Pavel Raiskup

> This should solve some issue when some platform needs some different defaults. This looks like the host-*.cfg is to some extent alternative to `site-defaults`, and thus it looks like...

Thank you for the report. Rather than implementing the options one by one, I would probably prefer something like `--dnf-options '--repo XYZ'`. Would that work for you?

This is a pretty complicated problem if I interpret it correctly. At this moment, we have very problematic design of executing nspawn'ed package manager in bootstrap "on demand". The bootstrap...

The locking mechanism with "readers/writers like" behavior though really look like these things should work. It is there just to protect against accidental concurrent initialization. I mean, there's IMO no...

> I mean, there's IMO no sane reason to block users from doing this, And not even saying the fact that having this working is extremely useful when debugging in...

https://github.com/rpm-software-management/mock/blob/9bf3cb57e4a14cdce4fb85d989171d5d48d5a6ef/mock/py/mockbuild/config.py#L521-L523 "cleanup_on_success=False" is enforced :-( so the workaround doesn't work

Interesting thoughts. Perhaps we could, no matter if --uniqueext is set, re-use the same cached bootstrap directory (per `config_opts['root']`). That would reduce the space consumption a lot.

Yes, we don't know how to run systemd-nspawn in container, yet. See pull request #337 which should lower the pain. Otherwise this is known problem.

Since this is about making systemd-nspawn work in docker, I'm flagging it as RFE.

Can you please retest against mock v1.4.20?