随着公司应用需求的增加，需要不断的扩展，服务器数量也随之增加，当服务器数量不断增加，我们会发现一台puppetmaster压力大，解析缓慢，而且时不时出“Timeout”之类的报错，那这时有什么优化的办法吗？我们在Puppet官网上找寻解决方案，发现puppetmaster可以配置多端口，结合WEB代理（推荐轻量级的负载均衡器Nginx），这样puppetmaster承受能力至少可以提升数倍以上，相当于在很大程度上优化了puppet的处理能力。

1.遵循前面的环境设定，我们这里的服务器环境及软件版本分别为:

服务器系统：CentOS5.8 x86_64

Ruby版本：ruby-1.8.5

Puppet版本：puppet-2.7.9

Nginx版本：nginx-0.8.46

2.Mongrel安装

要使用puppet多端口配置，需要指定mongrel类型，默认没有安装，需要安装：

yum install -yrubygem-mongrel

3.配置puppetmaster

在/etc/sysconfig/puppetmaster文件末尾添加如下两行，分别代表多端口、mongrel类型，内容如下所示：

PUPPETMASTER_PORTS=(8141 8142 8143 81448145)
PUPPETMASTER_EXTRA_OPTS="--servertype=mongrel--ssl_client_header=HTTP_X_SSL_SUBJECT"

4.安装Nginx服务

安装之前请确保系统已经安装pcre-devel正则库，然后再编译安装Nginx，需要添加SSL模块参数支持，Nginx的安装过程如下所示：

yum -y install pcre-devel
cd /usr/local/src
wgethttp://nginx.org/download/nginx-0.8.46.tar.gz
tar zxvf nginx-0.8.46.tar.gz
cd nginx-0.8.46
./configure --prefix=/usr/local/nginx--with-http_ssl_module
make && make install && cd../

添加www用户组及用户，命令如下所示：

groupadd www
useradd -g www www

5.我们依据puppet需求来修改配置文件nginx.conf，内容如下所示：

user www;
worker_processes    8;
events {
       worker_connections    65535;
}
http {
       include             mime.types;
       default_type   application/octet-stream;
       sendfile                on;
       tcp_nopush         on;
       keepalive_timeout    65;
       #定义puppet客户端访问puppet-server端日志格式
       log_format main '$remote_addr - $remote_user [$time_local]"$request" $request_length $request_time $time_local'
                                                '$status $body_bytes_sent$bytes_sent $connection $msec "$http_referer" '
                                               '"$http_user_agent" $http_x_forwarded_for$upstream_response_time $upstream_addr $upstream_status ';
       access_log   /usr/local/nginx/logs/access.log   main;
       upstream puppetmaster {
                server 127.0.0.1:8141;
                server 127.0.0.1:8142;
                server 127.0.0.1:8143;
                server 127.0.0.1:8144;
                server 127.0.0.1:8145;
                }
       server {
       listen 8140;
       root /etc/puppet;
       ssl on;
       ssl_session_timeout 5m;
       #如下为puppetmaster服务器端证书地址
       ssl_certificate /var/lib/puppet/ssl/certs/server.cn7788.com.pem;
       ssl_certificate_key/var/lib/puppet/ssl/private_keys/server.cn7788.com.pem;
       ssl_client_certificate /var/lib/puppet/ssl/ca/ca_crt.pem;
       ssl_crl /var/lib/puppet/ssl/ca/ca_crl.pem;
       ssl_verify_client optional;
       #File sections
       location /production/file_content/files/ {
       types { }
       default_type application/x-raw;
       #定义puppet推送路径别名
       alias /etc/puppet/files/;
       }
       # Modules files sections
       location ~ /production/file_content/modules/.+/ {
       root /etc/puppet/modules;
       types { }
       default_type application/x-raw;
       rewrite ^/production/file_content/modules/(.+)/(.+)$ /$1/files/$2 break;
       }
       location / {
       ##设置跳转到puppetmaster负载均衡
       proxy_pass http://puppetmaster;
       proxy_redirect off;
       proxy_set_header Host $host;
       proxy_set_header X-Real-IP $remote_addr;
       proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
       proxy_set_header X-Client-Verify $ssl_client_verify;
       proxy_set_header X-SSL-Subject $ssl_client_s_dn;
       proxy_set_header X-SSL-Issuer $ssl_client_i_dn;
       proxy_buffer_size 10m;
       proxy_buffers 1024 10m;
       proxy_busy_buffers_size 10m;
       proxy_temp_file_write_size 10m;
       proxy_read_timeout 120;
    }
}
}

6.修改完nginx.conf文件以后，我们要启动nginx及puppet-server，这时应该如何操作呢？

我们首先关闭puppetmaster进程，然后先启动nginx，不然nginx是会启动失败的，命令如下所示：

/usr/local/nginx/sbin/nginx

nginx占用puppetmaster默认的8140端口后，我们可以用如下命令来检查8140端口是否被nginx接管，如下所示：

lsof -i:8140

此命令显示结果表明8140被nginx进程接管，如下所示：

COMMAND   PID    USER     FD    TYPE DEVICE SIZE/OFF NODE NAME
nginx    4121    root        6u   IPv4    20668            0t0    TCP *:8140 (LISTEN)
nginx    4122    www    6u   IPv4    20668            0t0    TCP *:8140 (LISTEN)

我们再启动puppetmaster，命令如下所示：

service puppetmaster start

如果ruby版本为1.8.5的话，等会运行puppetmaster会有如下警告，如下所示：

Starting puppetmaster:
Port: 8141** Ruby version is not up-to-date;loading cgi_multipart_eof_fix
                                                                                                                                                                                                                                        [        OK        ]
Port: 8142** Ruby version is notup-to-date; loading cgi_multipart_eof_fix
                                                                                                                                                                                                                                      [        OK        ]
Port: 8143** Ruby version is notup-to-date; loading cgi_multipart_eof_fix
                                                                                                                                                                                                                                        [        OK        ]
Port: 8144** Ruby version is notup-to-date; loading cgi_multipart_eof_fix
                                                                                                                                                                                                                                       [        OK        ]
Port: 8145** Ruby version is notup-to-date; loading cgi_multipart_eof_fix
                                                                                                                                                                                                                                      [        OK        ]

这段警告值的意思为

It's just a warning. Mongrel wants a Rubyversion of at least 1.8.6.
But it still runs just fine with previousversions. Just ignore the warning.

翻译为中文的意思是：

Mongrel需要ruby至少是1.8.6以上的版本，但它仍然在当前版本运行，请忽咯当前警告，为了保证整个puppet运行环境的稳定，我这里选择还是沿用1.8.5版本的ruby，大家有任何疑问，欢迎随时通过此技术博客或电子邮箱[email protected]与抚琴煮酒进行交流。