I had an issue where a drive was corrupted, and the xfs_repair tool was unable to find any suitable superblocks.

 

Now I'd assume that a "good" disk would return something about good superblocks found and good magic numbers when you initially run it (xfs_repair -n /dev/sdX).  However all of mine are returning bad superblocks and bad magic numbers.

 

My question is two-fold:

1.  Does it matter?

2.  Is there some routine maintenance I should be running?

Nevermind I'm an idiot

29 minutes ago, whoopn said:

Nevermind I'm an idiot

 

Don't be so hard on yourself. Was it the -n flag that was tripped you up? In case anyone else stumbles across this thread with a similar problem, the -n flag means no-modify, so to actually make the repair you have to run the command without it. The other thing I would say is that /dev/sdX if the identifier for the actual disk. In a parity protected array you should be using /dev/mdX in order to maintain parity.

xfs_repair /dev/mdX

 

35 minutes ago, John_M said:

The other thing I would say is that /dev/sdX if the identifier for the actual disk. In a parity protected array you should be using /dev/mdX in order to maintain parity.

 

And if you did repair the sdX device and it was part of the array, then you have invalidated parity so you will have to resync.

1 hour ago, whoopn said:

However all of mine are returning bad superblocks and bad magic numbers.

The above sentence seems to indicate that the tool was run on disks that aren't using XFS.

2 hours ago, whoopn said:

I had an issue where a drive was corrupted, and the xfs_repair tool was unable to find any suitable superblocks.

 

Now I'd assume that a "good" disk would return something about good superblocks found and good magic numbers when you initially run it (xfs_repair -n /dev/sdX).  However all of mine are returning bad superblocks and bad magic numbers.

 

My question is two-fold:

1.  Does it matter?

2.  Is there some routine maintenance I should be running?

If you use the /dev/sdX identifiers then you need to also include the partition number (i.e. /dev/sdX1).    However as was mentioned this invalidates parity so you would normally put the array into Maintenance mode and use /dev/mdX where X is the disk number.   Using the /dev/mdX names means that you do not need to include the partition and it maintains valid parity.