Logical Volume Manger and Restoring

Hi, If I had an LVM of two 12TB drives for a total of 24TB, and one of those drives fails, what’s the best way to restore. Lets just say the the drives were completely full when one of them failed. I don’t want to have to restore 24TB, 12TB is already so much to restore from the cloud. Basically is there way for Duplicati to know which files come from a certain drive within an LVM, so I only have to restore the files that have been lost to the drive failure.

If possible I’d like to avoid having to restore all files.

Maybe I’m thinking about this all the wrong way, not experienced at all with this sort of stuff.

Welcome to the forum @dillonbrannick

Duplicati is file-based. It uses files like any normal application, so data placement is up to OS+FS.

I don’t use LVM, but Internet posts seem to say that the volume breaks if one of its drives breaks…
Possibly an LVM expert will stop by, but it feels to me like you’re also increasing the failure chance.
Fedora and LVM is an old commentary on the issue, and it links to another article discussing LVM.

This would be a single filesystem of some flavor, on a single volume on two physical drives, right?
As an alternative, perhaps you could design mounts to control what paths physically reside where.
Symbolic links might also work, but make sure you have a look at Duplicati symlink-policy options.

Hi, thanks, yeah I think you’re right about the LVM. At least it seems it’s a high chance that you would lose all data from every drive in it, which would make it useless for my use case.

I’ve since dropped the idea of an LVM and decided to use MergerFS to create a pool while keeping the seperate drives intact so it’s easy to point duplicati to an individuual drive with the pool rather than an array of drives. Although currently I only have one drive in the pool so it’s not fully tested, but it seems like it should work as intended.


1 Like