CentOS 6.3安装及配置RabbitMQ cluster

Why not ZeroMQ

ZeroMQ更快不是吗？是的，RabbitMQ更丰富。看场景了，还是那句话，没有最好的，只有最合适的。

Erlang编译安装

参考文档：http://www.erlang.org/doc/installation_guide/INSTALL.html#How-to-Build-and-Install-ErlangOTP

"By default, Erlang/OTP will be installed in /usr/local/{bin,lib/erlang}. To instead install in <BaseDir>/{bin,lib/erlang}, use the --prefix=<BaseDir> option."  注：编译安装的目录结构如上描述。yum安装R15B03-1这个版本默认安装到了/usr/lib/erlang，而不是/usr/local/lib/erlang. yum安装简单： 1.wget -O /etc/yum.repos.d/erlang_solutions.repo http://binaries.erlang-solutions.com/rpm/centos/erlang_solutions.repo 2. yum install esl-erlang 下文为编译安装过程：

[root@cx-t1 ~] # cd /usr/local/src/ [root@cx-t1 src] # wget http://www.erlang.org/download/otp_src_R15B03-1.tar.gz [root@cx-t1 src] # tar -xzf ./otp_src_R15B03-1.tar.gz ; cd otp_src_R15B03 [root@cx-t1 src] # ./configure [root@cx-t1 src] # make [root@cx-t1 src] # make install

 

RabbitMQ编译安装

• 准备工作：修改机器名及安装依赖包

[root@cx-t1 ~] # hostname rabbitmq-1 [root@cx-t1 ~] # vi /etc/sysconfig/network HOSTNAME= cx-t1 [root@cx-t1 ~] # vi /etc/hosts # 有内部DNS的省事多了，墙裂推荐搞个内部DNS，以FQDN的形式保存主机名（别学我） 192.168.7.101     cx-t1 192.168.7.102     cx-t2 [root@cx-t2 ~] # yum install xmlto

• 编译安装

[root@cx-t1 ~] # cd /usr/local/src/ [root@cx-t1 src] # wget http://www.rabbitmq.com/releases/rabbitmq-server/v3.0.1/rabbitmq-server-3.0.1.tar.gz [root@cx-t1 src] # tar -xzf ./rabbitmq-server-3.0.1.tar.gz; cd rabbitmq-server-3.0.1 [root@cx-t1 rabbitmq-server-3.0.1] # make [root@cx-t1 rabbitmq-server-3.0.1] # make TARGET_DIR=/usr/local/rabbitmq-3.0.1 SBIN_DIR=/usr/local/rabbitmq-3.0.1/sbin MAN_DIR=/usr/local/rabbitmq-3.0.1/man install [root@cx-t1 rabbitmq-server-3.0.1] # vi /usr/local/rabbitmq-3.0.1/sbin/rabbitmq-defaults CONFIG_FILE= /data/rabbitmq/rabbitmq LOG_BASE= /data/log/rabbitmq MNESIA_BASE= /data/rabbitmq/mnesia ENABLED_PLUGINS_FILE= /data/rabbitmq/enabled_plugins CONF_ENV_FILE= /data/rabbitmq/rabbitmq-env .conf [root@cx-t1 rabbitmq-server-3.0.1] # mkdir -pv /data/rabbitmq/mnesia /data/log/rabbitmq [root@cx-t1 rabbitmq-server-3.0.1] # groupadd -g 5672 rabbitmq ; useradd -u 5672 -g 5672 -d /home/rabbitmq -m rabbitmq [root@cx-t1 rabbitmq-server-3.0.1] # chown -R rabbitmq:rabbitmq /data/log/rabbitmq /data/rabbitmq [root@cx-t1 rabbitmq-server-3.0.1] # ln -s /usr/local/rabbitmq-3.0.1 /usr/local/rabbitmq [root@cx-t1 rabbitmq-server-3.0.1] # vi /etc/profile export PATH= /usr/local/rabbitmq/sbin :$PATH [root@cx-t1 rabbitmq-server-3.0.1] # source /etc/profile [root@cx-t1 rabbitmq-server-3.0.1] # rabbitmqctl --help

通过查看安装包中的MakeFile可以发现几个配置参数TARGET_DIR、SBIN_DIR和MAN_DIR需要在make时指定，

而RABBITMQ_NODENAME、RABBITMQ_MNESIA_DIR和RABBITMQ_LOG_BASE等即可在make时指定亦可在install后修改sbin/rabbitmq-defaults这个文件。

• 配置Erlang cookie，激活Management Plugin

  more: http://www.rabbitmq.com/management.html

[root@cx-t1 ~] # echo "your cookie, as complex as possible" > /home/rabbitmq/.erlang.cookie [root@cx-t1 rabbitmq-server-3.0.1] # su - rabbitmq [rabbitmq@cx-t1 ~]$ rabbitmq-plugins enable rabbitmq_management

• 启动服务

[rabbitmq@cx-t1 ~]$ rabbitmq-server -detached [rabbitmq@cx-t1 ~]$ rabbitmqctl list_users

配置RabbitMQ Clustering

Clustering配置有两种方式：

• 使用rabbitmqctl

  这种方式轻便简洁，可以动态增减node（推荐）。

  下面这个脚本例子就是将node rabbit@cx-t2加入到rabbit@cx-t1这个node。这种方式即便rabbit@cx-t2中有数据也会被rabbit@cx-t1所覆盖，相当于rabbit@cx-t2自动reset了。

  [rabbitmq@cx-t2 ~]$ rabbitmqctl stop_app Stopping node 'rabbit@cx-t2' ... ... done . [rabbitmq@cx-t2 ~]$ rabbitmqctl join_cluster rabbit@cx-t1 Clustering node 'rabbit@cx-t2' with 'rabbit@cx-t1' ... ... done . [rabbitmq@cx-t2 ~]$ rabbitmqctl start_app Starting node 'rabbit@cx-t2' ... ... done . [rabbitmq@cx-t2 ~]$ rabbitmqctl cluster_status Cluster status of node 'rabbit@cx-t2' ... [{nodes,[{disc,[ 'rabbit@cx-t1' , 'rabbit@cx-t2' ]}]},   {running_nodes,[ 'rabbit@cx-t1' , 'rabbit@cx-t2' ]},   {partitions,[]}] ... done .

• 修改rabbitmq.config这个配置文件
  必须是fresh node——即mnesia数据库是空的，才可以正常启动，否则需要执行如下命令后rabbitmqctl reset才可以正常启动加入集群。

  rabbitmqctl stop_app rabbitmqctl reset rabbitmqctl start_app

At least one disk node should be running at all times to prevent data loss. RabbitMQ will prevent the creation of a RAM-only cluster in many situations, but it still won't stop you from stopping and forcefully resetting all the disc nodes, which will lead to a RAM-only cluster. Doing this is not advisable and makes losing data very easy. 集群中至少保证有一个disk node。实际上，两个disk node加若干个RAM node既能保证性能又能避免数据丢失(容灾)。 When the entire cluster is brought down, the last node to go down must be the first node to be brought online. If this doesn't happen, the nodes will wait 30 seconds for the last disc node to come back online, and fail afterwards. If the last node to go offline cannot be brought back up, it can be removed from the cluster using the forget_cluster_node command - consult the rabbitmqctl manpage for more information. 当集群中的node都stop后，再次启动时一定要第一个启动最后一个被stop的node。 more: http://www.rabbitmq.com/clustering.html

 

总结：

• RabbitMQ安装步骤在每个机器上都是一样的，management plugin只需在一个机器上激活即可。
• 一定要确保每个机器上的erlang cookie都是一样的，这样node间才可以互相访问。
• 服务一定要用rabbitmq这个用户启动，不要使用root。
• rabbitmq默认的端口为5672，management plugin的默认端口为15672。
• 首次登录管理界面(http://yourhost:15672)用guest这个账号，密码为guest。登录后修改guest密码或新建一个管理员账号再删除guest