ActiveMQ (seven) - ActiveMQ's Network

First, start multiple servers on a single Broker

Steps are as follows (in preparation for a cluster):
1: the entire conf folder copy, for example, is called CONF2
2: Modify inside activemq.xml file
(1) inside brokerName can not repeat with the original
store (2) data file names must be unique, such as:
<hahaDB Directory = "$ {} activemq.data / kahadb_2" />
(3) designed transportConnectors all ports have not the same with the foregoing
3: modify jetty.xml, mainly modify the ports , for example:
<Property name = "port" value = "8181" /> ports must be in front of not the same
4: to the bin below, copy a ActiveMQ, such known activemq2;
ID (. 1) to modify the program, not the previous repeat
ACTIVEMQ_PIDFILE = "$ ACTIVEMQ_DATA / activemq2-hostname.pid"
(2) modify the configuration file path
ACTIVEMQ_CONF = "$ ACTIVEMQ_BASE / CONF2"
(. 3) modify the port, which has a port of 61616 tcp, to change not the same, the best agreement with the port activemq.xml inside ctp
(4) can then be carried out.

Two, ActiveMQ static network links

ActiveMQ is what the networkConnector
in some scenarios, you need to do ActiveMQ Broker multiple clusters, so long it comes to communication Broker. This is called networkConnector of ActiveMQ.
The default ActiveMQ networkConnector is unidirectional, a Broker sends a message at one end, a further message is received at the other end of the Broker. This is called "bridging." ActiveMQ also supports two-way link to create a two-way channel used to receive messages for both Broker, but also send messages only from the same channel, is typically mapped as duplex connector, as follows:
"Discovery" concept
in general, Discovery is used to discover remote service client typically want to discover all available brokers; another meaning, which is based on the existing network to find other available Broker Brokers .
There are two configurations Client to the Broker linking, one way: by means of Client to connect Broker Statically configured another way: Client be dynamically discovered by discovery agents Brokers
Networks Static
Static networkConnector is used to create a static configuration for multiple Broker network. This protocol is used for a composite url, a composite comprising a plurality url url addresses. Format is as follows:
static: (URI1, URI2, URI 3, ...) Key value =?
. 1: configuration example as follows (activemq.xml-- commented persistenceFactory node):
```
<networkConnectors>
<networkConnector name="local network" uri="static://(tcp://remotehost1:61616,tcp://remotehost2:61616)"/>
</networkConnectors>
```
The original basic schematic of Static networkConnector:

figure above, two Brokers are linked to the network through a static protocol. A link to a Consumer BrokerB addresses when sending a message on Producer BrokerA at the same address, at which point it was transferred to the BrokerB. That is, BrokerA will forward the message to the BrokerB.
networkConnector configuration of available properties
1, name: the default is Bridge
2, dynamicOnly: default is false , creates a corresponding network only if it is true, long-lasting durable subscription subscription is activated. It is enabled by default when you start
3, decreaseNetworkConsumerPriority: The default is false. Consumers priority setting, if true, the consumer network priority reduced to -5. If false, the default is the same as with local consumers 0
4, networkTTL: The default is 1 , the network for news and subscribe broker amount of consumption of
5, messageTTL: The default is 1 , the number of broker networks for the message
6, consumerTTL: default is 1 , the number of broker network for consumption
7, conduitSubscriptions: defaults to true , whether the consumer with a plurality of as a broker to handle
8, dynamicallyIncludedDestinations: blank by default , to be a message including a dynamic address, similar to in excludedDestinations, such as
```
<dynamicallyIncludedDestinations>
<queue physicalName="include.test.foo"/>
<topic physicalName="include.test.bar"/>
</dynamicallyIncludedDestinations>
```
9, staticallyIncludedDestinations: blank by default , to include the static message address. Similar excludedDestinations, such as:
```
<staticallyIncludedDestinations>
<queue physicalName="always.include.queue"/>
</staticallyIncludedDestinations>
```
10, excludedDestinations: Default is null , the specified address exclusion, examples are as follows:

. 11, Duplex: defaults to false , setting whether bidirectional communication
12, prefetchSize: default is 1000 , the maximum number of unacknowledged messages held, must be greater than 0, because consumers can not own a network poll message
13, suppressDuplicateQueueSubscriptions: default false , if true, to produce a repeat of the subscription relationship ie stop
14, bridgeTempDestinations: defaults to true , whether to create a temporary broadcast advisory messages Where do you want
15, alwaysSyncSend: default false transmitting if true, will also be non-persistent messages using request / reply mode to the remote mode instead of oneway Broker
16, staticBridge: defaults to false , if true, only staticallyIncludedDestinations configured destination can be processed

三、“丢失”的消息
存在这样的场景，broker1和broker2通过networkConnector连接，一些consumers连接到broker1，消费broker2上的消息。消息先被broker1从broke2上消费掉，然后转发给这些consumers。不幸的是转发部分消息的时候broker1重启了，这些consumers发现broker1连接失败，通过failover连接到broker2上去了，但是又一部分他们还没有消费的消息被broker2已经分发到broker1上去了。这些消息，就好像是消失了，除非有消费者重新连接到broker1上来消费。
从5.6版起，在destinationPolicy上新增的选项replayWhenNoConsumers。这个选项使得broker1上有需要转发的消息但是没有消费者时，把消息回流到它原始的broker。同时把enableAudit设置为false，为了防止消息回流后被当做重复消息而不被分发，示例如下：

<destinationPolicy>
    <policyMap>
        <policyEntry queue=">" enableAudit="false">
            <networkBridgeFilterFactory>
                <conditionalNetworkBridgeFilterFactory replayWhenNoConsumers="true"/>
            </networkBridgeFilterFactory>
        </policyEntry>
    </policyMap>
</destinationPolicy>

四、容错的连接

Failover Protocol
之前的都是Client配置链接到指定的broker上。但是，如果Broker的链接失败怎么办？此时，Client有两个选项：要么立刻死掉，要么去连接到其它的broker上。
Failover协议实现了自动重新链接的逻辑。这里有两种方式提供了稳定的brokers列表对于Client链接。
第一种：提供一个static的可用的Brokers列表
第二种：提供一个dynamic发现的可用Brokers
Failover Protocol的默认方式
failover:(uri1,...,uriN)?key=value或者failover:uri1,...,uriN
& Failover Protocol的默认配置
默认情况下，这种协议用于随机的去选择一个链接去链接，如果链接失败了，那么会链接到其它的Broker上。默认的配置定义了延迟重新链接，意味着传输将会在10秒后自动的去重新链接可用的broker。所有的重新链接参数都是可以根据应用的需要而配置的。

Failover Protocol的使用示例，在客户端程序里面

ConnectionFactory connectionFactory = new ActiveMQConnectionFactory   ("failover:(tcp://localhost:61616,tcp://localhost:61617)
?randomize=false");

Failover Protocol可用的配置参数：
1、initialReconnectDelay：第一次尝试重连之前等待的时间（毫秒）默认10
2、maxReconnectDelay：最长重连的时间间隔（毫秒），默认30000
3、useExponentialBackOff：重连时间间隔是否以指数形式增长，默认true
4、backOffMultiplier：递增倍数，默认2.0
5、maxReconnectAttempts：默认-1|0，自版本5.6起，-1为默认值，代表不限重连次数；0代表从不重试（只尝试连接一次，并不重连），5.6版本之前，0为默认值，代表不限重试次数；大于0的数，代表最大重试次数。
6、startupMaxReconnectAttempts：初始化时的最大重连次数。一旦连接上，将使用maxReconnectAttempts的配置，默认0
7、Randomize：使用随机连接，以达到负载均衡的目的，默认true
8、Backup：提前初始化一个未使用连接，以便进行快速失败转移，默认false
9、timeout：设置发送操作的超时时间（毫秒），默认-1
10、trackMessages：设置是否缓存[故障发生时]尚未传送完成的消息，当broker一旦重新连接成功，便将这些缓存中的消息刷新到新连接的代理中，使得消息可以在broker切换前后顺序传送，默认false
11、maxCacheSize：当trackMessages启用时，缓存的最大字节，默认为128*1024 字节
12、updateURIsSupported：设定是否可以动态修改broker rui（自版本5.4起），默认true

五、动态网络连接（纯理论）
多播协议multicast
ActiveMQ使用Multicast协议将一个Service和其他的Broker的Service连接起来。IP multicast是一个被用于网络中传输数据到其他一组接收者的技术。
IP multicast传统的概念成为组地址。组地址是ip地址在224.0.0.0到239.255.255.255之间的ip地址。ActiveMQ broker使用multicast协议去建立服务与远程的broker的服务的网络连接。

基本的格式配置
multicast://ip address:port?transportOptions
1、group：表示唯一的组名称，缺省值default
2、minmumWireFormatVersion：被允许的最小的wireformat版本，缺省为0
3、trace：是否追踪记录日志，默认false
4、useLocalHost：表示本地机器的名称是否为localhost，默认true
5、datagramSize：特定的数据大小，默认值41024
6、timeToLive：消息的声明周期值，默认值-1
7、loopBackMode：是否启用loopback模式，默认false
8、wireFormat：默认用wireFormat命名
9、wireFormat.：前缀是wireFormat
配置示例
1：默认配置，默认情况下是不可靠的多播，数据包可能会丢失
multicast://default
2：特定的ip和端口
multicast://224.1.2.3:6255
3：特定的ip和端口以及组名
multicast://224.1.2.3:6255?group=mygroupname
ActiveMQ使用multicast协议的配置格式如下：
```
<broker xmlns="http://activemq.apache.org/schema/core" brokerName="multicast" dataDirectory="${activemq.data}/data">
<networkConnectors>
    <networkConnector name="default-nc" uri="multicast://default"/>
</networkConnectors>
<transportConnectors>
    <transportConnector name="openwire" uri="tcp://localhost:61616" discoveryUri="multicast://default"/>
</transportConnectors>
</broker>
```
说明：
1：uri="multicast://default"中的default是activemq默认的ip，默认动态的寻找地址
2："discoveryUri"是指在transport中用multicast的default的地址传递
3："uri"指动态寻找可利用的地址
4：如果防止自动的寻找地址？
（1）名称为openwire的transport，移除discoveryUri="multicast:
//default"即可。传输连接用默认的名称openwire来配置broker的tcp多点连接，这将允许其他broker能够自动发现和连接到可用的broker中。
（2）名称为"default-nc"的networkConnector，注释掉或者删除即可。
ActiveMQ默认的networkConnector基于multicast协议的连接的默认名称是default-nc，而且自动的去发现其他broker。去停止这种行为，只需要注销或者删除掉default-nc网络连接。
（3）使brokerName的名字唯一，可以唯一识别Broker的实例，默认是localhost
multicast协议和普通的tcp协议
它们是差不多的，不同的是multicast能够自动的发现其他broker，从而替代了使用static功能列表brokers。用multicast协议可以在网络中频繁的添加和删除ip不会有影响。
multicast协议
好处：能够适应动态变化的地址。
缺点：自动连接地址会过渡的消耗网络资源

Discovery协议
Discovery是在multicast协议的功能上定义的。功能类似与Failover功能。它将动态的发现multicast协议的broker的连接并且随机的连接其中一个broker。

基本配置如下：
discovery:(discoveryAgentURI)?transportOptions
1、reconnectDelay：再次寻址等待时间，缺省值10
2、initialReconnectDelay：初始化设定再次寻址等待时间，缺省值10
3、maxReconnectDelay：最大寻址等待时间，缺省值30000
4、useExponentialBackOff：是否尝试BackOff重连接，默认是true
5、backOffMultiplier：尝试Backoff的次数，默认是2
6、maxReconnectAttempts：如果异常，最大的重新连接个数，默认是0
7、Group：组唯一的地址，默认是default
示例：
discovery:(multicast://default)?initialReconnectDelay=100

Discovery协议的配置示例

<broker name="foo">
<transportConnectors>
    <transportConnector uri="tcp://localhost:0" discoveryUri="multicast://default"/>
</transportConnectors>
</broker>

Peer协议
ActiveMQ提供了peer transport connector提供了更加容易的去嵌入broker网络中。它创建一个优于vm连接的p2p网络连接。默认格式如下：
peer://peergroup/brokerName?Key=value

Peer协议的基本使用
当我们启用了peer协议时，应用将自动的启动内嵌broker，也将会自动的去配置其它broker来建立连接，前提是必须要有一个组。配置如下：
peer://groupa/broker1?Persistent=false
另外，生产者和消费者都各自连接到嵌入到自己应用的broker，并且在本地的同一个组名中相互访问数据。
Peer协议的基本原理

在本地机器断网的情况下，本地的client访问本地brokerA将依然正常。
在断网的情况下发送消息到本地brokerA，然后网路连接正常后，所有的消息将重新发送并连接到brokerB。
Fanout协议

Fanout协议是同时连接到多个broker，默认的格式如下：
fanout:(fanoutURI)?key=value
示例：fanout:(static:(tcp://host1:61616,tcp://host2:61616))
表示client将试图连接到两个static列表中定义的三个URI

Configuration parameters Fanout protocol are as follows:
. 1, initialReconnectDelay: again waits for a connection time, the default is 10
2, maxReconnectDelay: The maximum re-latency connection, the default is 30000
. 3, useExponentialBackOff: whether to attempt BackOff reconnection, defaults to true
. 4, backOffMultiplier : Backoff attempts, the default is 2
. 5, maxReconnectAttempts: if abnormal, the maximum number of re-connection, the default is 0
. 6, fanOutQueues: whether to convert the message queue message topic, default to false
. 7, minAckCount: Broker minimum number of connections, The default is 2
configuration examples:
fanout: (static: (tcp: // localhost: 61616, tcp: // remotehost: 61616))?
initialReconnectDelay = 100
Note:
ActiveMQ is not recommended to use fanout Consumer agreement. When sending a message to a plurality of Provider broker, the test may be repeated Consumer message.

ActiveMQ (seven) - ActiveMQ's Network

Guess you like