I assigned the 3 drives before preclearing. I'm trying to preclear now but the 3 drives aren't showing up. Array is stopped. What can I do to preclear at this point?

 

Edit: I can do it via ./preclear_disk.sh /dev/sdX but not with screen it seems.

You cant run a preclear on assigned drives. You need to unassigned them if you want to preclear them.

You need to search the unraid forum and read the first page of the preclear posts

By assigned drives, do you mean the parity, disk1 and disk2 on the main page?

Yes. If you assign a drive to parity or data you can't then preclear it. The idea behind preclearing a drive before assigning it to the array is to stress test the drive and hopefully spot a possible weak point in the drive before adding it, i.e. assigning it to the server, it also to write zero's to the drive and a signature that is recognised by unRAID. If you just assign the parity drive and or a data drive that is not already precleared then first it will clear the drive and then it will format it. This can mean the server being unavailable for many hours. If your have already setup you array then unless the drives are all new I would leave them and run smart check on the drives. If you have a data drive fail in the future and you add a non precleared drive to replace it then the server would be offline for many hours while it first preclears, then formats and then rebuilds the failed data drive, where as a drive that is already precleared would take a lot less time.

If you have a data drive fail in the future and you add a non precleared drive to replace it then the server would be offline for many hours while it first preclears, then formats and then rebuilds the failed data drive, where as a drive that is already precleared would take a lot less time.

Close, but not quite. Preclear allows you to add a new slot assignment to a parity protected array without the lengthy clearing downtime. Replacing an existing slot (either failed drive or upgrading) doesn't require zeroes, because it's written with the prior contents anyway. Unraid runs through the rebuild process without first verifying that the drive is good, so a preclear is HIGHLY recommended on a replacement drive (or any drive really) but it doesn't actually save any time on the rebuild process.

Yeah, I just assigned them when I booted the machine. This is the maiden voyage for this machine so I'm not too worried at the moment. I was able to start the preclearing process on all drives.

 

I'm really just in the testing phase so that I can learn about unraid before I get data on it that's important.

Thanks for helping out Johnathanm, I was always wondering if when replacing a failed drive unraid looks for a preclear signature or not. So if your replacing or upgrading a drive with a previously good working drive the perclear is not needed. Obviously on a new replacement drive or assigning a new drive it's advisable to avoid a lengthy downtime. That's saved me some work this weekend as a friend is replacing some of his 3tb drives on his unraid server, with some of my 4tb drives that were previously working perfect in my unraid server.

Thanks

That's saved me some work this weekend as a friend is replacing some of his 3tb drives on his unraid server, with some of my 4tb drives that were previously working perfect in my unraid server.

I recommend pulling smart reports on all the drives involved, and keeping an eye on the syslogs during the procedure. Because of the way unraid currently works to upgrade drives, you are asking all your remaining drives to be perfect for the duration. Part of the upgrade procedure should be a non-correcting parity check after the upgrade is done, that way you can be confident what was just written to the drive is being read correctly. If you are doing multiple drives, you need a check after each upgrade. If the first upgrade didn't write correctly, it would corrupt the subsequent upgrades.

 

Also, even though the 4TB drives test fine in your current box, the drives are only one link in the chain. Sata and power connections will be different in the new box, as well as a different controller.

 

I'm not saying you will have problems, everything will probably go fine, but it's always wise to keep an eye on things whenever you are making changes in the box.

 

Keep the drives that are being pulled out intact until you are sure everything went well.

Both servers are identical in every way apart from the drive sizes. The data is already backed up on my server and offsite. Can you explain the idea behind the non correcting parity check after the data rebuild as opposed to write corrections to parity. Thanks

Can you explain the idea behind the non correcting parity check after the data rebuild as opposed to write corrections to parity. Thanks

Sure. The data rebuild process works by reading all the drives, parity included, and doing the calculation to derive the missing bit and write it to the new drive. If that write process is accepted by the drive, but for some reason isn't able to be read correctly, like maybe a marginal sector that hasn't been caught yet, the non-correcting check will throw an error when it gets to that bit. Any errors on the check indicate the data wasn't written to the new drive correctly, so we know for sure that the error is on the new drive, not the parity drive. Unraid trusts the data drives by default, so a correcting check would read the data that was written in error and update parity to reflect that error, and you would have no way to correct it. If the non-correcting check fails, you should pull the newly replaced drive and try the rebuild operation again, probably on a different drive just to be thorough, or preclear the drive you just pulled to force any sector reallocation that needs to happen.

 

Clear as mud?

Yep, very easy to understand thanks.