This document records the setup of a production environment for a Flask application on Alibaba Cloud’s CentOS7.
Configuring CentOS7
After logging in as root, first create a new regular user and set a password:
adduser user
passwd user
Next, copy the user’s public key to ~/.ssh, name it authorized_keys, and modify /etc/ssh/sshd_config to disable root login via SSH, change the default SSH port, and use certificate login. Modify the following settings:
Port 65535
PasswordAuthentication no
PermitRootLogin no
After configuration, restart the SSH service:
systemctl restart sshd.service
In CentOS7, firewalld replaces iptables. You need to manually add port 80 and the modified SSH port to firewalld:
firewalld --add-port 80/tcp --permanent
firewalld --add-port 65535/tcp --permanent
firewalld --reload
Configuring Nginx
Nginx can be installed directly via yum:
yum install nginx
After installation, add a location configuration in /etc/nginx/default.d, pointing to port 8001, which Flask will listen to:
location / {
proxy_pass http://127.0.0.1:8001
}
Reload the Nginx configuration after setting it up:
systemctl reload nginx.service
Installing Python
CentOS comes with Python 2.7. If you need Python 3, install it separately:
wget https://www.python.org/ftp/python/3.4.3/Python-3.4.3.tgz
tar xf Python-3.4.3.tgz
cd Python-3.4.3
./configure --prefix=/usr/local --enable-shared
make
make altinstall
Next, set up a virtual environment in the project. Use virtualenv for Python 2 (install with pip install virtualenv) and pyvenv for Python 3 (Python 3 comes with pyvenv). For a Python 3 environment, create a virtual environment folder venv in the web project:
pyvenv venv
Activate the virtual environment in the project path:
source venv/bin/activate
Exit the virtual environment using Ctrl+C or the deactivate command.
Installing and Configuring Gunicorn
Install Gunicorn using pip install gunicorn. Make sure to use pip within the virtual environment to match the Python version of the virtual environment. After installation, create a Gunicorn configuration file, such as deploy_config.py, with the following content:
import os
bind='127.0.0.1:8001' # Bound port
workers=4 # Number of workers
backlog=2048
debug=True
proc_name='gunicorn.pid'
pidfile='/var/log/gunicorn/debug.log'
loglevel='debug'
Save the file and try to start Gunicorn in the virtual environment:
gunicorn -c deploy_config.py myapp:app
myapp is the entry Python file name, and app is the function name. If worker-related information is output, it indicates a successful start.
Installing and Configuring Supervisor
Supervisor can be installed directly via yum. Note that Supervisor only supports Python 2, so do not use pip to install Supervisor in the virtual environment.
yum install supervisor
After installation, create a configuration file xxx.ini in /etc/supervisord.d with the following content:
[program:xxx]
command=/var/proj/xxx/venv/bin/python /usr/bin/gunicorn -c /var/proj/xxx/deploy_config.py myapp:app
autorstart=true
directory=/var/proj/xxx
autorestart=true
startsecs=10
startretries=20
xxx is the project name. Ensure that the command paths are fully specified to distinguish between the system environment and the project virtual environment. After completion, start supervisord to apply the configuration:
supervisord -c /etc/supervisord.conf