python fabric implements remote operation and deployment

fabric

The title is development, but at the same time, I have to do development, testing and operation and maintenance... Why is the task*3 not salary * 3 (o(╯□╰)o)

Recently, I have taken over more and more things. The release and operation and maintenance work is quite mechanical, and the frequency is quite high, which leads to a waste of time but has many advantages.

Fix bugs, test, submit the repository (2 minutes), ssh to the test environment for pull deployment (2 minutes), rsync to online machines A, B, C, D, E (1 minute) , ssh to the five machines of ABCDE respectively, and restart one by one (8-10 minutes) = 13-15 minutes

The frustrating thing is that every operation is the same, the command is the same, and the terrible thing is that in multiple On the machine, it is difficult to do it with a script on this machine. The main time is wasted on ssh, typing commands, and writing it as a script. It can be executed with one click. Take two minutes to look at the execution results

until I find out. Fabric

function

is a very powerful tool

can solidify the commands for automated deployment or multi-machine operation into a script

and some The operation and maintenance tools are very similar. The main reason for using it is that Python...

is simple, easy to use and easy to use

Of course, you can also combine various shell commands, including ancient artifacts and modern weapons. Difference

Environment configuration

Install the corresponding package on the local machine and the target machine (note, both are required)

sudo easy_install fabric

It is currently version 1.6 (or use pip install, the same )

After the installation is completed, you can check whether the installation is successful

[ken@~$] which fab
/usr/local/bin/fab

Then, you can do it

hello world

First perform simple operations on the machine, there are A preliminary understanding, the source of the example is from the official website

Create a new py script: fabfile.py

def hello():
    print("Hello world!")

Command line execution:

[ken@~/tmp/fab$] fab hello
Hello world!
Done.

Note that fabfile does not need to be used as the file name here. However, you need to specify the file when executing

[ken@~/tmp/fab$] mv fabfile.py test.py
fabfile.py -> test.py
[ken@~/tmp/fab$] fab hello
Fatal error: Couldn't find any fabfiles!
Remember that -f can be used to specify fabfile path, and use -h for help.
Aborting.
[ken@~/tmp/fab$] fab -f test.py hello
Hello world!
Done.

With parameters:

Modify the fabfile.py script:

def hello(name, value):
    print("%s = %s!" % (name, value))

Execute

[ken@~/tmp/fab$] fab hello:name=age,value=20
age = 20!
Done.
[ken@~/tmp/fab$] fab hello:age,20
age = 20!
Done.

Execute local operation

Simple local operation:

from fabric.api import local
def lsfab():
    local('cd ~/tmp/fab')
    local('ls')

Result:

[ken@~/tmp/fab$] pwd;ls
/Users/ken/tmp/fab
fabfile.py   fabfile.pyc  test.py      test.pyc
[ken@~/tmp/fab$] fab -f test.py lsfab
[localhost] local: cd ~/tmp/fab
[localhost] local: ls
fabfile.py  fabfile.pyc test.py     test.pyc
Done.

Start of actual combat:

Assume that you want to submit a configuration file settings.py to the repository every day (Conflicts are not considered here)

If it is a manual operation:

cd /home/project/test/conf/
git add settings.py
git commit -m 'daily update settings.py'
git pull origin
git push origin

In other words, you have to manually type these commands once a day. The so-called daily job means that you have to type these commands every day. Let’s see how to use fabric to accomplish repetitive, mechanized work with one click: (In fact, it can be done directly with shell scripts, but the advantage of fab is not here. The main thing here is to prepare for local + remote operations later. After all, both Write a script for the operation in each place to facilitate maintenance)

from fabric.api import local
def setting_ci():
    local("cd /home/project/test/conf/")
    local("git add settings.py")
    #后面你懂的，懒得敲了….

Mixing and integrating remote operations

At this time, suppose you want to go to the project directory corresponding to /home/ken/project of machine A. Update the configuration file

#!/usr/bin/env python
# encoding: utf-8
from fabric.api import local,cd,run
env.hosts=['user@ip:port',] #ssh要用到的参数
env.password = 'pwd'
def setting_ci():
    local('echo "add and commit settings in local"')
    #刚才的操作换到这里，你懂的
def update_setting_remote():
    print "remote update"
    with cd('~/temp'):   #cd用于进入某个目录
        run('ls -l | wc -l')  #远程操作用run
def update():
    setting_ci()
    update_setting_remote()

Then execute it:

[ken@~/tmp/fab$] fab -f deploy.py update
[user@ip:port] Executing task 'update'
[localhost] local: echo "add and commit settings in local"
add and commit settings in local
remote update
[user@ip:port] run: ls -l | wc -l
[user@ip:port] out: 12
[user@ip:port] out:
Done.

Note that if env.password is not declared, an interaction asking for a password will pop up when executing to the corresponding machine

多服务器混搭

操作多个服务器，需要配置多个host

#!/usr/bin/env python
# encoding: utf-8
from fabric.api import *
#操作一致的服务器可以放在一组，同一组的执行同一套操作
env.roledefs = {
            'testserver': ['user1@host1:port1',],  
            'realserver': ['user2@host2:port2', ]
            }
#env.password = '这里不要用这种配置了，不可能要求密码都一致的，明文编写也不合适。打通所有ssh就行了'
@roles('testserver')
def task1():
    run('ls -l | wc -l')
@roles('realserver')
def task2():
    run('ls ~/temp/ | wc -l')
def dotask():
    execute(task1)
    execute(task2)

结果：

[ken@~/tmp/fab$] fab -f mult.py dotask
[user1@host1:port1] Executing task 'task1'
[user1@host1:port1] run: ls -l | wc -l
[user1@host1:port1] out: 9
[user1@host1:port1] out:
[user2@host2:port2] Executing task 'task2'
[user2@host2:port2] run: ls ~/temp/ | wc -l
[user2@host2:port2] out: 11
[user2@host2:port2] out:
Done.

扩展

1.颜色

可以打印颜色，在查看操作结果信息的时候更为醒目和方便

from fabric.colors import *
def show():
    print green('success')
    print red('fail')
    print yellow('yellow')
#fab -f color.py show

2.错误和异常

关于错误处理

默认，一组命令，上一个命令执行失败后，不会接着往下执行

失败后也可以进行不一样的处理， 文档

目前没用到，后续用到再看了

3.密码管理

看文档

更好的密码管理方式，哥比较土，没打通，主要是服务器列表变化频繁，我的处理方式是：

1.host,user,port,password配置列表，所有的都写在一个文件

或者直接搞到脚本里，当然这个更........

env.hosts = [

'host1',

'host2'
]
env.passwords = { 
'host1': "pwdofhost1",
'host2': "pwdofhost2",
}

 或者

env.roledefs = {
'testserver': ['host1', 'host2'],
'realserver': ['host3', ]
}
env.passwords = {
'host1': "pwdofhost1",
'host2': "pwdofhost2",
'host3': "pwdofhost3", 
} 

2.根据key解析成map嵌套，放到deploy中

另外命令其实也可以固化成一个cmds列表的…..

The above is the detailed content of python fabric implements remote operation and deployment. For more information, please follow other related articles on the PHP Chinese website!