After creating a volume with the default inode ratio, then nearly filling the inode table, I found that a backup and restore cycle failed (see eg, JobID: 1102401). It appears that the logic for deciding the size of the restored volume does not take into account the inode count at all.

This also affects Gentoo volumes, as I'm sure you know, but can be triggered even from images without a modified inode ratio, thanks to the size reduction logic.

For this reason, and to help reduce fragmentation on restore, it would be helpful if the size of the restore volume could be specified on restore, to override the automatic selection.

We actually have a ticket open for this already, it just hasn't been implemented yet.

-James