Forum Replies Created
-
AuthorPosts
-
graywolfParticipant
Hello, in log I see you connected to share the physical display.
So do you have a remote display with a French layout and when connecting by NoMachine it switches to US? Did I understand it correctly?
May I ask you:
1. Does the problem occur in the login screen or in the desktop environment?
2. Which is the desktop in use (Gnome, KDE, Xfce, …)?
3. Which is the display manager (GDM, KDM, SDDM, LightDM, …)?
4. The precise version of Ubuntu in use?graywolfParticipantHello, I think you should rename the file to upload. File names having that suffix are not allowed.
graywolfParticipantAre you using NoMachine version 6.3.6 on both sides, client and server? Would you update to the more recent 6.5.6?
Are you using NoMachine to create a virtual display on the server or to share an existing display?
I don’t think you need to reboot to workaround the issue. After the remote session got shown, run the NoMachine menu, look at the buttons near the bottom of the window and press the one showing the tip “Resize remote screen”. The remote session will resize adapting to the client window.
March 4, 2019 at 18:23 in reply to: Non monochrome fonts in Qt applications are fuzzy and badly anti-aliased #21655graywolfParticipantSorry, my mistake. Correct form is:
ProxyExtraOptions "pack=16m-png"
March 1, 2019 at 16:58 in reply to: Non monochrome fonts in Qt applications are fuzzy and badly anti-aliased #21629graywolfParticipantWe have reproduced the asymettric halo and already working on fixing that. Thank you for reporting.
By the way, using PNG encoding would grant lossless compression without artifacts (with a little increase in bandwidth usage).
Please double check the value in ProxyExtraOptions:
ProxyExtraOptions "16m-png"
terminate all the running session, restart NoMachine to be sure:
/etc/NX/nxserver --restart
Then create a new session and check in session log whether PNG is actually used:
grep -H "Using pack method" /usr/NX/var/log/node/*C-*/session
graywolfParticipantThe Xorg log shows a video out has been staying connected for a while. It is possible the passthrough is not working as expected, although no problem is reported. For example, look at this compatibility list:
Compatibility Guide – Shared Pass-Through Graphics
It seems that hypervisors provide passthrough for a limited set of video cards.
February 27, 2019 at 14:52 in reply to: Non monochrome fonts in Qt applications are fuzzy and badly anti-aliased #21605graywolfParticipantWould you try this change in node.cfg:
DisplayServerExtraOptions "-extension Composite"
and this
DisplayServerExtraOptions "-extension MIT-SHM"
and their combination:
DisplayServerExtraOptions "-extension Composite -extension MIT-SHM"
If nothing changes, try this on the other key ProxyExtraOptions:
ProxyExtraOptions "16m-png"
February 27, 2019 at 10:39 in reply to: Black screen on connection to headless CentOS 7 NoMachine #21602graywolfParticipant# rmmod nvidia
rmmod: ERROR: Module nvidia is in use by: nvidia_modesetnvidia module depends on other modules they you need to remove before (order matters):
rmmod nvidia_drm nvidia_uvm nvidia_modeset nvidia
If it doesn’t help, use
nvidia-smi
to check processes using the GPU.graywolfParticipantLikely your video card changed to standby state if no display is attached. Search for errors in /var/log/Xorg.0.log.
Attaching a display would help to find out if the X server is correctly configured. I’d expect to see the guest X server output shown on the attached display.
Should things work fine with a display attached it could be replaced by a fake display dongle.
graywolfParticipantCheck the output of command
sudo -l
.
Search system log for message from “nx” sources:journalctl -b -0 | grep n
.
Look at NoMachine log in /usr/NX/var/log/node/C-*/session and /usr/NX/var/log/nxerror.log.graywolfParticipantI see that NoMachine launched a virtual display as its attempts to probe the X server failed. The virtual display runs in a process nxnode.bin, likely the one consuming the CPU.
By the way, I understand you want to use the video card, so falling back to the virtual display is not an option. Investigating the high CPU usage by nxnode.bin looks futile, in such a situation.
It seems that most part of the problems are due to the headless configuration. Would it be possible for you attaching a display to the output of the video card, restoring xorg.conf, restarting Gdm, restarting NoMachine, and trying to see what happens?
graywolfParticipantIs the user enabled to run sudo?
graywolfParticipant256k-png would be what you are looking for, but that 256k prefix is a legacy and actually ignored.
There is no way to change the number of colors in the current version of NoMachine. All the test we did at our labs showed that any gain in terms of reduced bandwitdh was little and not worth the quality degradation.
graywolfParticipantHello Thomas.
We are working on that. Please, register your email address to the Trouble Report so you’ll get a notification when the problem is fixed.
graywolfParticipantDo you get it under the same conditions (XFCE, virtual desktop session) ?
-
AuthorPosts