#1142978 systemd: daemon-reload causes ABRT or mark newly started service as dead

Package:
systemd
Source:
systemd
Description:
system and service manager
Submitter:
Georg Müller
Date:
2026-07-29 11:33:02 UTC
Severity:
normal
#1142978#5
Date:
2026-07-29 09:21:13 UTC
From:
To:
Dear Maintainer,

I ran into a problem by calling 'systemctl start ssh.service; systemctl enable ssh.service'.

My observation was that the ssh.service was in state 'inactive (dead)' even though
it was running. systemctl status still showed the correct MainPID and the process
in the cgroup.

This is reproducible (always resetted to a snapshot of a VM) in about 50% of my tries.

I once even ran into an ABRT in src/core/service.c service_sigchld_event() state switch,
pointing to an invalid state (default: assert_not_reached()), I unfortunately have
no backtrace of this.

The 'systemctl enable' command triggers a daemon-reload.

I think this is related to upstream bug https://github.com/systemd/systemd/pull/39703 .

ssh.service has alias sshd.service, which might be the cause of the problem.

This fix is backported to the upstream v258-stable branch and newer, but not to v257,
as this is probably too old.

I would like to have this patch added to the debian patches to fix the issue.
#1142978#10
Date:
2026-07-29 11:30:12 UTC
From:
To:
To assure that the systemd alias bug (fixed with https://github.com/systemd/systemd/pull/39703)
might be the one in charge, I checked by removing the "Alias=" line from ssh.service and
repeated the test (reset VM, start VM, send commands) and could not trigger the problem.

Am 29.07.26 um 11:21 schrieb Georg Müller: