Glances API sees GPU but not the dashboard

[charlesnadeau]

I recently upgraded my GPU. The old one displayed perfectly in the dash board but not the new one although the API sees it:

charles@paris:~$ python
Python 3.10.12 (main, Nov 4 2025, 08:48:33) [GCC 11.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.

from glances import api
gl = api.GlancesAPI()
Error while initializing the ports plugin (module 'netifaces' has no attribute 'default_gateway')
gl.gpu
{'nvidia0': {'key': 'gpu_id', 'gpu_id': 'nvidia0', 'name': 'NVIDIA GeForce RTX 5060 Ti', 'mem': 2.832444362700018, 'proc': 0, 'temperature': 36, 'fan_speed': 0}}

Also:

charles@paris:~$ glances --stdout gpu
Error while initializing the ports plugin (module 'netifaces' has no attribute 'default_gateway')
gpu: [{'key': 'gpu_id', 'gpu_id': 'nvidia0', 'name': 'NVIDIA GeForce RTX 5060 Ti', 'mem': 2.832444362700018, 'proc': 0, 'temperature': 36, 'fan_speed': 0}]

But when I start Glances with:

sudo glances --disable-plugin smart

It is nowhere to be found. What should I do/check to make it appear on the dashboard?
Thank you very much!
Charles

[nicolargo]

Hi @charlesnadeau

Strange... It should work, please only check that you have enough space in your terminal.

[charlesnadeau]

Nico,
Merci! Oui mon terminal est assez grand. Voir photo ci-jointe.
Charles

[nicolargo]

Ok, peux tu lancer Glances en mode debug (-d) et faire un pastebin du fichier de log ?

[charlesnadeau]

Voici: https://pastebin.com/CSU4uVN7
Merci!

Charles

[nicolargo]

Strange when i simulate your GPU, it is correctly displayed on my computer:

[nicolargo]

Can you copy/paste your glances.conf file ?

[charlesnadeau]

Le voici:

charles@paris:/etc$ cat /etc/glances/glances.conf
[quicklook]
list=cpu,load,mem,swap
[diskio]
hide=loop.,sd[a-z][\d],nvme[\d]n[\d]p[\d]$,.databases-data$
[fs]
hide=/snap.,/dev/loop.,/databases-data
[sensors]
hide=nct67.*
[amp_postgres]
enable=true
regex=.postgres.|pg_.*
refresh=5
countmin=12
countmax=36
[amp_zfs]
enable=true
regex=z_.|txg_.|arc_.|zvol.|zthr.|./zfs.|dp_.
refresh=5
countmin=12
countmax=36
[amp_boinc]
enable=true
regex=boinc.|dnetc.
refresh=5
countmin=8
countmax=16
[gpu]
proc_careful=33
proc_warning=50
proc_critical=90
mem_careful=33
mem_warning=50
mem_critical=90
#Temperature
temperature_careful=35
temperature_warning=50
temperature_critical=65

Merci!

Charles

[nicolargo]

I redo the test with your Glances conf file and no problem on my side, the plugin is displayed:

I do not understand why it is not displayed on your side... :(

[charlesnadeau]

After a few reboots still no GPU and also no base frequenc for my CPU. What should I do/check next?
Merci!

Charles

[nicolargo]

You need to add some debug message in the code...

In the /usr/local/lib/python3.10/dist-packages/glances/plugins/gpu/init.py add the two lines (with logger)

    def msg_curse(self, args=None, max_width=None):
        """Return the dict to display in the curse interface."""
        # Init the return message
        ret = []

        # Only process if stats exist, not empty (issue #871) and plugin not disabled
        if not self.stats or (self.stats == []) or self.is_disabled():
            return ret

        # Add the two following lines here
        from glances.logger import logger
        logger.info("ENTER IN THE GPU DISPLAY METHOD")

        # Check if all GPU have the same name
        same_name = all(s['name'] == self.stats[0]['name'] for s in self.stats)

Restart Glances and check the log file for "ENTER IN THE GPU DISPLAY METHOD" message.