环境安装
安装Python3.6.8(用root安装)
安装依赖库
yum -y install git gcc make patch zlib-devel gdbm-devel openssl-devel sqlite-devel bzip2-devel readline-devel
下载Python3.6.8安装包
wget https://www.python.org/ftp/python/3.6.8/Python-3.6.8.tgz
创建安装目录
mkdir -p /usr/local/python3
解压安装包
tar -zxvf Python-3.6.8.tgz
进入解压后的目录，编译安装
cd Python-3.6.8
./configure --prefix=/usr/local/python3
make && make install
建立python3的软链接
ln -s /usr/local/python3/bin/python3 /usr/bin/python3
将/usr/local/python3/bin加入环境变量
vi ~/.bashrc
export PATH=$PATH:/usr/local/python3/bin
source ~/.bashrc
修改Python3的用户以及组
chown -R maxwell /usr/local/python3/
chgrp -R maxwell /usr/local/python3/
安装项目依赖包(离线安装，事先准备好安装文件)
安装flask及其依赖包
pip3 install Click-7.0-py2.py3-none-any.whl
pip3 install Werkzeug-0.14.1-py2.py3-none-any.whl
pip3 install itsdangerous-1.1.0-py2.py3-none-any.whl
pip3 install MarkupSafe-1.1.1-cp36-cp36m-manylinux1_x86_64.whl
pip3 install Jinja2-2.10.1-py2.py3-none-any.whl
pip3 install Flask-1.0.3-py2.py3-none-any.whl
安装tensorflow及其依赖包
tar -xzvf PyYAML-5.1.2.tar.gz
cd PyYAML-5.1.2/
python3 setup.py install
tar -xzvf gast-0.2.2.tar.gz
cd gast-0.2.2/
python3 setup.py install
tar -xzvf termcolor-1.1.0.tar.gz
cd termcolor-1.1.0/
python3 setup.py install
tar -xzvf absl-py-0.7.1.tar.gz
cd absl-py-0.7.1/
python3 setup.py install
pip3 install six-1.12.0-py2.py3-none-any.whl
pip3 install numpy-1.17.0-cp36-cp36m-manylinux1_x86_64.whl
pip3 install h5py-2.9.0-cp36-cp36m-manylinux1_x86_64.whl
pip3 install scipy-1.3.0-cp36-cp36m-manylinux1_x86_64.whl
pip3 install Keras-2.1.6-py2.py3-none-any.whl
pip3 install Keras_Applications-1.0.6-py2.py3-none-any.whl
pip3 install grpcio-1.22.0-cp36-cp36m-manylinux1_x86_64.whl
pip3 install wheel-0.33.4-py2.py3-none-any.whl
pip3 install astor-0.8.0-py2.py3-none-any.whl
pip3 install protobuf-3.9.0-cp36-cp36m-manylinux1_x86_64.whl
pip3 install Markdown-3.1.1-py2.py3-none-any.whl
pip3 install setuptools-41.0.1-py2.py3-none-any.whl
pip3 install tensorboard-1.12.0-py3-none-any.whl
pip3 install Keras_Preprocessing-1.1.0-py2.py3-none-any.whl
pip3 install tensorflow-1.12.0-cp36-cp36m-manylinux1_x86_64.whl
安装pandas及其依赖包
pip3 install python_dateutil-2.8.0-py2.py3-none-any.whl
pip3 install pytz-2019.2-py2.py3-none-any.whl
pip3 install pandas-0.24.2-cp36-cp36m-manylinux1_x86_64.whl
安装gunicorn
pip3 install gunicorn-19.9.0-py2.py3-none-any.whl
安装requests及其依赖包
pip3 install urllib3-1.24-py2.py3-none-any.whl
pip3 install idna-2.8-py2.py3-none-any.whl
pip3 install certifi-2019.6.16-py2.py3-none-any.whl
pip3 install chardet-3.0.4-py2.py3-none-any.whl
pip3 install requests-2.22.0-py2.py3-none-any.whl
安装xlwt
pip3 install xlwt-1.3.0-py2.py3-none-any.whl
安装xlrd
pip3 install xlrd-1.2.0-py2.py3-none-any.whl
安装ibm_db
tar -xzvf ibm_db-3.0.1.tar.gz
cd ibm_db-3.0.1/
python3 setup.py install # 此步需要联网，需要下载驱动
安装gevent及其依赖包
pip3 install greenlet-0.4.15-cp36-cp36m-manylinux1_x86_64.whl
pip3 install gevent-1.4.0-cp36-cp36m-manylinux1_x86_64.whl
安装supervisor及其依赖包
pip3 install meld3-1.0.2-py2.py3-none-any.whl
pip3 install supervisor-4.0.4-py2.py3-none-any.whl
gunicorn配置，在项目根目录创建gc.py，编辑如下内容
import gevent.monkey
gevent.monkey.patch_all()
import multiprocessing
# 调试用，生产环境设为False
# debug = True
# 绑定地址:端口
bind = "0.0.0.0:8888"
# gunicorn进程ID日志
pidfile = "log/gunicorn.pid"
# loglevel用于控制errorlog的信息级别，可以设置为debug、info、warning、error、critical
loglevel = "debug"
# accesslog是访问日志，可以通过access_log_format设置访问日志格式
accesslog = "log/access.log"
errorlog = "log/debug.log"
# True意味着开启后台运行，默认为False
daemon = False
# 请求超时时间
timeout = 100
# 启动的进程数
# workers = multiprocessing.cpu_count()
workers = 4
# 进程的模式类型，默认为sync模式
worker_class = "gevent"
x_forwarded_for_header = 'X-FORWARDED-FOR'
supervisor配置
导出配置模板到项目根目录
echo_supervisord_conf > supervisor.conf
编辑supervisor.conf，并在尾部添加内容如下
[program:risk] # risk是自己给进程取的名，随意
command=gunicorn -c gc.py manager:flask_app
;directory=/home/maxwell/Downloads/riskcontrol # 项目目录
startsecs=0
stopwaitsecs=0
autostart=true
autorestart=true
关闭系统防火墙
查看防火墙状态
firewall-cmd --state
关闭防火墙
systemctl stop firewalld.service
启动服务
方式一：通过gunicorn运行服务(缺点是需要通过查找pid杀掉进程)
gunicorn -c gc.py manager:flask_app
方式二：通过supervisor运行服务(推荐)
> 加载配置文件并启动supervisor
supervisord -c supervisor.conf
> 修改完配置文件后重新启动supervisor
supervisorctl -c supervisor.conf reload
> 开启服务
supervisorctl -c supervisor.conf start risk
supervisor命令参考
supervisorctl -c supervisor.conf reload :修改完配置文件后重新启动supervisor
supervisorctl -c supervisor.conf status :查看supervisor监管的进程状态
supervisorctl -c supervisor.conf start 进程名 ：启动XXX进程 进程名就是你刚刚配置文件的进程名
supervisorctl -c supervisor.conf stop 进程名 ：停止XXX进程
supervisorctl -c supervisor.conf stop all：停止全部进程，注：start、restart、stop都不会载入最新的配置文件。
supervisorctl -c supervisor.conf update：根据最新的配置文件，启动新配置
参考资料
Centos安装Python
Supervisor命令详解
Gunicorn部署Flask，并用Supervisor托管
网友评论