RabbitMQ Clusterhtml
1安裝準備node
單節點MQ的安裝參考其餘文檔,此處僅以兩個節點爲例,多個節點也如法炮製。cookie
² 已經安裝好rabbitmq的兩個單節點rabbitmq-broker2、rabbitmq-broker3。app
² 保證兩臺機器上的rabbitmq爲後臺啓動狀態。ide
² 安裝erlang、rabbitmq版本一致。不然,版本不一致會出現問題。spa
2 .erlang.cookie拷貝orm
.erlang.cookie該文件通常存放在安裝用戶目錄下,如:/home/sca_bj/.erlang.cookie。從rabbitmq-broker2機器上將其拷貝到rabbitmq-broker3目錄下。server
3配置前查看狀態htm
rabbitmqctl cluster_statusrabbitmq
狀態以下:
Cluster status of node 'sca@rabbitmq-broker3' ...
[{nodes,[{disc,['sca@rabbitmq-broker3']}]},
{running_nodes,['sca@rabbitmq-broker3']},
{cluster_name,<<"sca@ocdata09">>},
{partitions,[]}]
...done.
4 join cluster
將rabbitmq-broker3加入到rabbitmq-broker2中,在rabbitmq-broker3上操做
² rabbitmqctl stop_app
² rabbitmqctl join_cluster root@rabbitmq-broker2
² rabbitmqctl start_app
5查看加入後的狀態
rabbitmqctl cluster_status
² 配置cluster後狀態以下:
Cluster status of node 'sca@rabbitmq-broker3' ...
[{nodes,[{disc,['root@rabbitmq-broker2','sca@rabbitmq-broker3']}]},
{running_nodes,['root@rabbitmq-broker2','sca@rabbitmq-broker3']},
{cluster_name,<<"root@ocdata07">>},
{partitions,[]}]
...done.
² 中止rabbitmq-broker2上的MQ
rabbitmqctl stop
² 在rabbitmq-broker3查看狀態
rabbitmqctl cluster_status
Cluster status of node 'sca@rabbitmq-broker3' ...
[{nodes,[{disc,['root@rabbitmq-broker2','sca@rabbitmq-broker3']}]},
{running_nodes,['sca@rabbitmq-broker3']},
{cluster_name,<<"root@ocdata07">>},
{partitions,[]}]
² 再將rabbitmq-broker2從新啓動
rabbitmq-server -detached
² 也能夠經過界面管理端來查看其狀態
6 RAM集羣節點
咱們默認配置的disk 集羣節點,大的集羣也能夠RAM集羣節點。
在rabbitmq-broker3操做
² rabbitmqctl stop_app
² rabbitmqctl join_cluster --ram root@rabbitmq-broker2
² rabbitmqctl start_app
7去除節點的方法
移除rabbitmq-broker3,在rabbitmq-broker3上操做
² rabbitmqctl stop_app
² rabbitmqctl reset
² rabbitmqctl start_app
或者在rabbitmq-broker2上操做
² rabbitmqctl stop_app
² rabbitmqctl forget_cluster_node sca@rabbitmq-broker3
² 此時rabbitmq-broker2還會認爲和rabbitmq-broker3成集羣狀態,這時
rabbitmqctl start_app會報錯
Removing node 'sca@rabbitmq-broker3' from cluster ...
Error: {offline_node_no_offline_flag,"You are trying to remove a node from an offline node. That is dangerous, but can be done with the --offline flag. Please consult the manual for rabbitmqctl for more information."}
² 繼續執行命令
rabbitmqctl reset
² rabbitmqctl start_app
注意:整個集羣的啓動和中止順序恰好相反
8普通集羣的弊端
Queues的消息只存在於被建立的那臺機器上。
對於publish,客戶端任意鏈接集羣的一個節點,轉發給建立queue的節點存儲消息的全部信息;對於consumer,客戶端任意鏈接集羣中的一個節點,若是數據不在該節點中,則從存儲該消息data的節點拉取。可見當存儲有queue內容的節點失效後,只要等待該節點恢復後,queue中存在的消息才能夠獲取消費的到。顯然增長集羣的節點,能夠提升整個集羣的吞吐量,可是在高可用方面要稍微差一些。
附加:Rabbitmq HA:http://www.rabbitmq.com/clustering.html