Call me anal, but I have a minor hodgepodge of disks on my server and I want all the same disks to be numbered together (as far as the logical disk numbers assigned by unRAID, e.g. disk1, disk2, etc.)

 

I am not using parity, and I do not have any shares that span more than 1 disk, so I am hoping this is easy?

It's simple.

 

Stop the array.

 

Go to the Devices tab.

 

Re-organize your disks how you'd like them.

 

Go back to the main page and check the "restore" box and then "Start."

 

This will record your new array configuration and all the data on your disks (along with your shares) will be untouched.

That will work, but will leave the array exposed to disk failure, not to mention take hours.

 

Here is a much easier way:

 

http://lime-technology.com/forum/index.php?topic=364.msg2539#msg2539

 

I just added this link to the "Best of the Forums" - "Hail to the chief (Messages from Tom)" section!  (See sig for link)

As long as the drives are re-assigned within the same set of unRAID disk names, I think you can stop the array, go to the devices page, re-assign any "data" disk to any other "data" slot, go back to the main page.  It will state something like "Disk positions have changed" and will want you to check a checkbox below the start button.  Check it, the new disk positions will be recorded, and the array will not even check parity, as it will still be valid.    (If you did have a parity drive, you would have to make sure you did not re-assign it)

 

I experimented by re-assigning 4 of my data drives a week or so ago when one post in a thread on the unRAID forum mentioned it might result in faster performance if adjacent disks in the array were not on the same cable.  I stopped the array, re-assigned every other drive to move them around to other controllers, and then re-started.  I used the "Start" button after checking its "are you sure" checkbox.   I did not press "Restore" as it is only needed it you are removing a disk from your array and not replacing it.    As I said, by using the "Start" button, the array started and I did not need to re-calculate parity.

 

Since you have no parity drive, pressing "Restore" has no effect different than just re-assigning your drives, checking the box below "Start" and re-starting the array.

If you did have a parity drive, it would make a huge difference.  "Restore" would throw away the parity data and set about recalculating it.  Some number of hours later (4/6/8?) it would have a new parity disk calculated.   Pressing "Start" as I did would have you up and protected in seconds.

 

Clairification:  If you currently have disks1 through 5 assigned, and wish to move the assignments around re-using disk1 through disk5, you can just restart the array.  If however, your new assignments use a previously unassigned slot (disk10 for example), or vacate a currently assigned slot you must use the "restore" button to delete the existing super.dat data so unRAID can store a new configuration.  Otherwise, unRAID will state invalid configuration.

 

Joe L.

That seems to have worked.  I took the server down, installed the new driver, brought it up, put all the drives in their normal slots, started the array, formatted the new drive, stopped the array, reorganized the disks in the devices tab, webt back to main, clicked on "I am sure I want to do this", and clicked start.

 

Everything came up, but my shares still listed the "old" disk assignments, so I went through and changed the disk assignments one share at a time, for the affected shares.  I hope that works ok.  Thanks for the help guys!

What am I doing wrong? All I want to do is match the disk numbers to the actual slots in my tower.

I don't see anything wrong, so if there is, you will have to clarify.

 

What I see is 5 drives moved 1 slot, and the picture is showing the new configuration plus the changes from the old configuration.  If you were trying to follow the advice given above, that was based on not having a parity drive, which is a very different scenario to what you have.  Parity is based on a particular disk configuration, in this case, the previous one, that had a drive in Disk 3, and none in Disk 8, and different drives in Disks 4 through 7.  So although the parity drive is still green (perhaps wrongly), it is no longer valid, since it is based on the previous disk config.  In this case, *you* know that it is the very same group of disks, but unRAID does not, it sees a different disk config.  By a new 'trick' that Tom has revealed, we can still preserve the parity drive though.

 

Once you check the I'm sure I want to do this box, and click the Restore button, you will have deleted the old configuration, and the new disk configuration will be saved, but parity will be considered 'Invalid'.  To re-validate your parity drive, you will need to go to the console (or Telnet box) and type Tom's special command, see here for instructions beginning with step 2.  As he says, a parity check will begin, which you can abort or allow to complete, your choice.

 

Added to FAQ (link back only), here.

I would try assigning disk8 to the disk3 slot.  It thinks disk3 is missing and disk8 new.

 

It will probably let you start the array that way, then stop it and try moving disk3 to disk8.

 

It should let you move the disk from 3 to 8 as long as it is part of the array.  Looking at the code of the md device, it can handle moving disks between currently assigned slots, but not from a currently assigned slot to a previously unassigned slot. If the logic was made a tiny bit smarter it could, and it probably is an improvement Tom can look into.  Once it thinks a disk is "missing" it will not let you restart the array.

 

As it is in your screen shot, the only way to get your array back online is to check the checkbox below "Restore" and then press restore to delete the existing super.dat and create a new one based on the current disk assignments. This will also force a new parity calc, as the current one will be invalid as soon as you press restore.   

 

I see RobJ described the special way to have parity considered correct after pressing restore, but before pressing start.  Use it only if you are sure all the hardware is working correctly.  I'd let the resulting parity check complete, even if it finds no errors.

 

Joe L.

I did exactly as described, and it seems to be working fine, no errors on the parity check so far. It would be nice to have the drive logic able to find existing good drives moved to previously empty slots without having to go through this again. My anal retentive organization wants the parity drive physically at the top of my machine, and by default that means the drives below that need to be in order largest to smallest. The bottom set of drives are my IDE drives, my top nine are my 3 removable tray 3 bay drive modules. Every time I can afford a new 1TB drive, it goes next in line at the top, and everybody else moves down 1. I suppose it would be pushing my luck to stop the parity check and add my new 1TB to the array right now. I'll wait for a clean parity result.