Jump to content

Removing a hard drive from the server


nukhem

Recommended Posts

remove data from drive while still part of the array

stop array

unassign drive from devices page

start array..not really needed but I like to check before shutting down  ;D

stop array and shutdown

remove hardware

That will NOT remove the drive from the array.  It will treat it exactly the same as if the drive had failed.  You will still be able to both read-and write to the un-assigned disk AND you are not protected from any other disk failure. Running in this mode will just use parity when writing to the un-assigned disk, and will re-create any contents from parity AND the other disks when reading from the un-assigned disk.

 

To remove a drive and not replace it you must perform one last step AFTER un-assigning it and BEFORE re-starting the array.

You must press the button labeled "Restore" (you will need to check the box under it to enable it)

The button is very poorly labeled, as it really is a "Set initial disk configuration" button.  It immediately invalidates parity.  You will not be parity protected again until you re-start the array and do a full parity calculation with the removed drive un-assigned. 

 

You can disconnect the old drive physically any time after you un-assign it, as long as you do it with the power removed.  unRAID is NOT hot-pluggable.

Link to comment

Option #1 - Parity Exposure but Maintains Disk Contents

 

 

 

Removing a disk from the array causes the need to rebuild parity, so this creates some exposure for a disk failure in the process.  Running the parity check at the beginning helps to minimize the chances of a problem.  The contents of the removed disk are preserved.

 

 

 

1 - Perform a full parity check

 

2 - Stop the array

 

3 - Go to the devices page and remove the disk (rearrange disks if desired)

 

4 - Go to the main page and press the "Restore" button (usually a bad thing to do but needed to remove a disk from the array)

 

(The circles next to all the drives will be blue)

 

5 - Start the array - this will cause parity to be BUILT

 

6 - Perform a full parity check (whenever you build parity you should check parity immediately afterwards)

 

Option #2 - Maintain Parity but Removed Disk Contents are Lost

 

This technique allows you to remove the disk without breaking parity.  It requires filling the disk to be removed with binary zeros.  This updates parity at the same time.  If you understand parity, you'll realize that a disk full of binary zeros makes it invisible from a parity perspective.  So once that is complete, you can remove the disk, and use the "set invalidslot" feature to reconfigure the array, and parity will be accurate. 

 

I wrote a detailed post including screenshots on this process, but cannot look it up in my current environment.  Maybe someone would be kind enough to post a link here.  Note that there was an optimization to allow the disk to be filled with zeros faster using a larger buffer later in the thread.  One negative side effect - because you have to fill up your disk with zeros, its contents are lost.

Link to comment

Option #2 - Maintain Parity but Removed Disk Contents are Lost

 

I wrote a detailed post including screenshots on this process, but cannot look it up in my current environment.  Maybe someone would be kind enough to post a link here.  Note that there was an optimization to allow the disk to be filled with zeros faster using a larger buffer later in the thread.  One negative side effect - because you have to fill up your disk with zeros, its contents are lost.

It is here: http://lime-technology.com/forum/index.php?topic=2591.msg20919#msg20919

Link to comment

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...