User is supposed to *UNMONITOR* LVs before quiting dmeventd. 'dmveventd' is protected against signal - so processed commands which *MAY* freeze your system (i.e. suspend devices) aren't leaving system in unusable state. So I'm not sure what '--force reboot' is doing - but it should run 'vgchange --monitor n' first. This is not seen as bug in dmeventd - rather some logical problem with reboot sequence.
Hi Zdenek, From your comment, I understand it's really necessary to go through the normal shutdown sequence, so I think there is nothing to do on this side. Specifically here we need lvm2-monitor.service to stop, which will lead to stopping dmeventd.service as well. This is what I see on normal shutdown: -------- 8< ---------------- 8< ---------------- 8< ---------------- 8< -------- Jun 28 10:04:36.507032 vm-thin8 systemd[1]: Stopped Monitoring of LVM2 mirrors, snapshots etc. using dmeventd or progress polling. ... Jun 28 10:04:36.514671 vm-thin8 dmeventd[873]: dmeventd detected break while being idle for 0 second(s), exiting. Jun 28 10:04:36.514675 vm-thin8 dmeventd[873]: dmeventd shutting down. -------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------