[ Openstack ] Openstack-Mitaka 高可用之 Rabbitmq-server 集群部署

环境介绍

openstack内部是遵循AMQP（高级消息队列协议）的基础上采用消息队列进行通信。 nova对请求应答进行异步调用，当请求接收后立即出发一个回调。由于使用了异步通信，不会有用户的动作被长置于等待状态。异步使整个系统变得更加高效。

集群搭建

以下操作三个节点都需要执行：

安装rabbitmq

# yum install rabbitmq-server  -y

配置监听地址（每个controller节点都需要配置本地监听地址）：

[root@controller1 ~]# vim /etc/rabbitmq/rabbitmq-env.conf        # 在RHEL7.2系统上默认是不存在该文件的
RABBITMQ_NODE_IP_ADDRESS=192.168.0.11
RABBITMQ_NODE_PORT=

只启动第一个节点：

[root@controller1 ~]# systemctl start rabbitmq-server

拷贝.erlang.cookie文件到controller2 controller3，注意该文件的权限是400

[root@controller1 ~]# scp /var/lib/rabbitmq/.erlang.cookie controller2:/var/lib/rabbitmq/
.erlang.cookie                                                                                                                    %        .0KB/s   :
[root@controller1 ~]# scp /var/lib/rabbitmq/.erlang.cookie controller3:/var/lib/rabbitmq/
.erlang.cookie                                                                                                                    %        .0KB/s   :
 
[root@controller1 ~]# ll /var/lib/rabbitmq/.erlang.cookie
-r--------  rabbitmq rabbitmq  Nov  : /var/lib/rabbitmq/.erlang.cookie
[root@controller2 ~]# chown -R rabbitmq:rabbitmq /var/lib/rabbitmq/.erlang.cookie
[root@controller3 ~]# chown -R rabbitmq:rabbitmq /var/lib/rabbitmq/.erlang.cookie

启动controller2、controller3 rabbitmq服务

# systemctl start rabbitmq-server

将controller2、controller3 与controller1组成集群

Controller2:

[root@controller2 ~]# rabbitmqctl stop_app
Stopping node rabbit@controller2 ...
[root@controller2 ~]# rabbitmqctl join_cluster rabbit@controller1
Clustering node rabbit@controller2 with rabbit@controller1 ...
[root@controller2 ~]# rabbitmqctl start_app
Starting node rabbit@controller2 ...

Controller3:

[root@controller3 ~]# rabbitmqctl stop_app
Stopping node rabbit@controller3 ...
[root@controller3 ~]# rabbitmqctl join_cluster rabbit@controller1
Clustering node rabbit@controller3 with rabbit@controller1 ...
[root@controller3 ~]# rabbitmqctl start_app
Starting node rabbit@controller3 …

在任意节点执行 rabbitmqctl cluster_status 查看集群

[root@controller1 ~]# rabbitmqctl cluster_status
Cluster status of node rabbit@controller1 ...
[{nodes,[{disc,[rabbit@controller1,rabbit@controller2,rabbit@controller3]}]},
 {running_nodes,[rabbit@controller3,rabbit@controller2,rabbit@controller1]},
 {cluster_name,<<"rabbit@controller1">>},
 {partitions,[]},
 {alarms,[{rabbit@controller3,[]},
          {rabbit@controller2,[]},
          {rabbit@controller1,[]}]}]

设置镜像队列
在任意一个节点上执行：

[root@controller1 rabbitmq]# rabbitmqctl set_policy ha-all '^(?!amq\.).*' '{"ha-mode": "all"}'
Setting policy "ha-all" for pattern "^(?!amq\\.).*" to "{\"ha-mode\": \"all\"}" with priority "" ...
...done.

将所有队列设置为镜像队列，即队列会被复制到各个节点，各个节点状态保持一致。

在rabbitmq中创建openstack用户

[root@controller1 rabbitmq]# rabbitmqctl add_user openstack openstack
Creating user "openstack" ...
...done.
[root@controller1 rabbitmq]# rabbitmqctl set_permissions openstack ".*" ".*" ".*"
Setting permissions for user "openstack" in vhost "/" ...
...done.

到此，rabbitmq高可用搭建完毕。

在配置Mariadb-galera和rabbitmq监听地址的时候，rabbitmq配置监听地址尤其要注意，配置如下：

[root@controller1 ~]# chown rabbitmq:rabbitmq /etc/rabbitmq/rabbitmq-env.conf
[root@controller1 ~]# systemctl restart rabbitmq-server
[root@controller1 ~]# netstat -ntplu | egrep
tcp               192.168.0.11:       0.0.0.0:*               LISTEN      /beam.smp
tcp               0.0.0.0:           0.0.0.0:*               LISTEN      /beam.smp

rabbitmq-env.conf 说明：
RABBITMQ_NODE_IP_ADDRESS= //IP地址，空串bind所有地址，指定地址bind指定网络接口
RABBITMQ_NODE_PORT=       //TCP端口号，默认是5672
RABBITMQ_NODENAME=        //节点名称。默认是rabbit
RABBITMQ_CONFIG_FILE= //配置文件路径，即rabbitmq.config文件路径
RABBITMQ_MNESIA_BASE=     //mnesia所在路径
RABBITMQ_LOG_BASE=        //日志所在路径
RABBITMQ_PLUGINS_DIR=     //插件所在路径