Let's say I have a 120 GB IDE drive that fails.  Can I put a larger SATA drive in its place simply by mapping the SATA drive to that slot in Devices?

 

Thanks!!

Robbie

Let's say I have a 120 GB IDE drive that fails.  Can I put a larger SATA drive in its place simply by mapping the SATA drive to that slot in Devices?

 

Thanks!!

Robbie

Yes... it is that easy....    unless the new drive is larger than the parity drive... and then there is just one more step.

 

If the new replacement drive is larger than the current parity drive, then assign the current parity drive in the old data slot, and the new larger drive in the parity slot, and the array will recognize you swapped them and first copy the old parity to the new, and then rebuild the data drive with the old contents.

Using that example, if I move the parity drive to the slot where the failed drive is, would the data from the failed drive not be lost (as there would be no parity drive at that moment)?

 

Thanks for helping me understand how unRAID really works 

Bump.  Am I not better to first replace the parity drive with a larger drive before moving the old parity drive to the failed drive's slot?

Bump.  Am I not better to first replace the parity drive with a larger drive before moving the old parity drive to the failed drive's slot?

You will not be able to do that if you have a failed disk in the array... (no way to build parity on the new drive from the failed drive in combination with the working drive) 

 

If you were to just plug in a larger parity drive, in place of the existing parity drive, while a data drive has failed, the unRAID array would not even let you start the array. It will simply say too many wrong or missing disks.  (data drive failed, and the parity drive is wrong)

 

In the "parity-swap" process, unRAID recognizes the disk serial numbers have moved and the prior parity drive is in a data drive slot, and a new serial number is in the parity slot in the array, and that a data drive is "missing"

 

It uses that special set of conditions to recognize the need to copy from the old parity drive (in the data slot) to the new parity drive (in the parity slot) and then

to use the new parity (a mirror of the old for its equivalent disk partitions) and the other still working data drives to reconstruct the data on the drive being replaced.

 

If all your drives are currently working, you can up-size the parity drive simply by replacing it and letting the array build parity on it when you press "Start"

If one of your data drives is not working, you can up-size the parity drive and reuse the old parity drive to replace the failed data drive in one operation as described in the "parity swap" procedure above.   

 

So.. to answer your question... You only need the "parity-swap" when a data disk has failed and your only possible replacement drive is bigger than the current parity drive.  Then it is the only way to handle the process...  (well... you could move the "simulated drive's" contents to disks that are still working, then un-assign the failed disk, then press "Restore" to delete the failed drive from the array and completely rebuild parity without it.  I did that once when I was re-organizing my data and disks and removing a very small disk from the array)  You are still not protected from a second disk failure while re-building parity, but once built you would be.  And if you could not get a replacement drive for several weeks, it might be a logical choice if you had enough free capacity to move things around.

 

Now... Just to let you play... if you un-assign a drive, it is exactly the same as a failure to unRAID. (At least until you either replace it with a new drive, or use the button labeled "restore" (it is actually a Initialize new configuration button) which saves the currently working and assigned disks as a new configuration, and starts a new parity build based on them.   

 

So... to play, un-assign a disk...  Start the array, you will still be able to read and write to it, as it is being simulated.  Re-assign it, and "Start" the array, the contents, including any changes made to it will be rebuilt onto it.   During the simulation and rebuild you are not protected from a second failed drive... It is that reason why replacement of a failed drive should be performed as soon as possible.

 

If you wanted to practice the parity swap process.

1. Un-assign a drive  (to simulate its failure)

2. Replace the existing parity drive with new one as big as it, or bigger  (assign the new parity drive on the "Devices" page to the parity role.)

3. Assign the old parity drive to the data drive slot you un-assigned in step 1.

4. "Start" the array.   Some number of hours later the new parity drive will have had its contents loaded from the old and the new data drive the data restored to it.  The array will be off-line as the parity drive contents are copied... I think it will be on-line for the second part of the process... (it has been a while since I did it as an exercise.. I don't remember if it was on-line while rebuilding the data disk for sure)

 

Joe L.