The tool developed to collect the gpus info of the linux cluster in my lab and display them in wechat mini-program "考研备忘录"

1. hostname
2. ram_available
3. gpu_name, gpu_memory_total, gpu_memory_free, gpu_memory_used

hostname
cat /proc/meminfo | grep MemAvailable | tr -cd "[0-9]"
nvidia-smi --query-gpu=name,memory.total,memory.free,memory.used --format=csv,noheader,nounits

The info listed above are collected using the project.
Check the code for more details
About nvidia-smi

1. hostname
2. ram_available
3. gpu_name, gpu_memory_total, gpu_memory_free, gpu_memory_used

The info listed above are displayed in mini-program "考研备忘录", which is not an open-source project yet.

numpy
requests

git clone https://github.com/2812427914/ServerTool.git
or
git clone https://github.com.cnpmjs.org/2812427914/ServerTool.git
(faster, mirror)

cd ServerTool

before run bash wgpu.sh

1. set the python_path in wgpu.sh . (The python version contains the packages required)
2. set the cron_freq in wgpu.sh .(Optional; the time in crontab tasks, in minutes; default 3 minutes and recommended)
3. set the group_name = "bdaa_edu" in main_v1.py.(Later feature; "bdaa_edu" supported only currently)

bash wgpu.sh

sis cluster python_path=~/anaconda3/envs/learn/bin/python3.6
pangpang clusterpython_path=/usr/bin/python3.6
huzx clusterpython_path=/usr/bin/python3.7

git clean -f -d
git fetch --all
git reset --hard origin/master

Then do not forget to:

1. set the python_path in wgpu.sh(!!!).
2. set the cron_freq in wgpu.sh .
3. set the group_name in main_v1.py.

rm access_token.txt record_id.txt cron ; rm -r gpustat_v1/ __pycache__/
git add -A
git commit -m 'fix bugs'
git pull origin master
git push -u origin master

1. Add linux group feature. (one can set the group_name and check the gpus info in the "考研备忘录" according to group_name)
2. Change the way of setting the python_path to the level of naive
3. Support setting the variables (cron_freq, alive_servers) of a linux cluster through logining in one of the them.