On Fri, Feb 03, 2017 at 10:04:48AM -0300, Felipe Sateler wrote: > On 3 February 2017 at 08:03, Carsten Brandt <deb...@cebe.cc> wrote: > > # apt-get dist-upgrade […] > > The full output can be found at the following URL: > > https://gist.github.com/cebe/67c135b990d5c7a92149e4e7113c555c
The "interesting" part for context later: | […] | Vorbereitung zum Entpacken von .../keyboard-configuration_1.158_all.deb ... | Entpacken von keyboard-configuration (1.158) über (1.123) ... | Vorbereitung zum Entpacken von .../nfs-common_1%3a1.3.4-2_amd64.deb ... | Entpacken von nfs-common (1:1.3.4-2) über (1:1.2.8-9) ... | […] | (Lese Datenbank ... 307696 Dateien und Verzeichnisse sind derzeit installiert.) | Entfernen von systemd (215-17+deb8u6) ... | […] > Dear apt maintainers, any idea on what conditions could result in apt > deciding to temporarily remove systemd? There are situations where the official stanza thenever this comes up in apt/dpkg discussions is "borderline packaging bugs". That can't really be it through as apt doesn't like to perform these temporary removes (as they are a pain in the … backside) and forbids this in cases essentials or their dependencies are involved and systemd is one of those. I haven't a huge amount of time at the moment, so feel free to pick up where I will leave – what I found so far: Downloading the status file and with an apt/jessie build (apt/stretch has various changes in the dpkg interaction, so installation plan is slightly different) run as user (see README.md file in apt sources in case you want to know what on earth this commandline does) [my testsystem is unstable; as such the repository state is different and hence the effective solution is different, but "similar"] $ apt-get dist-upgrade -o dir::state::status=./download -o Debug::pkgAcqArchive::NoQueue=1 -o Debug::pkgDpkgPm=1 -o Debug::NoLocking=1 […] /usr/bin/dpkg --status-fd 183 --unpack --auto-deconfigure /var/cache/apt/archives/hdparm_9.51+ds-1_amd64.deb /var/cache/apt/archives/netbase_5.4_all.deb /var/cache/apt/archives/kbd_2.0.3-2_amd64.deb /var/cache/apt/archives/keyboard-configuration_1.158_all.deb /var/cache/apt/archives/keyutils_1.5.9-9_amd64.deb /var/cache/apt/archives/nfs-common_1%3a1.3.4-2_amd64.deb /usr/bin/dpkg --status-fd 185 --force-depends --force-remove-essential --remove systemd-shim:amd64 […] The removal of systemd-shim is advertised and in the command before you see that nfs-common and keyboard-configuration is unpacked, just like in the full output above and the rest is in there, too… So, as strange as that sounds, it looks for me more like either apt in a real run or dpkg accidently write/read systemd where systemd-shim should be… but that sounds even more unlikely then a temp removal… I notice also a bunch of segfaults in mono stuff further above in the log, which doesn't look particularly reassuring, so perhaps that is even something as simple as dying/faulty memory? apt & dpkg have a tendency to show hardware issues as they are more demanding than most apps… If someone could replicate that system and run the update for real that would probably shine more light on the issue – unreproducible for me until then. Best regards David Kalnischkies
signature.asc
Description: PGP signature
