2 minutes delay at poweroff with systemd-247

After upgrading to systemd-247, poweroff reprodicably experiences a 2 minute delay. It seems relative to gnome-session-restart-dbus.service.

The issue is first reported by Debian:

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=975400

And I can reproduce this issue with my Linux From Scratch build.

The relevant lines in journalctl (copied from Debian bug report because my journal was broken when I forcefully restarted the system):

Nov 17 01:05:38 nuc8 systemd[1]: user@1000.service: State 'stop-sigterm'
timed out. Killing.
Nov 17 01:05:38 nuc8 systemd[1]: user@1000.service: Killing process 1141
(systemd) with signal SIGKILL.
Nov 17 01:05:38 nuc8 systemd[1]: user@1000.service: Killing process 1286
(ssh-agent) with signal SIGKILL.
Nov 17 01:05:38 nuc8 systemd[1]: user@1000.service: Killing process
163646 (dbus-daemon) with signal SIGKILL.
Nov 17 01:05:38 nuc8 systemd[1]: user@1000.service: Main process exited,
code=killed, status=9/KILL
Nov 17 01:05:38 nuc8 systemd[1]: user@1000.service: Failed with result
'timeout'.

And

Nov 17 01:03:38 nuc8 systemd[1141]: Requested transaction contradicts existing jobs: Transaction for exit.target/start is destructive (gnome-session-restart-dbus.service has 'start' job queued, but 'stop' is included in transaction).

This is very reproducible on my system, but it seems a timing issue and may be not reproducible on other systems. I'm currently using the workaround suggested by Debian bug report (commenting out two lines about gnome-session-restart-dbus.service in gnome-session-shutdown.target).

I'm not sure if it's a gnome-session bug or a systemd bug. If you guys believe this is for systemd I'll report to them.