fra81

[fra81]

@LittleWoods, what is your KDE Plasma version?

[fra81]

Hi,

did you also check the content of the /usr/libexec/plasma-dbus-run-session-if-needed script?

Besides, you can try the following steps:

1. Create the /etc/xdg/plasma-workspace/env directory

2. Place a nx-surceenv.sh script in it containing only this line: export LD_PRELOAD=/usr/NX/lib/libnxegl.so

3. Reboot

4. Make sure the libnxegl library is preloaded (the output of the sudo netstat -anlp | grep egl command shouldn’t be empty)

 

Let us know the results 😉

[fra81]

@jjgalvez,

please show the content of the ‘/usr/libexec/plasma-dbus-run-session-if-needed’ script. Note that this file could be in a different path, so you can use the ‘locate’ command to find it.

Also show the output of:

ls -l /etc/xdg/plasma-workspace/env

[fra81]

Hi,

do you mean that the 1280×1024 monitor is “physically” attached to the Mac mini and turned on? Or do you mean that the 1280×1024 monitor is it what you see in the system settings but no real monitor is actually plugged in?

To investigate further we would need to see the logs. You can gather both server and client side logs by following the instructions in https://kb.nomachine.com/DT07S00243. You can send them to forum[at]nomachine[dot]com, by referencing this topic.

[fra81]

Hi,

for maximum image quality, besides setting the slider all the way up, I would check ‘Disable network-adaptive’ but leave ‘Disable multi-pass’ unchecked. I’d also check ‘frame buffering’ and ‘post-processing’. Hardware decoding shouldn’t be relevant for quality. For this scenario, you may also want to disable hardware encoding on the server, since hardware encoders are surely faster but they could encode with lowest quality compared to the software encoder, even when using the same parameters.

For video playback, I’d choose a lower quality in the slider (medium quality should be a good trade off), uncheck ‘network-adaptive’ and ‘multi-pass encoding’, check ‘frame-buffering’ and ‘post-processing’. Hardware decoding is usually faster.

 

[fra81]

Hi,

we’re trying to reproduce the encoder issue with the new drivers, but it seems you also have a different problem. In fact, the fact hardware encoding fails isn’t enough to explain the black screen. Software encoding should be used as a fallback. Can you check, on the machine where you have the black screen, if unchecking ‘Use hardware encoding’ (Server settings -> Performance panel) solves the black screen problem?

Can you also send server side logs gathered while reproducing the black screen? For gathering the logs you can follow the instructions in https://kb.nomachine.com/DT07S00243 and send them to forum[at]nomachine[dot]com.

[fra81]

Sorry for the delay. I checked the logs and I see there is no difference in the scroll events before and after the connection, so we can exclude that NoMachine leaves the input events system in an inconsistent state. Even the modifiers state is exactly the same. I don’t know why those two applications stop handling scroll events, but it doesn’t seem anything that the NoMachine does wrong. One last thing you could try is comparing the outputs of the ‘xev -q’ command before and after the connection, in order to see if any of the X server settings has changed in the meanwhile.

[fra81]

Hi,

sorry for the delay. Disabling CapsLock by XKB settings on the Linux client will simply disable it as a modifier, but the NoMachine player will still receive it and forward to the Mac. Furthermore, when you disable CapsLock on the Mac, this will only affect the physical keyboard of the Mac, but it won’t have any effect on the remote input events.

What I’d suggest you is to disable the CapsLock key completely on the Linux machine, by using this command:

xmodmap -e 'keycode 66=NoSymbol'

[fra81]

Hi OldMan,

can you try to run the ‘xev’ utility in the remote desktop, send the Ctrl key to it through the NoMachine remote session and show us what is reported in the output? I would start with sending any other key first, just to check if everything is set up correctly and the keys are reported in the output. Then you can type just the Ctrl key, to see if it is sent through. Finally, you can try with the problematic Ctrl-<key> combinations and show us the xev output.

[fra81]

Hi,

is the server headless, or is its monitor turned off? Do you experience the same lag when the monitor is attached and turned on?

This doesn’t happen when connected to a physical display.

Please clarify that. Do you mean that it doesn’t happen when you’re sitting in front of the server machine and using the local mouse?

Some tips for headless machines are also available here https://kb.nomachine.com/AR03P00973

[fra81]

I see that there are no X events reported by xev even before the connection. You have to keep the cursor on the xev window to make them appear. Can you try again in that way? No need to send the whole xev output, it would be enough to show just the scroll events, before and after the connection.

[fra81]

Please run the ‘xev’ utility on the server and show us its output when scrolling. You should see a few ButtonPress/ButtonRelease events (of button 4 and 5), and these are the ones we’re interested in. Do this on the server before the connection (when scrolling works in all the apps) and then again after the connection (when you experience the problem).

[fra81]

Hi,

we don’t reproduce in our lab with version 8.9.1. Can you make sure you have rebooted the system after updating NoMachine? Can you also confirm what is the distribution you are using on the server?

If that doesn’t help, please show us the output of the following commands:

sudo netstat -anlp | grep egl

sudo getcap /usr/bin/kwin_wayland

Please also check if any of these files exist:

/usr/lib/plasma-sourceenv.sh
/usr/lib/plasma-dbus-run-session-if-needed

[fra81]

Hi,

it looks like a problem with the Nvidia drivers, so I would try to reinstall them. Alternatively, as a workaround, you could rename temporarily the libcuda.so library, so that it won’t be loaded and the installation can complete.

[fra81]

Hi,

it looks like the NumLock is off on the Ubuntu server’s keyboard, and the keypad on the Mac works as expected in this regard. You should be able to turn it on remotely by pressing the “Clear” key on the keypad of the Mac.

[Author]

Posts