3.7.2 线上环境中的Fabric应用实例
笔者线上的核心业务机器统一都是AWS EC2主机，机器数量较多，每个数据中心都部署了Fabric跳板机（物理拓扑图可参考图3-3），系统为Amazon Linux，内核版本为3.14.34-27.48.amzn1.x86_64，Python版本为Python 2.6.9。
如果公司项目组核心开发人员离职，线上机器就都要更改密钥，由于密钥一般是以组的形式存在的，再加上机器数量繁多，因此单纯通过技术人员手工操作，基本上是一项不可能完成的任务，但若是通过Fabric自动化运维工具的话，这就是一项简单的工作了，由于现在的线上服务器多采用SSH Key的方式管理，所以对于大多数系统运维人员来说SSH Key分发也是工作内容之一，故而建议大家掌握此脚本的用法。示例脚本内容如下：
#!/usr/bin/python2.6
# -*- coding:
utf-8 -*-
from fabric.api
import *
from
fabric.colors import *
from
fabric.context_managers import *
#这里为了简化工作，脚本采用纯Python的写法，没有采用Fabric的@task修饰器
env.user =
'ec2-user'
env.key_filename
= '/home/ec2-user/.ssh/id_rsa'
hosts=['budget','adserver','bidder1','bidder2','bidder3','bidder4','bidder5','bidder6','bidder7','bidder8','bidder9',redis1','redis2','redis3','redis4','redis5','redis6']
#机器数量众多，这里只罗列了部分
def
put_ec2_key():
with settings(warn_only=False):
put("/home/ec2-user/admin-master.pub","/home/ec2-user/admin-master.pub")
sudo("\cp
/home/ec2-user/admin-master.pub /home/ec2-user/.ssh/authorized_keys")
#\cp的作用是取消其别名作用，即不让cp-i生效
sudo("chmod 600
/home/ec2-user/.ssh/authorized_keys")
def
put_admin_key():
with settings(warn_only=False):
put("/home/ec2-user/admin-operation.pub",
"/home/ec2-user/admin-operation.pub")
sudo("\cp /home/ec2-user/admin-operation.pub /home/admin/.ssh/authorized_keys")
sudo("chown admin:admin
/home/admin/.ssh/authorized_keys")
sudo("chmod 600
/home/admin/.ssh/authorized_keys")
def
put_readonly_key():
with settings(warn_only=False):
put("/home/ec2-user/admin-readonly.pub",
"/home/ec2-user/admin-readonly.pub")
sudo("\cp
/home/ec2-user/admin-readonly.pub /home/readonly/.ssh/authorized_keys")
sudo("chown readonly:readonly
/home/readonly/.ssh/authorized_keys")
sudo("chmod 600
/home/readonly/.ssh/authorized_keys")
for host in
hosts:
env.host_string = host
put_ec2_key()
put_admin_key()
put_readonly_key()
大家可以输入如下命令查看系统中定义的别名（CentOS 6.4 x86_64）。
alias
命令显示结果如下所示：
alias cp='cp -i'
alias l.='ls -d
.* --color=auto'
alias ll='ls -l
--color=auto'
alias ls='ls
--color=auto'
alias mv='mv -i'
alias rm='rm -i'
alias
which='alias | /usr/bin/which --tty-only --read-alias --show-dot --show-tilde'
Amazon Linux系统与CentOS 6.4略有差别，已经取消了cp的别名定义。
如果线上的Nagios 客户端的监控脚本因为业务需求又发生了改动，而bidder业务集群约有23台（下面只列出了其中10台），且其中的一个业务需求脚本前前后后改动了4次，这时，手动操作肯定会耗费大量人力及时间成本，因此这里用Fabric推送此脚本并执行，代码如下：
#!/usr/bin/python2.6
## -*- coding:
utf-8 -*-
from fabric.api
import *
from
fabric.colors import *
from
fabric.context_managers import *
user =
'ec2-user'
hosts=['bidder1','bidder2','bidder3','bidder4','bidder5','bidder6','bidder7','bidder8','bidder9','bidder10']
#机器数量比较多，这里只列出其中10台
@task
#这里用到了@task修饰器
def put_task():
print yellow("Put Local File to Nagios
Client")
with settings(warn_only=True):
put("/home/ec2-user/check_cpu_utili.sh",
"/home/ec2-user/check_cpu_utili.sh")
sudo("cp
/home/ec2-user/check_cpu_utili.sh /usr/local/nagios/libexec")
sudo("chown nagios:nagios
/usr/local/nagios/libexec/check_cpu_utili.sh")
sudo("chmod +x
/usr/local/nagios/libexec/check_cpu_utili")
sudo("kill `ps aux | grep nrpe | head -n1 | awk '{print
$2}' `")
sudo("/usr/local/nagios/bin/nrpe
-c /usr/local/nagios/etc/nrpe.cfg -d")
print green("upload File success
and restart nagios service!")
#这里以绿色字体打印结果是为了方便查看脚本执行结果
for host in
hosts:
env.host_string = host
put_task()
执行上面的脚本以后，Fabric也会返回清晰的显示结果，大家可以根据显示结果得知哪些机器已经成功运行，哪些机器失败，非常直观，结果如下所示：
Put Local File
to remote
[bidder1] put:
/home/ec2-user/check_cpu_utili.sh -> /home/ec2-user/check_cpu_utili.sh
[bidder1] sudo:
cp /home/ec2-user/check_cpu_utili.sh
/usr/local/nagios/libexec/check_cpu_utili.sh
[bidder1] sudo:
chown nagios:nagios /usr/local/nagios/libexec/check_cpu_utili.sh
[bidder1] sudo:
chmod +x /usr/local/nagios/libexec/check_cpu_utili.sh
[bidder1] sudo:
kill `ps aux | grep nrpe | head -n1 | awk '{print $2}' `
[bidder1] sudo:
/usr/local/nagios/bin/nrpe -c /usr/local/nagios/etc/nrpe.cfg -d
upload File
success and restart nagios service!
Put Local File
to remote
[bidder2] put:
/home/ec2-user/check_cpu_utili.sh -> /home/ec2-user/check_cpu_utili.sh
[bidder2] sudo:
cp /home/ec2-user/check_cpu_utili.sh
/usr/local/nagios/libexec/check_cpu_utili.sh
[bidder2] sudo:
chown nagios:nagios /usr/local/nagios/libexec/check_cpu_utili.sh
[bidder2] sudo:
chmod +x /usr/local/nagios/libexec/check_cpu_utili.sh
[bidder2] sudo:
kill `ps aux | grep nrpe | head -n1 | awk '{print $2}' `
[bidder2] sudo:
/usr/local/nagios/bin/nrpe -c /usr/local/nagios/etc/nrpe.cfg -d
upload File
success and restart nagios service!
Put Local File
to remote
[bidder3] put:
/home/ec2-user/check_cpu_utili.sh -> /home/ec2-user/check_cpu_utili.sh
[bidder3] sudo:
cp /home/ec2-user/check_cpu_utili.sh
/usr/local/nagios/libexec/check_cpu_utili.sh
[bidder3] sudo:
chown nagios:nagios /usr/local/nagios/libexec/check_cpu_utili.sh
[bidder3] sudo:
chmod +x /usr/local/nagios/libexec/check_cpu_utili.sh
[bidder3] sudo:
kill `ps aux | grep nrpe | head -n1 | awk '{print $2}' `
[bidder3] sudo:
/usr/local/nagios/bin/nrpe -c /usr/local/nagios/etc/nrpe.cfg -d
upload File
success and restart nagios service!
Put Local File
to remote
[bidder4] put:
/home/ec2-user/check_cpu_utili.sh -> /home/ec2-user/check_cpu_utili.sh
[bidder4] sudo:
cp /home/ec2-user/check_cpu_utili.sh
/usr/local/nagios/libexec/check_cpu_utili.sh
[bidder4] sudo:
chown nagios:nagios /usr/local/nagios/libexec/check_cpu_utili.sh
[bidder4] sudo:
chmod +x /usr/local/nagios/libexec/check_cpu_utili.sh
[bidder4] sudo:
kill `ps aux | grep nrpe | head -n1 | awk '{print $2}' `
[bidder4] sudo:
/usr/local/nagios/bin/nrpe -c /usr/local/nagios/etc/nrpe.cfg -d
upload File
success and restart nagios service!
Put Local File
to remote
[bidder5] put:
/home/ec2-user/check_cpu_utili.sh -> /home/ec2-user/check_cpu_utili.sh
[bidder5] sudo:
cp /home/ec2-user/check_cpu_utili.sh
/usr/local/nagios/libexec/check_cpu_utili.sh
[bidder5] sudo:
chown nagios:nagios /usr/local/nagios/libexec/check_cpu_utili.sh
[bidder5] sudo:
chmod +x /usr/local/nagios/libexec/check_cpu_utili.sh
[bidder5] sudo:
kill `ps aux | grep nrpe | head -n1 | awk '{print $2}' `
[bidder5] sudo:
/usr/local/nagios/bin/nrpe -c /usr/local/nagios/etc/nrpe.cfg -d
upload File
success and restart nagios service!
大家可以看到，短短几行代码就达到了自动化运维的效果，而且跟Fabric相关的代码都是纯Python代码和Shell代码，开发人员和运维人员很容易上手，在公司里推广应用，大家的认可程度也高。事实上，通过上面的举例大家应该能发现，Fabric特别适合于需要重复执行大量Shell命令的工作场景。