GPU Server Monitor

Display the status of your GPU servers in a web page.

Environment

Create the environment using conda:

conda create -n gpu-monitor-web python=3.7
conda activate gpu-monitor-web
pip install flask apscheduler markdown2 paramiko

Besides, make sure gpustat is installed on your GPU server:

/usr/bin/pip3 install gpustat

Configurations

Generate your SSH keys and copy them to your GPU server:

ssh-keygen
ssh-copy-id username@hostname -p port

Create your configuration from the template:

cp config_template.py config.py

Save your modifications to config.py.

Run

python main.py