll.... you can do this:

 

1) keep pre-cleared hot spare in the server.

2) on failure, add hot spare to array config.

3) copy from (simulated) failed drive to hot spare.

4) zero out bad drive.

5) stop array and remove zeroed-out bad drive from array config.

6) restart array, and trust parity.

 

You must copy,in step 3, using dd not cp.

One more thing... unRAID is NOT hotswap.  You must power down to swap out a drive.

 

Wellllll.... you can do this:

 

1) keep pre-cleared hot spare in the server.

2) on failure, add hot spare to array config.

3) copy from (simulated) failed drive to hot spare.

4) zero out bad drive.

5) stop array and remove zeroed-out bad drive from array config.

6) restart array, and trust parity.

 

You must copy,in step 3, using dd not cp.

actually, if the new drive has been added in step2, and formatted, then any method of copying will work.

 

Joe L.

You must copy,in step 3, using dd not cp.

 

No, as Joe pointed out cp -rp will work fine.

One more thing... unRAID is NOT hotswap.  You must power down to swap out a drive.

 

Wellllll.... you can do this:

 

1) keep pre-cleared hot spare in the server.

2) on failure, add hot spare to array config.

3) copy from (simulated) failed drive to hot spare.

4) zero out bad drive.

5) stop array and remove zeroed-out bad drive from array config.

6) restart array, and trust parity.

 

You must copy,in step 3, using dd not cp.

actually, if the new drive has been added in step2, and formatted, then any method of copying will work.

 

Joe L.

 

Your right. How does one zero out the bad drive?

One more thing... unRAID is NOT hotswap.  You must power down to swap out a drive.

 

Wellllll.... you can do this:

 

1) keep pre-cleared hot spare in the server.

2) on failure, add hot spare to array config.

3) copy from (simulated) failed drive to hot spare.

4) zero out bad drive.

5) stop array and remove zeroed-out bad drive from array config.

6) restart array, and trust parity.

 

Why such a convoluted recovery method? What's wrong with simply setting the failed disk pull-down to the newly installed replacement drive, checkig off the confirmation box and clicking start? The array is not protected as long as you run while simulating a failed drive. Letting unRAID rebuild to a new drive is the quickest way to get the parity protection back.

 

Peter

 

Why such a convoluted recovery method?

 

For one, you can keep the array up longer, and get back to a safer state with no shutdown.  I've had indexings running for several days, and I don't want to have to take the array offline.

 

With a little tweaking, unRAID could actually do it all hot, without taking the array offline with an AHCI mobo.

Huh, you have to stop the array to change drive assignments with either method. The array can still be accessed while the disk is rebuilding.

 

Peter

Yes, but WHEN you stop the array can be important.  Suppose you want to delay stopping the array for a few days, but want to protect the data from a second drive failure?

 

 

Copying the data from a failed drive to another drive outside the array does not protect you from a second array drive failing.

 

2) on failure, add hot spare to array config.

 

To do this step requires you to stop the array. At this point, you might as well just set the new drive as a replacement for the failed drive before you restart the array.

 

You could mount the drive outside of the array and copy the data off it. This still does not provide any protection against a second data drive failing unless you use the dd command to make a duplicate of the failed disk and DO NOT write to the failed drive after making the duplicate.

 

Peter

 

To do this step requires you to stop the array.

 

Correct.  The drive needs to already be in the array config, so step 2 needed to say that.

Another reason it is a good idea to preclear / burn in a new drive soon after receiving it - if you do have a DOA, it is usually easier to deal with returning it to Newegg (or other e-tailer) than it is returning it to the manufacturer.  Plus, manufacturer may send you a refurb and not a new unit.