This NCPA plugin for checks Nvidia GPU stats for all GPUS detected via the nvidia-smi executable on linux machines
- Python 3.5 or greater
- nagiosplugin module - https://pypi.org/project/nagiosplugin/
- pip3 install nagiosplugin
- install
check_nvidiasmi.py
into /usr/local/ncpa/plugins/check_nvidiasmi.py - ensure
/usr/local/ncpa/etc/ncpa.cfg
configured to use python3 binary for plugin scripts
usage: check_nvidiasmi.py [-h] [-a RANGE] [-A RANGE] [-u RANGE] [-U RANGE] [-m RANGE] [-M RANGE] [-t RANGE] [-T RANGE] [-p RANGE] [-P RANGE] [-v]
NCPA plugin to check Nvidia GPU status using nvidia-smi
optional arguments:
-h, --help show this help message and exit
-a RANGE, --avg_gpu_warning RANGE
warning if threshold is outside RANGE for average of all GPUS
-A RANGE, --avg_gpu_critical RANGE
critical if threshold is outside RANGE for average of all GPUS
-u RANGE, --gpu_warning RANGE
warning if threshold is outside RANGE for any given GPU
-U RANGE, --gpu_critical RANGE
critical if threshold is outside RANGE for any given GPU
-m RANGE, --mem_warning RANGE
warning if threshold is outside RANGE for any given GPU
-M RANGE, --mem_critical RANGE
critical if threshold is outside RANGE for any given GPU
-t RANGE, --temp_warning RANGE
warning if threshold is outside RANGE for any given GPU
-T RANGE, --temp_critical RANGE
critical if threshold is outside RANGE for any given GPU
-p RANGE, --procs_warning RANGE
warning if threshold is outside RANGE for any given GPU
-P RANGE, --procs_critical RANGE
critical if threshold is outside RANGE for any given GPU
-v, --verbose increase verbosity (use up to 3 times)