Quoting Tim Retout (2017-04-18 11:00:54) > On 17 April 2017 at 18:59, Jonas Smedegaard <[email protected]> wrote: >> I don't recall, but believe I installed using plain standard >> debian-installer with the "netboot" image. But possibly I might have >> used multistrap, which means the order of installing the various >> packages may be different from (c)debootrap-based installs. >> Something that is *not* a bug in multistrap - Debian Policy mandates >> package relations but no specific install order. > > Thanks - I'll break out some real armhf hardware later. So far I've > mainly been testing by reinstalling rather than upgrading > systemd-sysv.
Let me repeat: I do *not* think this issue is specific to armhf.
I *do* think this (sub)issue is tied to _upgrading_ systemd-sysv.
> I don't *think* multistrap install order should be an issue - I've
> even tried uninstalling systemd-sysv in a chroot (so that /sbin/halt
> doesn't exist), installing molly-guard, then installing systemd-sysv
> again - but this seemed to be handled fine.
Possibly this (sub)issue is tied to this:
1) install both packages _without_ executing postinst
while *not* running systemd (e.g. in a chroot)
2) execute postinst of each package (in varying order)
while *not* running systemd (e.g. in a chroot)
3) run systemd (e.g. boot the system on real hardware)
4) update systemd-sysv
I believe 1) through 2) mimicks multistrap, and since various packages
have invented their own probing mechanisms for whether being on real
hardware or not that might be relevant for triggering this (sub)issue.
> If you used multistrap, does this mean there was another package repo
> involved?
No. I use multistrap for each separation of unpacking and postinst
execution - since on slow archs that gives a speedup by doing the former
on a different more powerful host and the latter either on emulated
hardware (slow) or target hardware (cumbersome esp. when creating
several similar setups).
> The bit I can't get my head round is why dpkg-divert would not have
> reported molly-guard having any diversions in place (which seems to
> match the error message "also in package molly-guard 0.6.4"). For any
> recently-installed stretch system, molly-guard should definitely have
> been using diversions.
Right: 0.6.4 entered Stretch only 3 days prior to my bugreport, but that
bit of the log message indicates no earlier version was involved.
- Jonas
--
* Jonas Smedegaard - idealist & Internet-arkitekt
* Tlf.: +45 40843136 Website: http://dr.jones.dk/
[x] quote me freely [ ] ask before reusing [ ] keep private
signature.asc
Description: signature
