目录
[Management UI plugin安装](#Management UI plugin安装)
[Queue leader strategy](#Queue leader strategy)
[Node 1](#Node 1)
[Installation path](#Installation path)
[Node 2](#Node 2)
[Installation path](#Installation path)
[Node 3](#Node 3)
[Installation path](#Installation path)
[Queue and AnonymousQueue](#Queue and AnonymousQueue)
本文利用一台Linux示范三节点RabbitMQ集群。RabbitMQ版本:3.12.11.
重点:
- 关键技术点
- 节点间相互验证,.erlang.cookie
- nodename的唯一性
- port冲突与配置
- Management UI plugin安装
- Queue replicas
- Queue leader strategy
- 配置文件详细
- Springboot连接与编程
几个关键技术点
节点间相互验证,.erlang.cookie
.erlang.cookie是erlang框架实现分布式的必要文件。三节点集群需要每个节点的.erlang.cookie文件相同。文件的位置:$home/.erlang.cookie.
erlang安装之后文件就已经存在,如果使用三台不同的机器,需要将其中一台的文件拷贝三份。用一台机器做示范,我们不需要再做额外的动作。
nodename的唯一性
RabbitMQ的nodename有固定的格式<prefix>@<hostname>。hostname可以通过操作系统的hostname命令获得。prefix可以自己决定。但是集群中的nodename必须是唯一的。
如果是部署在不同的机器上,所有的hostname需要明确配置在/etc/hosts文件中。
nodename配置在rabbitmq-env.conf文件中。
port冲突与配置
部署在同一台机器上,需要解决三个端口的冲突。
- amqp通信端口5672 - 相关配置文件rabbitmq.conf or rabbitmq-env.conf
- management UI端口15672 - 相关配置文件rabbitmq.conf
- 内部管理Command Line Interface端口25672 - 相关配置文件rabbitmq-env.conf
Management UI plugin安装
信息要放在enabled_plugin文件中。这个文件要手工添加,插件在RabbitMQ第一次启动的时候初始化。
Queue(message)replicas
高可用需要同一条消息要同时复制到三个节点上面,这样应用连接的当前节点下线后,还可以从其它另外两个节点读取还未消费完的消息。没有replica,虽然Rabbitmq依然可用,但会有消息丢失的情况。
要实现消息node间复制,需要创建Queue的时候, queue type设置成Quorum。
两种方式:
- 通过管理界面创建Queue
- Java应用在启动的时候配置Queue
java
Map QUEUE_ARGS = new HashMap<String,Object>();
QUEUE_ARGS.put("x-queue-type", "quorum");
new Queue("queue_name", true, false, false, QUEUE_ARGS);
对于匿名Queue不能配置这个属性,当一个node down以后,应用会在新的节点重新创建临时Queue。可能产生message丢失的情况。
Queue leader strategy
在rabbitmq.conf中配置,两种client-local(by default)和balanced.
配置文件详细
Download rabbitmq package, rabbitmq-server-generic-unix-3.12.11.tar.xz. 解压后拷贝三份,作为三个node。
下面的文件除了.erlang.cookie都需要手动添加。
Node 1
Installation path
/opt/rabbitmq/rabbitmq-node-1
.erlang.cookie
$home/.erlang.cookie
rabbitmq-env.conf
File path: /opt/rabbitmq/rabbitmq-node-1/etc/rabbitmq/rabbitmq-env.conf
java
# Example rabbitmq-env.conf file entries. Note that the variables
# do not have the RABBITMQ_ prefix.
#
# Overrides node name
RABBITMQ_NODENAME=rabbitmq-node-1@DESKTOP-B5WANG
# Specifies new style config file location
RABBITMQ_CONFIG_FILE=/opt/rabbitmq/rabbitmq-node-1/etc/rabbitmq/rabbitmq.conf
# Specifies advanced config file location
# RABBITMQ_ADVANCED_CONFIG_FILE=/etc/rabbitmq/advanced.config
# CLI port
RABBITMQ_DIST_PORT=25672
rabbitmq.conf
File path: /opt/rabbitmq/rabbitmq-node-1/etc/rabbitmq/rabbitmq.conf
java
# port
listeners.tcp.default = 5672
# listeners.tcp.1 = 0.0.0.0:5672
# default user
default_vhost = /
default_user = admin
default_pass = admin
default_permissions.configure = .*
default_permissions.read = .*
default_permissions.write = .*
default_user_tags.administrator = true
default_user_tags.management = true
default_user_tags.custom_tag = true
# management plugin
management.tcp.port = 15672
management.tcp.ip = 0.0.0.0
# cluster
cluster_name = rabbitmq-cluster
cluster_keepalive_interval = 10000
cluster_formation.peer_discovery_backend = classic_config
cluster_formation.classic_config.nodes.1 = rabbitmq-node-1@DESKTOP-B5WANG
cluster_formation.classic_config.nodes.2 = rabbitmq-node-2@DESKTOP-B5WANG
cluster_formation.classic_config.nodes.3 = rabbitmq-node-3@DESKTOP-B5WANG
# queue location strategy
queue_master_locator = client-local
enabled_plugins
File path: /opt/rabbitmq/rabbitmq-node-1/etc/rabbitmq/enabled_plugins
java
[rabbitmq_management].
Node 2
Installation path
/opt/rabbitmq/rabbitmq-node-2
.erlang.cookie
$home/.erlang.cookie
rabbitmq-env.conf
File path: /opt/rabbitmq/rabbitmq-node-2/etc/rabbitmq/rabbitmq-env.conf
java
# Example rabbitmq-env.conf file entries. Note that the variables
# do not have the RABBITMQ_ prefix.
#
# Overrides node name
RABBITMQ_NODENAME=rabbitmq-node-2@DESKTOP-B5WANG
# Specifies new style config file location
RABBITMQ_CONFIG_FILE=/opt/rabbitmq/rabbitmq-node-2/etc/rabbitmq/rabbitmq.conf
# Specifies advanced config file location
# RABBITMQ_ADVANCED_CONFIG_FILE=/etc/rabbitmq/advanced.config
# CLI port
RABBITMQ_DIST_PORT=25673
rabbitmq.conf
File path: /opt/rabbitmq/rabbitmq-node-2/etc/rabbitmq/rabbitmq.conf
java
# port
listeners.tcp.default = 5673
# listeners.tcp.1 = 0.0.0.0:5673
# default user
default_vhost = /
default_user = admin
default_pass = admin
default_permissions.configure = .*
default_permissions.read = .*
default_permissions.write = .*
default_user_tags.administrator = true
default_user_tags.management = true
default_user_tags.custom_tag = true
# management plugin
management.tcp.port = 15673
management.tcp.ip = 0.0.0.0
# cluster
cluster_name = rabbitmq-cluster
cluster_keepalive_interval = 10000
cluster_formation.peer_discovery_backend = classic_config
cluster_formation.classic_config.nodes.1 = rabbitmq-node-1@DESKTOP-B5WANG
cluster_formation.classic_config.nodes.2 = rabbitmq-node-2@DESKTOP-B5WANG
cluster_formation.classic_config.nodes.3 = rabbitmq-node-3@DESKTOP-B5WANG
# queue location strategy
queue_master_locator = client-local
enabled_plugins
File path: /opt/rabbitmq/rabbitmq-node-2/etc/rabbitmq/enabled_plugins
java
[rabbitmq_management].
Node 3
Installation path
/opt/rabbitmq/rabbitmq-node-3
.erlang.cookie
$home/.erlang.cookie
rabbitmq-env.conf
File path: /opt/rabbitmq/rabbitmq-node-3/etc/rabbitmq/rabbitmq-env.conf
java
# Example rabbitmq-env.conf file entries. Note that the variables
# do not have the RABBITMQ_ prefix.
#
# Overrides node name
RABBITMQ_NODENAME=rabbitmq-node-3@DESKTOP-B5WANG
# Specifies new style config file location
RABBITMQ_CONFIG_FILE=/opt/rabbitmq/rabbitmq-node-3/etc/rabbitmq/rabbitmq.conf
# Specifies advanced config file location
# RABBITMQ_ADVANCED_CONFIG_FILE=/etc/rabbitmq/advanced.config
# CLI port
RABBITMQ_DIST_PORT=25674
rabbitmq.conf
File path: /opt/rabbitmq/rabbitmq-node-3/etc/rabbitmq/rabbitmq.conf
java
# port
listeners.tcp.default = 5674
# listeners.tcp.1 = 0.0.0.0:5674
# default user
default_vhost = /
default_user = admin
default_pass = admin
default_permissions.configure = .*
default_permissions.read = .*
default_permissions.write = .*
default_user_tags.administrator = true
default_user_tags.management = true
default_user_tags.custom_tag = true
# management plugin
management.tcp.port = 15674
management.tcp.ip = 0.0.0.0
# cluster
cluster_name = rabbitmq-cluster
cluster_keepalive_interval = 10000
cluster_formation.peer_discovery_backend = classic_config
cluster_formation.classic_config.nodes.1 = rabbitmq-node-1@DESKTOP-B5WANG
cluster_formation.classic_config.nodes.2 = rabbitmq-node-2@DESKTOP-B5WANG
cluster_formation.classic_config.nodes.3 = rabbitmq-node-3@DESKTOP-B5WANG
# queue location strategy
queue_master_locator = client-local
enabled_plugins
File path: /opt/rabbitmq/rabbitmq-node-3/etc/rabbitmq/enabled_plugins
java
[rabbitmq_management].
Springboot连接与编程
application.yml
java
spring:
rabbitmq:
# Connect rabbitmq cluster
addresses: localhost:5672,localhost:5673,localhost:5674
username: admin
password: admin
Queue and AnonymousQueue
Queue
java
Map arg = new HashMap<String,Object>();
arg.put("x-queue-type", "quorum");
arg.put("x-queue-leader-locator", "client-local");
new Queue(QUEUE_NAME_TEXT_MESSAGE, true, false, false, arg);
AnonymousQueue
java
AnonymousQueue aq = new AnonymousQueue(new Base64UrlNamingStrategy("queue-prefix-"));
AnonymousQueue本质上还是Queue,只是属性不同。所以不需要配置quorum,否则应用会报错。
java
public class AnonymousQueue extends Queue {
public AnonymousQueue() {
this((Map)null);
}
public AnonymousQueue(Map<String, Object> arguments) {
this(Base64UrlNamingStrategy.DEFAULT, arguments);
}
public AnonymousQueue(NamingStrategy namingStrategy) {
this(namingStrategy, (Map)null);
}
public AnonymousQueue(NamingStrategy namingStrategy, Map<String, Object> arguments) {
super(namingStrategy.generateName(), false, true, true, arguments);
if (!this.getArguments().containsKey("x-queue-master-locator")) {
this.setLeaderLocator("client-local");
}
}
}