Mastering Fabric: Automate Server Management and Deployments with Python

What is Fabric

Fabric is a Python‑based automation script tool for system management and application deployment that conveniently supports multiple servers. A single script file can execute the same or different tasks on many servers.

It provides a concise command set to perform common operations such as running shell commands locally or on remote servers, uploading/downloading files, and prompting user input for interactive operations. Advanced features include server grouping, parallel task execution, and error handling.

Use Cases

Instagram, a large‑scale image social app owned by Facebook, operates thousands of servers and performs more than 30 code deployments per day. Before adopting Facebook’s distributed deployment system, Instagram used Fabric for massive server management and application deployment, demonstrating Fabric’s powerful capabilities.

Usage Examples

(1) Hello World

Create a script file fabfile.py:

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

Run the task: $ fab hello You can rename the script and specify it explicitly:

$ mv fabfile.py hello.py
$ fab -f hello.py hello

(2) Passing Parameters

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

Execute with parameters: $ fab hello:name=hi,val=world (3) Executing Local Commands

from fabric.api import local, lcd

def ls():
    with lcd('~'):
        local('ls')

(4) Mixed Local and Remote Execution

from fabric.api import local, cd, run, env

env.hosts = ['[email protected]:28']
env.password = 'password'

def do_local():
    local('echo "in local"')

def do_remote():
    print "in remote"
    with cd('~'):
        run('ls -l')

def opt():
    do_local()
    do_remote()

Run opt to see both local and remote results.

(5) Different Tasks on Different Servers

#!/usr/bin/env python
# encoding: utf-8
from fabric.api import *

env.roledefs = {
    'testserver': ['[email protected]:28'],
    'realserver': ['[email protected]:22'],
}

env.passwords = {
    '[email protected]:28': "password",
    '[email protected]:22': "password",
}

@roles('testserver')
def task1():
    run('ls /home')

@roles('realserver')
def task2():
    run('ls /home')

def dotask():
    execute(task1)
    execute(task2)

Running dotask executes task1 on the test server and task2 on the real server.

Important Feature: Parallel Task Execution

Parallel execution is especially powerful when managing many servers, saving considerable time. By default tasks run sequentially. To run tasks in parallel, use the -P flag or the @parallel decorator.

Example of running two tasks on three hosts sequentially:

$ fab -H web1,web2,web3 update reload

Result:

update on web1
update on web2
update on web3
reload on web1
reload on web2
reload on web3

Enable parallel execution:

$ fab -H web1,web2,web3 -P update reload

Result:

update on web1, web2, and web3
reload on web1, web2, and web3

Alternatively, annotate tasks:

@parallel
def runs_in_parallel():
    # ...

@serial
def runs_serially():
    # ...

Execute both:

$ fab -H host1,host2,host3 runs_in_parallel runs_serially

Result shows the parallel task runs on all hosts simultaneously, while the serial task runs on each host one after another.

Installation

Fabric requires a Python environment. Install it via the system package manager:

Ubuntu: $ sudo apt-get install fabric CentOS: $ yum install fabric Official website: http://www.fabfile.org