I'm looking at finishing the Mock back-end enough to make it work.

Implementing healthCheck() is enough for the chroot init to actually start working. However, this reveals a bigger issue - the lowest common functional unit for chroot-plugin is a tarball (since it's modelled on pbuilder, which uses self-contained tarballs). However, Mock doesn 't use or produce tarballs, it produces directory structures.

I'm not certain whether it's better to wrap up Mock's output in tar, or alter chroot-plugin to operate on directories at a time - my suspicion is the former is the better route, as a sudo exception for tar is less bad than the security hell of trying to copy entire root filesystems full of device nodes around.

Both issues might be mitigated if the chroot could be initialised in its final destination directly, i.e. avoid the create-generic-chroot-and-copy-to-job workflow for mock.