machined thinks all machines terminate on "systemctl daemon-reload"

[aburch]

This applies at least to machines started via systemd-nspawn / machinectl start.

Executing systemctl daemon-reload results in systemd-machined believing the machine terminated, even though it's still running:

Jun 25 18:10:08 X systemd-machined[18606]: Machine X terminated.

I guess this happens due to the UnitRemoved signals emitted by systemd. match_unit_removed in src/machine/machined-dbus.c then releases the machine even though the unit comes back a bit later when a UnitNew signal is emitted.

I use systemd/220-7 from Debian's unstable suite. There's also a bug already filed in Debian: https://bugs.debian.org/788269

[aburch]

http://lists.freedesktop.org/archives/systemd-devel/2015-June/033357.html looks like the same issue and refers to 9b420b3 as the commit which might have introduced this as a regression.

[poettering]

There's a fix for this now in #898.