cc @tkopecek @ngompa

See commit message for some background.

Makes sense to me - we can squeze it to 1.31 so we don't even introduce --bind-dev @ngompa?

Wait, this doesn't make sense. Why doesn't Mock's nspawn mode expose device-mapper? We need it for building images with livemedia-creator too... @praiskup @msuchy

I'd rather just fix Mock so that we expose the device-mapper device like we do for btrfs-control

PR submitted to Mock for this: https://github.com/rpm-software-management/mock/pull/1005

@ngompa @ignatenkobrain so, do we now think (after mock merged Neals' code) that we should revert #3568 ?

@ngompa @ignatenkobrain so, do we now think (after mock merged Neals' code) that we should revert #3568 ?

Yes.

Reverted in https://pagure.io/koji/c/05851ac0c91f444dd467bd636138d09b1d4cddb7

@tkopecek I think it is still wrong…

1. Mounting /dev by koji breaks nspawn so if anything, it has to be removed
2. The code in mock does not make kiwi builds working anyway =(

Makes sense. I've tested old/new chroot vs. non/dm image and it needs this patch + mock's patch. Then I was able to build all four of them.

Commit de2df168 fixes this pull-request