I looked everywhere but could not find an answer to this question. This would be the same as "What do I do if 2 disks (or more) fail in the array at the same time?" I have a 16 disk array (14 data, 1 parity, 1 cache) and I want to replace 2 of the drives. The 2 drives are completely empty and are not assigned to any user shares. I tried physically connecting the 2 new drives and booting up, but the array will not start because I have too many "bad" disks. I want to tell the array "go ahead and start up anyway, without parity protection, and let me access the rest of the drives in the array. Format the 2 new drives, make them part of the array, and then build parity from scratch."

 

I know that I could replace one drive, wait 48 hours for parity to rebuild an EMPTY drive, and then repeat the procedure for the second drive, but that seems a bit much just to replace 2 EMPTY drives.

 

There must be a way of doing this or else the entire array would be wiped out if 2 or more drives were to fail at the same time, and I can't believe that this is the case since there is no striping taking place. Can someone please point me to the place where the correct procedure is outlined?

You just need to request the server initialize a new disk configuration.  (This also immediately invalidates parity based on the prior disk configuration)

 

1. Stop the array

2.  Un-assign the two disks you will remove on the Devices page.

3.  Log in via telnet or on the system console (with the array still stopped) and type

initconfig

Answer "Yes" to its prompt.  (Capital "Y" and lower case "es")

 

4. Then refresh the web-browser.  All the indicators will be blue.  When you next start the array a full initial parity calculation will be performed.

5. Start the array. ( a full parity calc will begin.  Let it complete)

6. Once it is complete also perform a parity "check" to make sure the parity you just wrote can be read back

 

Joe L.

I looked everywhere but could not find an answer to this question. This would be the same as "What do I do if 2 disks (or more) fail in the array at the same time?" I have a 16 disk array (14 data, 1 parity, 1 cache) and I want to replace 2 of the drives. The 2 drives are completely empty and are not assigned to any user shares. I tried physically connecting the 2 new drives and booting up, but the array will not start because I have too many "bad" disks. I want to tell the array "go ahead and start up anyway, without parity protection, and let me access the rest of the drives in the array. Format the 2 new drives, make them part of the array, and then build parity from scratch."

 

I know that I could replace one drive, wait 48 hours for parity to rebuild an EMPTY drive, and then repeat the procedure for the second drive, but that seems a bit much just to replace 2 EMPTY drives.

 

There must be a way of doing this or else the entire array would be wiped out if 2 or more drives were to fail at the same time, and I can't believe that this is the case since there is no striping taking place. Can someone please point me to the place where the correct procedure is outlined?

 

1.  Please read this post.  After reading it should be clear why you can't rebuild 2 drives at the same time.  With 2 "missing bits", parity will not be sufficient to determine the value of each.  There is discussion of adding a 2nd parity drive (technically inaccurate description but you get the idea) that would allow handling of 2 drive failures.  This would provide the ability to replace 2 drives at the same time.  But for today, you can only rebuild one at a time.

 

2.  Remember that even if, to you, a drive is empty, unRAID continues to see the 1s and 0s on the disk, and all those 1s and 0s affect parity.  The only exception is if you litterally zero every bit on a drive.  If you re-read that parity post and consider the case where one drive is ALWAYS zero, you will see that a zeroed disk is "invisible" in a sense (in other words, parity is correct with that drive present and with that drive removed).  This creates an opportunity to zero and drive (or even multiple drives), remove those drives, and then use a trick to reconfigure the array to exclude the zero'ed disks.  There are pros and cons to this method, but here is a link to a post on removing a disk from an array.

 

3.  The third method is the easiest and most straightforward.  The only drawback is that if you have a real disk failure in the process you could lose data.  The idea would be to remove the old disks, add the new disks, reiniitialize the the array configuration (run "initiconfig" or press "restore" button on older unRAID versions), and then start the array.  unRAID will rebuild parity.

 

You might want to look at this thread because it is a similar question I posted to the form.  Might help you better understand the pros and cons on different options, as well as tweaks to provide ability to have at least a chance to recover should an unrelated drive fail in the middle.  Remember, though, whatever you do, run a full parity check soon (a day or 2) before begninning.  Take a backup of your config directory (with array stopped), just before beginning.  And preclear any new disks before putting them in your array.

Thanks, guys! That was exactly what I was looking for. I thought that "initconfig" might be what I was looking to run, but I was not sure from the threads that I had read. I definitely want to rebuild parity from scratch anyway, as I think that the parity information for my failed disk2 was bad anyway.

 

Thanks again!