Yesterday I upgraded v6.3.3 to v6.3.4 and right after the upgrade restart I got errors for 16 reallocated sectors and 88 pending sectors on a 3TB drive.

Ran SMART extended self-test, and it "Completed without errors".

I stopped the array, de-assigned the drive, assigned a 6TB drive in it's slot, started the array and it is rebuilding now the 3TB drive data in the 6TB drive.

I tried to mount the 3TB drive and got "XFS (sdl1): Filesystem has duplicate UUID", and the drive was not mounted.

Is it because the data rebuilt is not over so it "remembers" the 3TB drive UUID,

or must I restart the server (after the rebuilt is over) in order for it to "forget" this drive UUID?

 

Thank you for helping.

That's normal , the rebuilt disk is going to have the same UUID as the old disk, you can't have both mounted at the same time.

Forgot to add, if you need to mount the old disk on the same server without re-formatting you can change the UUID, like so:

 

xfs_admin -U generate /dev/sdX1

 

Don't forget the 1 in the end.

Thanks! I was writing a question regarding this when got the sound for a new message in this post. 

Thanks @johnnie.black - just encountered the same error after a rebuild on a replacement disk. While I'm sure the rebuild was successful, I want to mount the old suspect drive (which appeared fine on my Ubuntu system) so I can use the Krusader 'Synchronize' function to compare and verify the files from the suspect disk with the replacement disk after the rebuild. The command given worked like expected and I'm now comparing.