How to speed up NM server startup at boot

[dje666]

Dear Forum,

Is there anyway to speed up (prioritise) the startup of the NM server on a remote Linux machine. On my Ubuntu 22.04 PC, I am seeing that the NM server does not startup for several minutes, and if I try to connect to the Linux box during that time, NM player on my Win10 laptop will assume that the Linux NM unit is headless, and start a virtual desktop instead.

If I wait for the NM server on the Linux machine to start up, then all is good, but I’m having to guess how long to wait before initialising a connection from the Win10 laptop.

Regards,

DJE666

 

[Britgirl]

This is strange behaviour and not one that we are aware. If you send us the logs that we requested in your other topic, we can check if there is anything that indicates why the nxserver is not starting.

NoMachine defaulting to headless rather than “real” desktop

[dje666]

Hello BritGirl,

Can you confirm if the NoMachine icon that appears on my remote Linux desktop (top right), is an indication of the server being active or an indication that the Linux box is being accessed remotely?

If I access the Linux machine directly the NoMachine server GUI reports that the NM server is active, but the NM icon can take a few minutes to appear, if it appears at all.

TTFN,

dje666

[Britgirl]

The NoMachine icon represents the NoMachine service i.e that there is a daemon running and that the server is either running or temporarily stopped. If you “shut down” the server (from NoMachine settings)  or “quit the service” from the !M icon, the !M icon will disappear, and you will have to open the program from the applications menu, go in to server settings and “start the server”.

Notifications appear from that icon. You will have noticed that when you connect, the message “User xyx is connected” pops up. And if someone is connected to your desktop in desktop sharing, a “desktop is viewed” message appears. More about the !M icon in the system tray can be found here in the dedicated document: https://kb.nomachine.com/DT07S00239.

If I access the Linux machine directly the NoMachine server GUI reports that the NM server is active, but the NM icon can take a few minutes to appear, if it appears at all.

This is what needs investigating. That icon should be visible/accessible permanently if the service is running. It should not be appearing/disappearing. If it is slow to appear, again, this is not normal. We’ll wait for the logs, they will shed some light on what is happening. Your other topic contains the link to the instructions on how to enable them.

Is the Ubuntu installation running Wayland? If you switch to X.org, do you see the same behaviour?

[epiphany123]

I am on 22.04.3 and facing the same issue with Wayland after last weeks’ distro LTS updates. Tested v7 and v8.1 servers and it’s the same behaviour. You have to reboot the server on login in order for it to start. Otherwise it takes 10-15 minutes. Not sure if it will start at all if you can’t physically log in, or how long that would take, but I don’t have the time to test this during the week.

See nxserver debug level logs attached, there’s nothing useful elsewhere.

Attachments:

1. debug-nxserver-log.zip

[epiphany123]

This is still occurring with NoMachine version 8.10.1

[dje666]

The update to v8.10.1 has improved matters for me. I shall post further updates if the behaviour reappears.

ttfn,

dje666

[Britgirl]

We have been able to reproduce the same issue and are looking at possible solutions.

[Britgirl]

A workaround that you can try is adding graphical.target to nxserver.service in After=.

nxserver.service can be in one of these paths:

/lib/systemd/system/nxserver.service
/usr/lib/systemd/system/nxserver.service

with root permissions you should add graphical.target to the following line:

After = syslog.target network.target network-online.target sshd.service http-daemon.target.service htd.service

to look like this

After = syslog.target network.target network-online.target sshd.service http-daemon.target.service htd.service graphical.target

Then, still as root

systemctl daemon-reload

and

systemctl restart nxserver.service

Does this change anything?

[epiphany123]

Hi @Britgirl,

Thanks for the tip!

graphical.target and default.target aren’t fixing it as systemd discards them both as it detects a wrong cycle, but multi-user.target does the trick which makes sense, because it’s invoked prior to logging in. Then nx server starts correctly at gdm3 and you can connect and log-in remotely.

This should fix the problems for these users too – https://forum.nomachine.com/topic/no-available-desktop-on-this-server-with-dummy-displayport as logins

[Author]

Posts