Connection fails with client error 104

[NickLeaf]

I had NoMachine running between two Ubuntu 20.04 machines, but some recent software update seems to have broken my setup. Both machines are using Gnome3 and gdm3.

On the client machine, I can see the server in the connection browser, select it, and authenticate (via password)–this is the same process that was successful prior. But then the client just seems to enter a connect-fail-retry loop for a bit before finally giving me a “Error is 104: Connection reset by peer.”

On the server, I briefly see the “user xxxx has connected” message pop up, as you would expect after successful authentication. In nxserver.log I spotted the following message:

nxserver.log:2020-08-07 10:04:27 611.473  7705 NXSERVER WARNING! 93E0C3ECACCA41FD1BAEA1AD176C7D71 nxnode died with exit code 11.

And that seems to correspond with the following from /var/log/kern.log:

Aug  7 10:04:27 nleaf-hpcvis kernel: [  132.763381] nxnode.bin[7846]: segfault at 4 ip 00007fdadd9489af sp 00007fdacfffe450 error 6 in libnxdimi.so[7fdadd923000+5e000]

Aug  7 10:04:27 nleaf-hpcvis kernel: [  132.763393] Code: 8e 56 02 00 00 4c 8d af 08 8d 00 00 31 ed 45 31 e4 41 bf 01 00 00 00 eb 33 66 90 49 8b 45 00 c6 40 04 ff 49 8b 85 80 00 00 00 <c6> 40 04 ff 8d 45 01 49 83 c5 08 48 83 c5 01 39 83 d4 8a 00 00 48

I’ve tried a full reinstall of both the client and server, including wiping out all of the settings & log files that I could find (/usr/NX, $HOME/.nx, /var/lib/gdm3/.nx). I attached the client and server logs, let me know if there’s anything else that I’m missing.

Attachments:

1. nomachine-server-logs.tar.gz
2. nomachine-client-logs.tar.gz

[NickLeaf]

Small update: it seems like it might be related to the NVIDIA driver version. My setup worked fine with the 440.100 driver, but I ran into the same 104 error when I upgraded the driver to 450.57. This happens regardless of whether I have a physical display connected to the card or not. The crash report points at a segfault in X11Poller::getOutputsCount().

Link to the .crash report: https://drive.google.com/file/d/1yIv7WphtTC-VZzBXmkkYxINb9nZ4uYyC/view?usp=sharing

[graywolf]

Hello. Would you run the command xrandr for both the cases, with driver 440.100 and with 450.57?

[graywolf]

We can’t reproduce with latest nvidia driver 450.57.

Please collect info about your video card using command lshw -C display.

[NickLeaf]

Here’s the output of  lshw -C display:

*-display
description: VGA compatible controller
product: TU102B
vendor: NVIDIA Corporation
physical id: 0
bus info: pci@0000:01:00.0
version: a1
width: 64 bits
clock: 33MHz
capabilities: pm msi pciexpress vga_controller bus_master cap_list rom
configuration: driver=nvidia latency=0
resources: irq:52 memory:fa000000-faffffff memory:c0000000-cfffffff memory:d0000000-d1ffffff ioport:e000(size=128) memory:fb000000-fb07ffff
*-display
description: VGA compatible controller
product: TU102B
vendor: NVIDIA Corporation
physical id: 0
bus info: pci@0000:06:00.0
version: a1
width: 64 bits
clock: 33MHz
capabilities: pm msi pciexpress vga_controller bus_master cap_list rom
configuration: driver=nvidia latency=0
resources: irq:53 memory:f8000000-f8ffffff memory:a0000000-afffffff memory:b0000000-b1ffffff ioport:d000(size=128) memory:f9000000-f907ffff

And here’s xrandr with the 450.57 driver. Note that this is with a physical screen connected and in use (though NoMachine still fails to connect) I’ll try to get the output from 440.100 tonight when I have the chance to swap drivers.

nleaf@nleaf-hpcvis:~$ nvidia-smi | head -n 3 | tail -n 1
| NVIDIA-SMI 450.57       Driver Version: 450.57       CUDA Version: 11.0     |
nleaf@nleaf-hpcvis:~$ DISPLAY=:1 xrandr
Screen 0: minimum 8 x 8, current 1920 x 1080, maximum 32767 x 32767
DP-0 disconnected (normal left inverted right x axis y axis)
DP-1 disconnected (normal left inverted right x axis y axis)
DP-2 disconnected (normal left inverted right x axis y axis)
DP-3 disconnected (normal left inverted right x axis y axis)
DP-4 disconnected (normal left inverted right x axis y axis)
DP-5 disconnected (normal left inverted right x axis y axis)
DP-6 disconnected (normal left inverted right x axis y axis)
DP-7 connected primary 1920x1080+0+0 (normal left inverted right x axis y axis) 510mm x 290mm
1920x1080 60.00*+
1680x1050 59.95
1440x900 59.89
1280x1024 60.02
1280x800 59.81
1152x864 75.00
1024x768 70.07 60.00
800x600 60.32 56.25
640x480 59.94
USB-C-0 disconnected (normal left inverted right x axis y axis)
DP-1-0 disconnected
DP-1-1 disconnected
DP-1-2 disconnected
DP-1-3 disconnected
DP-1-4 disconnected
DP-1-5 disconnected
DP-1-6 disconnected
DP-1-7 disconnected
USB-C-1-0 disconnected

[NickLeaf]

Here’s the same info with the 440.100 drivers:

nleaf@nleaf-hpcvis:~$ nvidia-smi | head -n 3 | tail -n 1
| NVIDIA-SMI 440.100      Driver Version: 440.100      CUDA Version: 10.2     |
nleaf@nleaf-hpcvis:~$ DISPLAY=:1 xrandr
Screen 0: minimum 8 x 8, current 1920 x 1080, maximum 32767 x 32767
DP-0 disconnected (normal left inverted right x axis y axis)
DP-1 disconnected (normal left inverted right x axis y axis)
DP-2 disconnected (normal left inverted right x axis y axis)
DP-3 disconnected (normal left inverted right x axis y axis)
DP-4 disconnected (normal left inverted right x axis y axis)
DP-5 disconnected (normal left inverted right x axis y axis)
DP-6 disconnected (normal left inverted right x axis y axis)
DP-7 connected primary 1920x1080+0+0 (normal left inverted right x axis y axis) 510mm x 287mm
1920x1080     60.00*+
1680x1050     59.95
1440x900      59.89
1280x1024     60.02
1280x800      59.81
1152x864      75.00
1024x768      70.07    60.00
800x600       60.32    56.25
640x480       59.94
USB-C-0 disconnected (normal left inverted right x axis y axis)

And lshw -C display:

nleaf@nleaf-hpcvis:~$ sudo lshw -C display
*-display
description: VGA compatible controller
product: TU102B
vendor: NVIDIA Corporation
physical id: 0
bus info: pci@0000:01:00.0
version: a1
width: 64 bits
clock: 33MHz
capabilities: pm msi pciexpress vga_controller bus_master cap_list rom
configuration: driver=nvidia latency=0
resources: irq:52 memory:fa000000-faffffff memory:c0000000-cfffffff memory:d0000000-d1ffffff ioport:e000(size=128) memory:fb000000-fb07ffff
*-display
description: VGA compatible controller
product: TU102B
vendor: NVIDIA Corporation
physical id: 0
bus info: pci@0000:06:00.0
version: a1
width: 64 bits
clock: 33MHz
capabilities: pm msi pciexpress vga_controller bus_master cap_list rom
configuration: driver=nvidia latency=0
resources: irq:53 memory:f8000000-f8ffffff memory:a0000000-afffffff memory:b0000000-b1ffffff ioport:d000(size=128) memory:f9000000-f907ffff

[graywolf]

Hello. Would you try a custom NoMachine package in order to help the debugging?

[NickLeaf]

Sure, I’d be willing to give it a shot.

[graywolf]

Thank you very much for your cooperation. We identified the problem. It’ll be fixed in one of the next releases.

[Author]

Posts