What to restart to get rid of unresponsive notifications after a Plasma crash?

A couple of years ago, I got myself a short script to restart Plasma:

kbuildsycoca5 && kquitapp5 plasmashell && kstart5 plasmashell

I also use kwin_x11 --replace to restart kwin specifically. More recently, I learned systemctl --user restart plasma-plasmashell.service.

However, none of these seem to help with the most recent issue that I’m having. After waking a laptop from sleep, several notifications were shown. Then Plasma or Kwin or whatever seems to have crashed, and not fully recovered. This caused notifications to get stuck – I cannot interact with the notifications, yet they are rendered above all other windows (including if checkbox “Keep above others” is enabled).

Running any of the three commands didn’t help. systemctl complained:

Warning: The unit file, source configuration file or drop-ins of plasma-plasmashell.service changed on disk. Run 'systemctl --user daemon-reload' to reload units.

running the suggested command didn’t help much, systemctl --user restart plasma-plasmashell.service still didn’t get rid of the notifications. Here’s the status output, if that helps:

$ systemctl --user status plasma-plasmashell.service
● plasma-plasmashell.service - KDE Plasma Workspace
     Loaded: loaded (/usr/lib/systemd/user/plasma-plasmashell.service; enabled; preset: enabled)
     Active: deactivating (stop-sigterm) (Result: timeout) since Sun 2026-05-17 HH:MM:SS TZ; 5 days ago
   Main PID: 391496 (plasmashell)
      Tasks: 5 (limit: 38235)
     Memory: 20.7M (peak: 22.9M)
        CPU: 231ms
     CGroup: /user.slice/user-1000.slice/user@1000.service/session.slice/plasma-plasmashell.service
             └─391496 /usr/bin/plasmashell --no-respawn

May 23 HH:MM:SS computername systemd[4241]: plasma-plasmashell.service: Scheduled restart job, restart counter is at 7.
May 23 HH:MM:SS computername systemd[4241]: Starting plasma-plasmashell.service - KDE Plasma Workspace...
May 23 HH:MM:SS computername systemd[4241]: plasma-plasmashell.service: start operation timed out. Terminating.

I can always just sudo reboot now, but could you please let me know: What else can I try to restart Plasma/Kwin/whatever other part of the stack I don’t know about?

What’s the most recent, definitely up-to-date, way of restarting Plasma in KDE 6 from command line?


System information

Operating System: KDE neon User Edition
KDE Plasma Version: 6.6.5
KDE Frameworks Version: 6.26.0
Qt Version: 6.10.2
Kernel Version: 6.17.0-23-generic (64-bit)
Graphics Platform: X11
Processors: 8 × Intel® Core™ i5-8250U CPU @ 1.60GHz
Memory: 32 GiB of RAM (31.2 GiB usable)
Graphics Processor 1: Intel® UHD Graphics 620
Graphics Processor 2: AMD Radeon R5 M465 Series
Manufacturer: Dell Inc.
Product Name: Inspiron 5570

There is the ungraceful option. From a bash shell, Reload systemd:

systemctl --user daemon-reload

forcefully kill the frozen notification daemon, Plasma workspace, and KWin to break the lock on the UI:

pkill -9 kwin_x11

pkill -9 plasmashell

pkill -9 kded6

Once killed, they should restart automatically. If they don’t, manually trigger a fresh session with:

systemctl --user restart plasma-plasmashell.service

But all of that doesn’t get to the actual cause of your problem. Maybe a graphics driver is causing problems. Check the journal logs for clues:

journalctl --user -b -p err

I did a search myself for the problem you are experiencing and found this:

"Stuck notifications during sleep/wake cycles often happen when the compositor abruptly drops out due to GPU power management faults or D-Bus timeouts. "

One suggestion was to turn on “Do not disturb” in your notification settings before letting machine sleep, but once again, that doesn’t really get to the underlying cause.