1.获取gpu信息 [root@ks-devops python]# cat gpu_info.py import pynvml UNIT = 1024 * 1024 def GpuInfo(): pynvml.nvmlInit() #初始化 gpuDeriveInfo = pynvml.nvmlSystemGetDriverVersion() print("Drive: ", str(gpuDeriveInfo, encoding='utf-8')) #显示驱动信息 gpuDeviceCount = pynvml.nvmlDeviceGetCount()#获取Nvidia GPU块数 print("GPU", gpuDeviceCount ) gpu_list = [] for i in range(gpuDeviceCount): handle = pynvml.nvmlDeviceGetHandleByIndex(i)#获取GPU i的handle,后续通过handle来处理 gpuName = str(pynvml.nvmlDeviceGetName(handle), encoding='utf-8') memoryInfo = pynvml.nvmlDeviceGetMemoryInfo(handle) print("GPU", i, ":", gpuName, memoryInfo.total/UNIT,"MB") gpu_list.append(str(i)+":"+gpuName+" "+str(memoryInfo.total/UNIT)+"MB") return gpu_list if __name__ == "__main__": gpu_info = GpuInfo() pynvml.nvmlShutdown() #print(gpu_devices)
2.硬件序列号设定(自定义的,随机大小写字母+时间) [root@ks-devops python]# cat sn_code.py import string import random import time def sncode(): # Randomly choose a letter from all the ascii_letters str = "" for num in range(9): randomLetter = random.choice(string.ascii_letters) str += randomLetter #print(randomLetter,end="") now = time.localtime() now_time = time.strftime("%Y%m%d%H%M%S", now) #print(now_time) print(str+now_time) return str+now_time #sncode() if __name__ == '__main__': #sncode() pass
3.flask框架接口定义 [root@ks-devops python]# cat app.py from flask import Flask,jsonify from gpu_info import GpuInfo from sn_code import sncode app = Flask(__name__) @app.route('/gpuinfo') def gpu_info(): gi = jsonify("gpuinfo",GpuInfo()) return gi @app.route('/sncode') def sn_code(): sn = jsonify("sn",sncode()) return sn @app.route('/health') def status(): return "ok" @app.route('/') def index(): return "test lipc" if __name__ == '__main__': app.run(host='0.0.0.0', port=5000)
4.制作python底层镜像 (1)下载centos:7基础镜像 docker pull centos:7 (2)运行临时容器 docker run -it --rm --name python-test centos:7 (3)安装python环境和flask库 yum -y install python3 python3 -m pip install --upgrade --force pip pip3 install sanic pip3 install flask pip3 install pynvml yum clean all (4)将容器打包成镜像 #查看容器id docker ps |grep python-test #将容器打包成镜像 docker commit f185eb722b13 172.16.4.17:8888/base/centos-py3:v1 5.制作python运行镜像 #当前目录下的文件 [root@ks-devops python]# ls app.py Dockerfile gpu_info.py py_base_images.sh sn_code.py start.sh (1)编写dockerfile [root@ks-devops python]# cat Dockerfile FROM 172.16.4.17:8090/base/centos-py3:v2 WORKDIR /data RUN virtualenv venv RUN source ./venv/bin/activate COPY . /data/venv HEALTHCHECK --interval=5s --timeout=3s --retries=3 \ CMD curl -fs http://127.0.0.1:5000/health || exit 1 ENTRYPOINT ["sh","/data/venv/start.sh"] (2)制作镜像 docker build . -t 172.16.4.17:8888/test/up-lipc:v1.1 (3)运行容器 docker run -itd --name flask-test --network host --gpus all 172.16.4.17:8888/test/up-lipc:v1.1 (4)查看容器运行状态 docker ps
6.访问测试
标签:__,硬件资源,flask,app,python,获取,pynvml,gpu,-- From: https://www.cnblogs.com/Leonardo-li/p/16748394.html