VNC — Optimize Connection to the Thin Client

Symptom

~~Translation~~On ~~pending.~~some Intel hardware (e.g., BLACK.DICE, 1688, 1658), the utilization of a CPU core jumps to 100% during an active VNC connection. When dragging a window, the image updates within several s.

Workarounds

Use the Xorg-Intel driver

By default, thin clients use the Kernel Mode Setting (KMS) driver. Instead, Intel devices can also use the xorg-Intel driver.

Open the WebUI of the openthinclient Management Server Create a new hardware type or select an existing hardware type In the hardware type settings, set the "Use specific graphics driver" switch to "Use Xorg-Intel driver instead of KMS". Verify that the hardware type is assigned to the affected client; if necessary, assign the hardware type to the thin client.

Disable GLAMOR.

ℹ️ Expert Tip

The following instructions are intended for experts. Unless there's a compelling reason, use the Xorg-Intel driver as described above.

It's also possible to use the KMS driver and disable GLAMOR. Note the following warnings regarding these instructions:

It applies to all thin clients that boot from the server. Set the "Use specific graphics driver" switch to "Default" and use it as configured. Intel hardware continues to boot after these changes.

To disable GLAMOR for all thin clients, place the file <Manager-Home>/nfs/root/custom/rootfs/etc/X11/xorg.conf.d/20-modesetting.conf with the following content on the openthinclient Management Server:

Section "Device"
	Identifier "My GPU"
  	Driver "modesetting"
	Option "AccelMethod" "none"
EndSection

Then reboot the thin clients. If the changes succeed, the VNC connection should work again without the CPU load on a single core rising to 100%. Additionally, the following lines should be present in the file /var/log/Xorg.0.log :

...
[...] (**) modeset(0): Option "AccelMethod" "none"
...
[...] (**) modeset(0): glamor disabled
...