On Thu, Dec 01, 2016 at 10:43:18AM +0100, Thomas Goirand wrote: >On 11/30/2016 08:35 PM, Sam Hartman wrote: >> Do the openstack images use debootstrap as a package resolver? > >Yes. > >> If so, yeah, that broke for most nontrivial installations in stretch. > >After some tests, it is libpam-systemd that is pulling systemd-shim. So >I wonder 2 things: > >1/ Do we really need libpam-systemd?
I honestly don't know, tbh. >2/ If we do, what is the way to get systemd-sysv installed *without* >systemd-shim? libpam-systemd has as dependency: systemd-shim (>= 8-2) | >systemd-sysv. How come it's not the opposite way around, and the -shim >version has priority? It was done that way explicitly so that systemd-shim would solve the dependency for people without systemd. I've no idea what's actually broken here, I'm just seeing the (limited) installation failure messages from debootstrap. >3/ If systemd-shim and systemd-sysv are conflicting, how come this isn't >declared in the packages? > >Finally, does anyone have a trick to fix this issue? > >I'll upload a version without installing libpam-systemd at the >debootstrap stage, and install it *after* the debootstrap phase. >Hopefully, this will be a good fix. I'd suggest maybe trying doing a stretch debootstrap locally using the same set of packages as the build script first, and see what's *Actually* breaking... -- Steve McIntyre, Cambridge, UK. st...@einval.com The two hard things in computing: * naming things * cache invalidation * off-by-one errors -- Stig Sandbeck Mathisen