KDE Plasma takes very long to load after login in but only on Wayland. How to troubleshoot?

I have an EndeavourOS installation which is about 2 years old (I tinkered a lot with it but it works without issues).

I decided to install plasma-wayland-session and select the Wayland option via SDDM, after login in the following happens:

  1. It takes about 10 seconds to load the desktop wallpaper
  2. It takes additonall about 40 seconds to load the taskbar + desktop icons (I can’t click anything yet!)
  3. It takes about 30 seconds more until I can click on anything.

Sometimes between step 2-3 the whole screen turns black/white-ish.

So overall it takes me about 80 seconds to load my desktop and to be able to interact with it.
Compared on X11 it takes about 3 seconds and everything is loaded.


I am using a full AMD system: Ryzen 3600 + RX 6700 XT + NVME SSD…

Where would I start troubleshooting this issue? Or is it normal to have ~27x longer loading times on Wayland?

Thanks for the help.

1 Like

Is this on bare metal or in a VM?

bare metal

My autostart looks like this, if this helps:
image

That usually means that something hangs or crashes on the plasma startup and the system waits until it eventually times out and proceeds.

I would start troubleshooting with:

  • Create a completely new user (eliminate side effects from existing autostart, configs, caches, …) and check if it works
  • Check journalctl if there are error messages

I am now encountering this same issue/behavior on my EndeavourOS host, using SDDM and plasma-wayland. SDDM is a black screen with my mouse that takes a while to load, then after signing into EndeavourOS using a plasma-wayland session it’s just my wallpaper and my mouse for another couple of minutes before the desktop environment actually loads.

Here are my logs from running journalctl:

Any ideas/suggestions?

This won’t be of any help for you:

My issue just went way.

After a KDE / system update some months ago the time to login got reduced to 2-3 seconds.

I guess the problem was a hidden bug that simply got fixed.


As suggested, I tried looking through journalctl / other logs but I could never figure out what caused the very long loading times to begin with.