大数据集群运维（64）Airflow安装教程

 

1.安装环境

CentOS-6.5
Python-2.7.12
setuptools-29.0.1
pip-9.0.1

2.编译Python

sudo yum install -y gcc
sudo yum install -y gcc-c++
sudo yum install -y wget
sudo yum install -y mysql
sudo yum install -y mysql-devel
sudo yum install -y python-devel
sudo yum install -y zlib-devel
sudo yum install -y openssl-devel
sudo yum install -y sqlite-devel
wget https://www.python.org/ftp/python/2.7.12/Python-2.7.12.tgz

sudo mkdir /usr/local/python27
sudo tar zxfv Python-2.7.12.tgz -C /usr/local/
cd /usr/local/Python-2.7.12/
./configure --prefix=/usr/local/python27
make
make install
sudo mv /usr/bin/python /usr/bin/python2.6
sudo ln -sf /usr/local/python/bin/python /usr/bin/python2.7

vim /usr/bin/yum
#!/usr/bin/python2.6

vim /etc/profile
export PYTHON_HOME=/usr/bin/python2.6
export PATH=$PYTHON_HOME/bin:$PATH

wget https://pypi.python.org/packages/59/88/2f3990916931a5de6fa9706d6d75eb32ee8b78627bb2abaab7ed9e6d0622/setuptools-29.0.1.tar.gz#md5=28ecfd0f2574b489b9a18343879a7324
tar zxfv setuptools-29.0.1.tar.gz
cd setuptools-29.0.1
python setup.py install

wget https://pypi.python.org/packages/11/b6/abcb525026a4be042b486df43905d6893fb04f05aac21c32c638e939e447/pip-9.0.1.tar.gz#md5=35f01da33009719497f01a4ba69d63c9
tar zxfv pip-9.0.1.tar.gz
cd pip-9.0.1
python setup.py install
pip install --upgrade pip

wget https://pypi.python.org/packages/a5/e9/51b544da85a36a68debe7a7091f068d802fc515a3a202652828c73453cad/MySQL-python-1.2.5.zip#md5=654f75b302db6ed8dc5a898c625e030c
unzip MySQL-python-1.2.5.zip
cd MySQL-python-1.2.5
python setup.py install

#第三方包 /usr/local/python27/lib/python2.7/site-packages

3.安装

    airflow通过pip可以方便的安装到系统中。

# airflow needs a home, ~/airflow is the default,
# but you can lay foundation somewhere else if you prefer
# (optional)
export AIRFLOW_HOME=/usr/local/airflow

# install from pypi using pip
pip install airflow
pip install airflow[hive]

# initialize the database
airflow initdb

# start the web server, default port is 8080
airflow webserver -p 8080

4.设置mysql为元数据库

#首先要安装mysql客户端
sudo yum install -y mysql
sudo yum install -y mysql-devel

CREATE USER airflow;
CREATE DATABASE airflow;
CREATE DATABASE celery_result_airflow;

GRANT all privileges on airflow.* TO 'airflow'@'%' IDENTIFIED BY 'airflow';
GRANT all privileges on celery_result_airflow.* TO 'airflow'@'%' IDENTIFIED BY 'airflow';

#安装mysql模块
wget https://pypi.python.org/packages/a5/e9/51b544da85a36a68debe7a7091f068d802fc515a3a202652828c73453cad/MySQL-python-1.2.5.zip#md5=654f75b302db6ed8dc5a898c625e030c
unzip MySQL-python-1.2.5.zip
cd MySQL-python-1.2.5
python setup.py install

#在airflow的配置文件中配置mysql为元数据的存储库
sudo vi $AIRFLOW_HOME/airflow.cfg

#更改数据库链接：
sql_alchemy_conn = mysql://airflow:airflow@localhost:3306/airflow

#对应字段解释如下：
dialect+driver://username:password@host:port/database

#初始化元数据库
airflow initdb

#重置元数据库
airflow resetdb

5.安装登录模块

#安装password模块
pip install airflow[password]

#在airflow的配置文件中修改需要认证
sudo vi $AIRFLOW_HOME/airflow.cfg
[webserver]
authenticate = True
filter_by_owner = True
auth_backend = airflow.contrib.auth.backends.password_auth

运行以下代码将用户名密码写入元数据库中

import airflow
from airflow import models, settings
from airflow.contrib.auth.backends.password_auth import PasswordUser

user = PasswordUser(models.User())
user.username = 'quzhengpeng'
user.email = 'quzhengpeng@163.com'
user.password = 'quzhengpeng'
session = settings.Session()
session.add(user)
session.commit()
session.close()
exit()

5.启动守护进程

    启动后台守护进程了之后，Airflow才能实时监控任务的调度情况。将任务脚本放到${AIRFLOW_HOME}/dags下在web UI 就能看到任务执行情况。

airflow scheduler

6.启动web服务

#启动web进程
airflow webserver -p 8080

#关闭CentOS6的防火墙
sudo service iptables stop

#关闭CentOS6的SELinux
setenforce 0

#关闭CentOS7的防火墙
systemctl stop firewalld.service

#禁止firewall开机启动
systemctl disable firewalld.service

Celery+MySQL

#Celery文档 http://docs.jinkan.org/docs/celery/index.html
#Celery4.0.0在airflow中有一些问题，所以安装Celery3
pip install -U Celery==3.1.24
pip install airflow[celery]

修改配置文件

vi airflow.cfg

[core]
executor = CeleryExecutor

[celery]
broker_url = sqla+mysql://airflow:airflow@localhost:3306/airflow

celery_result_backend = db+mysql://airflow:airflow@localhost:3306/airflow

启动airflow

airflow webserver -p 8080

airflow scheduler

#以非root用户运行
airflow worker

#启动Celery WebUI 查看celery任务
airflow flower 
http://localhost:5555/

Celery+RabbitMQ

wget http://www.rabbitmq.com/releases/rabbitmq-server/v3.6.5/rabbitmq-server-3.6.5-1.noarch.rpm

#安装RabbitMQ的依赖包
yum install erlang

yum install socat

#如果下载了rabbitmq的yum源 sudo yum install -y rabbitmq-server
rpm -ivh rabbitmq-server-3.6.5-1.noarch.rpm

启动RabbitMQ服务

#启动rabbitmq服务
sudo service rabbitmq-server start 
#或者
sudo rabbitmq-server

#添加 -detached 属性来让它在后台运行（注意：只有一个破折号）
sudo rabbitmq-server -detached

#设置开机启动rabbitmq服务
chkconfig rabbitmq-server on

#永远不要用 kill 停止 RabbitMQ 服务器，而是应该用 rabbitmqctl 命令
sudo rabbitmqctl stop

设置RabbitMQ

#创建一个RabbitMQ用户
rabbitmqctl add_user airflow airflow

#创建一个RabbitMQ虚拟主机
rabbitmqctl add_vhost vairflow

#将这个用户赋予admin的角色
rabbitmqctl set_user_tags airflow admin

#允许这个用户访问这个虚拟主机
rabbitmqctl set_permissions -p vairflow airflow ".*" ".*" ".*"

# no usage
rabbitmq-plugins enable rabbitmq_management 

修改airflow配置文件支持Celery

vi $AIRFLOW_HOME/airflow/airflow.cfg

#更改Executor为CeleryExecutor
executor = CeleryExecutor

#更改broker_url
broker_url = amqp://airflow:airflow@localhost:5672/vairflow
Format explanation: transport://userid:password@hostname:port/virtual_host

#更改celery_result_backend
celery_result_backend = amqp://airflow:airflow@localhost:5672/vairflow
Format explanation: transport://userid:password@hostname:port/virtual_host

安装airflow的celery和rabbitmq模块

pip install airflow[celery]
pip install airflow[rabbitmq]

airflow使用DAG(Directed Acyclic Graph,有向无环图为)来管理作业流的

#创建DAG
from datetime import datetime, timedelta
from airflow.models import DAG
args = {
    'owner': 'airflow',
    'start_date': seven_days_ago,
    'email': ['airflow@airflow.com'],
    'email_on_failure': True,
    'email_on_retry': True,
    'retries': 3,
    'retries_delay': timedelta(seconds=60),
    'depends_on_past': True
}

dag = DAG(
    dag_id='dag',
    default_args=args,
    schedule_interval='0 0 * * *',
    dagrun_timeout=timedelta(minutes=60)
)

创建任务将任务添加到DAG中

from airflow.operators.bash_operator import BashOperator
from airflow.operators.dummy_operator import DummyOperator

demo = DummyOperator(
    task_id='demo',
    dag=dag
)

last_execute = BashOperator(
    task_id='last_execute',
    bash_command='echo 1',
    dag=dag
)

配置任务的依赖关系

demo.set_downstream(last_execute)