I've been building my unraid server byt migrating drives and data from my old server.  Because of case space constraints, I've had to pull working drives all set up with data from the unraid server to have room to add more new ones.  Whenever I do that I get an invalid expansion error (presumably because the new drive is occupying the location of a previously set up drive).  When this happens I've just been hitting the restore button and then it allows me to start the array and format the newly added drive.  What concerns me is that when I do this it nevers clears the drive first.  Will that be a problem I need to worry about?  I've read that the clearing procees is to ensure fault tolerance.  What shold I do?

When you click 'restore' it "restores" the configuration to the default state where all drives appear "new" (i.e., a newly constructed array).  Then clicking 'start' initiates parity-sync.  So if during this parity-sync time, a data disk happens to fail, you will not be able to rebuild it's contents (because parity is not completely valid).

 

The 'clear' function was implemented to work around this problem.  Suppose you have some number of drives in your system and parity is valid.  In this state if a drive fails you can replace it and rebuild the data on the drive.  Now you want to expand the array by adding another drive.  You can do this in two ways:

 

1. By the method you are using (add new drive, click 'restore', click 'start').

Advantages: fast - your shares are immediately available even though parity-sync is running in backgroud.

Disadvantage: risk of data loss if one of the original data disks fails during the parity sync.

 

2. By the method of adding a new drive to an unused slot (click 'start' and drive is cleared first).

Advantage: no risk of losing data if a drive fails before the clear completes.

Disadvantage: you have to wait for the clear to complete before being able to use the server.

 

By clearing the drive we are of course writing it with all zeros.  After this completes, we can seemlessly add the disk to the array and parity is still valid for the entire array (because data which is zeros does not change the overall parity).

 

The unRAID software currently executes a 'clear' before bring the entire array on-line.  This was done because it was expedient to code it this way (i.e., as an unRAID user myself I didn't mind waiting for clear to finish since I don't add drives very often).  However we are changing the code so that the array will start immediately and clear will commence in background - then after clear completes we'll bring the additional disk on-line.

Unfortunately, I didn't have much choice but to do it that way.  It was an existing 10 disk NTFS server that I was "converting" over to an unraid system.  To do this I started building the new server in a small desktop case so I could add the drives one at a time as I copied over their contents.  The case wasn't big enough to hold very many drives, so I eventually was forced to swap out drives so I could continue.  In doing this, unraid never offered me the option to clear the drives.  I could only restore, start and format.

 

If I understand your explanation of clearing correctly, then there are no long term consequences from not clearing?  It is only a means to avoid constant parity-syncs and that particular period of vulnerability?

If I understand your explanation of clearing correctly, then there are no long term consequences from not clearing?  It is only a means to avoid constant parity-syncs and that particular period of vulnerability?

Correct.

Thanks!