My hardware has some fault, that every few months it has an event of multiple errors on multiple disks,

followed by multiple disks dropped from the array, and it turns out that the fs on these disks (physical + their array copy) is corrupted,

and then my data is at the mercy of xfs-repair.
I suspect my on-board HBA (the only HBA on my server) has a warming issue, so I try not to overload it for now.

 

I cannot build another backup server at this time, but I have some disks to backup my array data.

I am looking for a simple, reliable, automated (scheduled) solution to backup (actually file-sync) my Unraid array to unassigned devices,
I was offered to create a logical volume from the unassigned devices using Linux LVM, than rsync the array to it.

 

Is it theoretically possible to do? Any thoughts / suggestions / other solutions before I dive into learning Linux LVM and rsync parameters?

 

Another issue is that rsync might be destructive if one of the source disks and it's array copy have fs corruption,

and the automated rsync would delete this disk content from the backup logical volume.

So maybe run a dry run of rsync and some (automated) examination of the expected results before running the actual sync?

This solution is getting more and more complicated.

Edited May 30, 20197 yr by Gico

24 minutes ago, Gico said:

Any thoughts / suggestions / other solutions before I dive into learning Linux LVM and rsync parameters?

One option would be creating a btrfs pool of unassigned devices, with or without redundancy, then rsync from the array, you could do a snapshot before each sync in case rsync deletes something it wasn't supposed to.