Nvidia_smi plugin: Parsing of section nvidia_smi failed

Troubleshooting
1

CMK version: 2.3.Op6
OS version: ubuntu 22.04

Hello Team,

Iam running a few Ubuntu 22.04 servers with Nvidia GPU devices installed, for the same to monitor I have installed the “nvidia_smi” plugin in /usr/lib/check_mk_agent/plugins added required parameters for nvidia_smi as per the git doc mentioned checkmk/agents/plugins/nvidia_smi at 889a09e9380d781ebc8509c5cc665e61307ebe6a · Checkmk/checkmk · GitHub
but Iam experiencing “Parsing of section nvidia_smi failed - please submit a crash report!“ once the service discovery is done on the host, kindly let me know is this is affected for the version 2.3.Op6 any workarounds to solve the issue.

Appreciate your help

Regards,
Aru

2

Can you share the crash file? Your Checkmk version is very old. Maybe whatever produces your crash might be fixed in a more recent version.

3

Hi,

we’re having a similar issue, presumably.

We get the following crash report & agent_output from nvidia_smi:
agent_output.txt (215.4 KB)
crash.txt (418.1 KB)

Seems like the power_state value has been deprecated.
Additionally, I might add, the temperature group has changed and added tlimit in the names of the sub sections.

Thanks for having a look!

4

From your output:

<power_state>Requested functionality has been deprecated</power_state>

You should upgrade to 2.4.0p20 or later. This includes Werk #19132: Nvidia SMI: Allow deprecation message for power_state

1 Like