docker-compose 不能等待 mysql 数据库

Question

我在尝试获取 docker-compose 脚本来启动 mysql 数据库和 Django 项目时遇到了真正的问题，但是让 Django 项目等到 mysql 数据库准备好。

我有两个文件，一个 Dockerfile 和一个 docker-compose.yml，我在下面复制了它们。

当我运行 docker-compose.yml 并检查 Web 容器的日志时，它说它无法连接到数据库 mydb。然而，我第二次运行它（没有清除容器和图像）它正确连接并且 Django 应用程序工作正常。

我花了一整天的时间尝试了很多东西，例如脚本、健康检查等，但我无法让它发挥作用。

Dockerfile

FROM python:3.6

ENV PYTHONUNBUFFERED 1

RUN mkdir /code
WORKDIR /code

COPY ./ /code/

RUN pip install -r requirements.txt
RUN python manage.py collectstatic --noinput

docker-compose.yml

version: '3'

services:
  mydb:
    environment:
      - MYSQL_ROOT_PASSWORD=password
      - MYSQL_USER=django
      - MYSQL_PASSWORD=secret
      - MYSQL_DATABASE=dbMarksWebsite
    image: mysql:5.7
    ports:
      # Map default mysql port 3306 to 3308 on outside so that I can connect
      # to mysql using workbench localhost with port 3308
      - "3308:3306"
  web:
    environment:
      - DJANGO_DEBUG=1
      - DOCKER_PASSWORD=secret
      - DOCKER_USER=django
      - DOCKER_DB=dbMarksWebsite
      - DOCKER_HOST=mydb
      - DOCKER_PORT=3306
    build: .
    command: >
      sh -c "sleep 10 &&
             python manage.py migrate &&
             python manage.py loaddata myprojects_testdata.json &&
             python manage.py runserver 0.0.0.0:8080"
    ports:
      - "8080:8080"
    depends_on:
      - mydb

首次运行（没有现有图像或容器）：

...
  File "/usr/local/lib/python3.6/site-packages/MySQLdb/__init__.py", line 84, in Connect
    return Connection(*args, **kwargs)
  File "/usr/local/lib/python3.6/site-packages/MySQLdb/connections.py", line 179, in __init__
    super(Connection, self).__init__(*args, **kwargs2)
django.db.utils.OperationalError: (2002, "Can't connect to MySQL server on 'mydb' (115)")

第二次运行：

System check identified no issues (0 silenced).
March 27, 2020 - 16:44:57
Django version 2.2.11, using settings 'ebdjango.settings'
Starting development server at http://0.0.0.0:8080/
Quit the server with CONTROL-C.

Answer 1

我在 entrypoint.sh 中使用以下函数解决了这个问题：

function wait_for_db()
{
  while ! ./manage.py sqlflush > /dev/null 2>&1 ;do
    echo "Waiting for the db to be ready."
    sleep 1
  done
}

Answer 2

对于任何感兴趣的人，我找到了解决方案：

1 - 我写了一个python脚本来每秒连接到数据库， 但超时。我将此超时设置为相当高，为 60 秒，但这似乎在我的电脑上工作。

2 - 我将等待的命令添加到我的撰写文件中。

这应该意味着我可以为我的网站调出一组测试容器，我可以在其中指定使用的 Python 和 MySQL 的确切版本。

相关文件如下：

Dockerfile：

FROM python:3.6

ENV PYTHONUNBUFFERED 1

RUN mkdir /code
WORKDIR /code

COPY ./ /code/

RUN pip install -r requirements.txt
RUN python manage.py collectstatic --noinput

docker-compose.yml

version: '3'

services:
  mydb:
    container_name: mydb
    environment:
      - MYSQL_ROOT_PASSWORD=password
      - MYSQL_USER=django
      - MYSQL_PASSWORD=secret
      - MYSQL_DATABASE=dbMarksWebsite
    image: mysql:5.7
    ports:
      # Map default mysql port 3306 to 3308 on outside so that I can connect
      # to mysql using workbench localhost with port 3308
      - "3308:3306"
  web:
    container_name: web
    environment:
      - DJANGO_DEBUG=1
      - DOCKER_PASSWORD=secret
      - DOCKER_USER=django
      - DOCKER_DB=dbMarksWebsite
      - DOCKER_HOST=mydb
      - DOCKER_PORT=3306
    build: .
    command: >
      sh -c "python ./bin/wait-for.py mydb 3306 django secret dbMarksWebsite 60  &&
             python manage.py migrate &&
             python manage.py loaddata myprojects_testdata.json &&
             python manage.py runserver 0.0.0.0:8080"
    ports:
      - "8080:8080"
    depends_on:
      - mydb

等待.py

'''
I don't like adding this in here, but I cannot get the typical wait-for scripts
to work with MySQL database in docker, so I hve written a python script that
either times out after ? seconds or successfully connects to the database
The input arguments for the script need to be:

    HOST, PORT, USERNAME, PASSWORD, DATABASE, TIMEOUT

'''
import sys, os
import time
import pymysql

def readCommandLineArgument():
    '''
    Validate the number of command line input arguments and return the
    input filename
    '''

    # Get arguments
    if len(sys.argv)!=7:
        raise ValueError("You must pass in 6 arguments, HOST, PORT, USERNAME, PASSWORD, DATABASE, TIMEOUT")

    # return the arguments as a tuple
    return (sys.argv[1], sys.argv[2], sys.argv[3], sys.argv[4], sys.argv[5], sys.argv[6]) 

def connectToDB(HOST, PORT, USERNAME, PASSWORD, DATABASE):
    '''
    for now, just try to connect to the database.
    '''

    con = pymysql.connect(host=HOST, port=PORT, user=USERNAME, password=PASSWORD, database=DATABASE)

    with con:
        cur = con.cursor()
        cur.execute("SELECT VERSION()")

def runDelay():
    '''
    I don't like passing passwords in, but this is only used for a test docker
    delay script
    '''

    # Get the database connection characteristics.
    (HOST, PORT, USERNAME, PASSWORD, DATABASE, TIMEOUT) = readCommandLineArgument()

    # Ensure timeout is an integer greater than zero, otherwise use 15 secs a default
    try:
        TIMEOUT = int(TIMEOUT)
        if TIMEOUT <= 0:
            raise("Timeout needs to be > 0")
    except:
        TIMEOUT = 60

    # Ensure port is an integer greater than zero, otherwise use 3306 as default
    try:
        PORT = int(PORT)
        if PORT <= 0:
            raise("Port needs to be > 0")
    except:
        PORT = 3306

    # Try to connect to the database TIMEOUT times
    for i in range(0, TIMEOUT):

        try:
            # Try to connect to db
            connectToDB(HOST, PORT, USERNAME, PASSWORD, DATABASE)

            # If an error hasn't been raised, then exit
            return True

        except Exception as Ex:
            strErr=Ex.args[0]
            print(Ex.args)
            # Sleep for 1 second
            time.sleep(1)

    # If I get here, assume a timeout has occurred
    raise("Timeout")


if __name__ == "__main__":

    runDelay()

Answer 3

另一种选择是使用脚本来控制启动顺序，并包装网络服务的命令。

docker-compose's documentation 中的“wait-for-it”是推荐的工具之一，但其他存在。

Answer 4

出于测试/开发目的，您可以使用具有运行状况检查的 MySQL 映像版本（我相信有一个 healthcheck/mysql 映像），或配置您自己的（参见此处的示例：Docker-compose check if mysql connection is ready）。

对于生产用途，您不想在启动时升级数据库架构，也不想假设数据库已启动。升级模式会自动鼓励您不要考虑部署错误并需要回滚时会发生什么，并且并行模式升级将不起作用。加长版：https://pythonspeed.com/articles/schema-migrations-server-startup/