Skip to content

Conversation

@mr-creditcard
Copy link

General information

In the NVIDIA smi schema v13 the elements containing the power draw & power limit of the GPU have been changed. This PR contains a workaround for this, similar to an already existing workaround in the nvidia_smi plugin.

Currently the workaround is implemented using the same logic of an already existing workaround (where power_readings was renamed to gpu_power_readings, see f99722a & #681). Just adding on to the pile doesn't feel as elegant, but it works. While doing some research I came across Telegraf, which also supports parsing the XML from NVIDIA smi. They fetch the schema version from the DOCTYPE and then use a parser for that specific version (https://github.com/influxdata/telegraf/blob/master/plugins/inputs/nvidia_smi/nvidia_smi.go#L112). Depending on how these schemas evolve in the future, this may be a more elegant solution.

Bug reports

After upgrading the NVIDIA driver on one of my machines from 530 to 580 I noticed that the power monitoring in CheckMK had less information available. The 'Power Draw' figure disappeared from the host services overview, as did the graph in the service detail view and the power limit reported None.

The nvidia-smi man page mentions the following:

Changes between nvidia-smi v575 Update and v570

[...]

       • Updated 'nvidia-smi -q' to print both 'Instantaneous Power Draw' and 'Average Power Draw' in all cases where 'Power Draw' used to be printed.

Proposed changes

  • What is the expected behavior?
    The power draw and power limit are available in the web interface
  • What is the observed behavior?
    The aforementioned values are not available in the web interface

@github-actions
Copy link

github-actions bot commented Jan 29, 2026

All contributors have signed the CLA ✍️ ✅
Posted by the CLA Assistant Lite bot.

In the nvidia smi schema v13 the elements containing the power draw &
power limit of the GPU have been changed. This commit contains a
workaround for this, similar to an already existing workaround in the
nvidia_smi plugin.
@mr-creditcard mr-creditcard force-pushed the support-newer-nvidia-drivers branch from 5fee1ce to 8b1fc9e Compare January 29, 2026 08:13
@mr-creditcard
Copy link
Author

I have read the CLA Document and I hereby sign the CLA or my organization already has a signed CLA.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants