Demand background
The message RabbitMQ cluster at the same VPC Ali cloud migration from one segment to another segment cluster cluster. Messaging middleware is a message for immediate consumption, why are historical news, because it is a question left over by history. Therefore, to migrate
The entire network topology is as follows
note:
If the network for cross-VPC
1. Ensure that each host network interworking
2. configure each host name
Security groups on both sides of the direction of development: 15672,25672,5672,4369 port
No cause problems in joining the cluster
Resource List
| CPU name |
IP addresses |
Character |
Remark |
| node171 |
172.20.0.171 |
Old MQ cluster _1 |
|
| node172 |
172.20.0.172 |
Old MQ cluster _2 |
|
| by node 173 |
192.168.0.173 |
New in MQ cluster _1 |
|
| node174 |
192.168.0.174 |
The new MQ cluster _2 |
Infrastructure software and information environment
Erlang: Erlang / OTP 20 [ore-9.3.3.3]
RabbitMQ：rabbitmq_server-3.7.8
Cluster deployment
node171, node172 up the cluster A
node173, node174 up the cluster B
Here's a little deployment environment
Create a test account
[Operation] in the node171
rabbitmqctl add_user root root123
rabbitmqctl add_vhost kcvhost
rabbitmqctl set_permissions -p kcvhost root ".*" ".*" ".*"
rabbitmqctl add_user admin admin123
rabbitmqctl set_permissions -p kcvhost admin ".*" ".*" ".*"
rabbitmqctl set_user_tags admin administrator
rabbitmq-plugins enable rabbitmq_management
rabbitmqctl stop_app
rabbitmqctl start_app
Generate test data
News producer Code:
private static final String EXCHANGE_NAME = "exchange_test_3";
private static final String ROUTING_KEY = "routingkey_demo";
private static final String QUEUE_NAME = "queue_test_3";
private static final String IP_ADDRESS = "172.20.0.171";
private static final int PORT = 5672; //RabbitMQ服务默认端口号为5672
static void main public (String [] args) throws IOException, a TimeoutException, {InterruptedException
the ConnectionFactory Factory the ConnectionFactory new new = ();
factory.setHost (IP_ADDRESS);
factory.setPort (PORT);
factory.setUsername ( "the root");
Factory. the setPassword ( "root123");
connection connection factory.newConnection = (); // create a connection
channel channel = connection.createChannel (); // Create channel
// Create a type = "direct", persistent, non-automatic deletion switch
channel.exchangeDeclare (EXCHANGE_NAME, "Direct", to true, false, null);
// create a persistent, non-exclusive, non-automatic deletion queue
channel.queueDeclare (qUEUE_NAME, true, false, false, null);
// switch to the queue via the routing key bindings
channel.queueBind (queue_name, EXCHANGE_NAME, ROUTING_KEY);
// send a persistent message: hello world!
for (int I = 1; I <= 100000; I ++) {
String MSG = "_1 exchanger and queue 1 binding: the Message _" + I;
channel.basicPublish (EXCHANGE_NAME, ROUTING_KEY, MessageProperties.PERSISTENT_TEXT_PLAIN, Msg.getBytes ( ));
}
// close resource
channel.close ();
Connection.close ();
}
}
Consumer Code
package com.zjkj.rabbitmq.demo;
import java.io.IOException;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.TimeoutException;
import com.rabbitmq.client.AMQP;
import com.rabbitmq.client.Address;
import com.rabbitmq.client.Channel;
import com.rabbitmq.client.Connection;
import com.rabbitmq.client.ConnectionFactory;
import com.rabbitmq.client.Consumer;
import com.rabbitmq.client.DefaultConsumer;
import com.rabbitmq.client.Envelope;
/**
* 消息的消费者
* @author zjkj
*
*/
public class Rabbitmq_Consumer {
private static final String QUEUE_NAME = "queue_test_3";
private static final String IP_ADDRESS = "192.168.6.171";
private static final int PORT = 5672;
static void main public (String [] args) throws IOException, a TimeoutException, {InterruptedException
the Address [] = new new the Address Addresses [] {
new new the Address (IP_ADDRESS, PORT)
};
the ConnectionFactory Factory the ConnectionFactory new new = ();
factory.setUsername ( "the root ");
factory.setPassword (" root123 ");
// connection here with the demo producers slightly different, note the difference
connection connection = factory.newConnection (addresses); // create a connection
final Channel channel = connection.createChannel (); // Create channel
channel.basicQos (64); // set up the client has not been received ack message number
/ **
* This use of inheritance DefaultConsumer way to achieve consumption, have had experience with RabbitMQ developer
* may prefer to use QueueingConsumer way to achieve consumption
* Use QueueingConsumer because there will be some problems.
* While RabbitMQ Java client begins QueueingConsumer 4.0.0 @Deprecated a marked
* /
Consumer Consumer = new new DefaultConsumer (Channel) {
@Override
public void handleDelivery (String consumerTag, Envelope Envelope, AMQP.BasicProperties Properties, byte [] body)
throws IOException {
System.out.println ( "the recv Message:" + new new String (body));
the try {
TimeUnit.SECONDS.sleep (. 1);
the catch} (InterruptedException E) {
e.printStackTrace ();
}
channel.basicAck (envelope.getDeliveryTag (), to false);
}
};
channel.basicConsume (queue_name, to true, Consumer);
// wait for a callback function completes after, Close the resource
TimeUnit.SECONDS.sleep (. 5);
channel.close ();
Connection.close ();
}
}
Such as the number of users view the cluster, the number of switches, the number of queues, etc.
[root@node171 rabbitmq]# rabbitmqctl list_users
Listing users ...
admin [monitoring]
guest [administrator]
root []
[root@node171 rabbitmq]# rabbitmqctl list_exchanges
Listing exchanges for vhost / ...
amq.topic Topic
amq.headers headers
exchange_test_3 direct
amq.direct direct
exchange_test_2 direct
amq.rabbitmq.trace topic
amq.match headers
direct
exchange_test_1 direct
amq.fanout fanout
[root@node171 rabbitmq]# rabbitmqctl list_queues
Timeout: 60.0 seconds ...
Listing queues for vhost / ...
queue_test_3 100000
queue_test_2 200
queue_test_1 10000
Migration Scenarios
Migration Steps
1. Stop all producers and consumers of applications
2. The machine cluster B is added sequentially one of a cluster A, and confirm that all queues complete image
3. A culling cluster in one machine
4. Application of the cluster B to point
Scenario 1 [feasible]
The cluster B is added to the cluster A of a machine, and then the cluster B is another machine he has joined the cluster, the cluster A and then removed in one machine, and then remove the cluster A in another machine
This program is for ordinary RabbitMQ cluster that is, Cluster mode is invalid
1. Stop all connected to the cluster A
2. B is a cluster node to the cluster A
The .erlang.cookie value of the cluster A to cluster B copies of node173
[root@node171 rabbitmq]# cat .erlang.cookie
ORMTFBMHOXOGFKRLQSPU[root@node171 rabbitmq]#
[root@node173 plugins]# cp /var/lib/rabbitmq/.erlang.cookie /var/lib/rabbitmq/erlang.cookie.bak
[root@node173 plugins]# chmod 700 /var/lib/rabbitmq/.erlang.cookie
[root@node173 plugins]# vi /var/lib/rabbitmq/.erlang.cookie
ORMTFBMHOXOGFKRLQSPU
[root@node173 plugins]# chmod 400 /var/lib/rabbitmq/.erlang.cookie
[root@node173 plugins]# ls -lrth /var/lib/rabbitmq/.erlang.cookie
-r-------- 1 rabbitmq rabbitmq 21 Oct 24 18:51 /var/lib/rabbitmq/.erlang.cookie
3. node173 cluster B is added to the cluster A,
[root@node173 rabbitmq]# service rabbitmq-server start
Redirecting to /bin/systemctl start rabbitmq-server.service
[root@node173 rabbitmq]# rabbitmqctl stop_app
Stopping rabbit application on node mq_173@node173 ...
[root@node173 rabbitmq]# rabbitmqctl reset
Resetting node mq_173@node173 ...
[root@node173 rabbitmq]# rabbitmqctl join_cluster mq171@node171
Clustering node mq_173@node173 with mq171@node171
[root@node173 rabbitmq]# rabbitmqctl start_app
Starting node mq_173@node173 ...
completed with 3 plugins.
4. The same method as in the cluster B of the cluster A is added to the node174
[root@node174 rabbitmq]# rabbitmqctl cluster_status
Cluster status of node mq_174@node174 ...
[{nodes,[{disc,[mq_174@node174]}]},
{running_nodes,[mq_174@node174]},
{cluster_name,<<"mq_174@node174">>},
{partitions,[]},
{alarms,[{mq_174@node174,[]}]}]
[root@node174 rabbitmq]# rabbitmqctl stop_app
Stopping rabbit application on node mq_174@node174 ...
[root@node174 rabbitmq]# rabbitmqctl reset
Resetting node mq_174@node174 ...
[root@node174 rabbitmq]# rabbitmqctl join_cluster mq171@node171
Clustering node mq_174@node174 with mq171@node171
[root@node174 rabbitmq]# rabbitmqctl start_app
Starting node mq_174@node174 ...
completed with 0 plugins.
5. A cluster of cluster node171 removed
[root@node171 rabbitmq]# rabbitmqctl stop
Stopping and halting node mq171@node171 ...
Then visit the Web cluster management node172
The same Web management interface on node173 View
So far for the average cluster model, this approach is not acceptable.
Scenario 2 [feasible]
If mirroring RabbitMQ queue, the data migration message to the cluster A, cluster B,
A cluster of node171, node172 queue mirroring
A mirror image queue build cluster environment
1. A first cluster of the cluster added node172
[Operation] on node172
[root@node172 ~]# rabbitmqctl stop_app
Stopping rabbit application on node mq172@node172 ...
[root@node172 ~]# rabbitmqctl reset
Resetting node mq172@node172 ...
[root@node172 ~]# rabbitmqctl join_cluster mq171@node171
Clustering node mq172@node172 with mq171@node171
ra[root@node172 ~]# rabbitmqctl start_app
Starting node mq172@node172 ...
2. Set the mirror policy
[Operation] on node171
[root@node171 ~]# rabbitmqctl set_policy ha-all -p kcvhost "^" '{"ha-mode":"all","ha-sync-mode":"automatic"}'
Setting policy "ha-all" for pattern "^" to "{"ha-mode":"all","ha-sync-mode":"automatic"}" with priority "0" for vhost "kcvhost" ...
[root@node171 ~]# rabbitmqctl set_policy rabbit_mirror "^" '{"ha-mode":"all"}'
Setting policy "rabbit_mirror" for pattern "^" to "{"ha-mode":"all"}" with priority "0" for vhost "/" ...
Mirroring queue migration starts in cluster A
1. Stop all message producer and consumer related applications and services
2. Stop all machines in the cluster A and backup of the original data
[Node171, node172] must operate
Node172 performed as follows:
[root@node172 ~]# service rabbitmq-server stop
Redirecting to /bin/systemctl stop rabbitmq-server.service
[root@node172 ~]# cd /var/lib/rabbitmq/
[root@node172 rabbitmq]# ls
mnesia
[root@node172 rabbitmq]# cp -rf mnesia mnesia.20181025.bak
[root@node172 rabbitmq]# service rabbitmq-server start
Redirecting to /bin/systemctl start rabbitmq-server.service
node171 performed as follows:
[root@node171 ~]# service rabbitmq-server stop
Redirecting to /bin/systemctl stop rabbitmq-server.service
[root@node171 ~]# cd /var/lib/rabbitmq/
[root@node171 rabbitmq]# cp -rf mnesia mnesia.20181025.bak
[root@node171 rabbitmq]# service rabbitmq-server start
Redirecting to /bin/systemctl start rabbitmq-server.service
2. First node173 machine cluster B is added to the cluster A,
[root@node173 network-scripts]# service rabbitmq-server stop
Redirecting to /bin/systemctl stop rabbitmq-server.service
[root@mq04 rabbitmq]# cp -rf /var/lib/rabbitmq /var/lib/rabbitmq.bak
[root@mq04 rabbitmq]# cd /var/lib/rabbitmq
[root@mq04 rabbitmq]# rm -rf .erlang.cookie mnesia/
[root@mq01 rabbitmq]# scp .erlang.cookie root@mq04:/var/lib/rabbitmq
The authenticity of host 'mq04 (192.168.0.232)' can't be established.
ECDSA key fingerprint is SHA256:zgAicKOpvRLLCyhdUbpNvyanKYrPt/Pp9g+Sdq9mAoo.
ECDSA key fingerprint is MD5:15:7d:1e:c2:86:d5:4a:40:63:df:f5:4e:65:c4:24:62.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added 'mq04' (ECDSA) to the list of known hosts.
root@mq04's password:
Permission denied, please try again.
root@mq04's password:
.erlang.cookie 100% 20 19.6KB/s 00:00
[root@mq04 rabbitmq]# chmod 400 .erlang.cookie
[root@mq04 rabbitmq]# chown -R rabbitmq:rabbitmq .erlang.cookie
[root@mq04 rabbitmq]# service rabbitmq-server start
Redirecting to /bin/systemctl start rabbitmq-server.service
[root@mq04 rabbitmq]# rabbitmqctl stop_app
Stopping rabbit application on node mq04@mq04 ...
[root@mq04 rabbitmq]# rabbitmqctl reset
Resetting node mq04@mq04 ...
For Ali cloud ECS must first open a temporary port security group 15672,25672,5672,4369
[root@node173 network-scripts]# rabbitmqctl join_cluster mq171@node171
Clustering node mq_173@node173 with mq171@node171
[root@node173 network-scripts]# rabbitmqctl start_app
Starting node mq_173@node173 ...
completed with 3 plugins.
3. B then node174 machine cluster to the cluster A,
Using the same method as above, is added to the cluster to node174
4. A cluster excluding the node171, node172 machine
Execution on Node172
[root@mq02 rabbitmq]# rabbitmqctl stop_app
Stopping rabbit application on node mq02@mq02 ...
[root@mq02 rabbitmq]# service rabbitmq-server stop
Redirecting to /bin/systemctl stop rabbitmq-server.service
[root@mq02 rabbitmq]# cp -rf /var/lib/rabbitmq /var/lib/rabbitmq.bak
[root@mq02 rabbitmq]# service rabbitmq-server start
Redirecting to /bin/systemctl start rabbitmq-server.service
[root@mq02 rabbitmq]# rabbitmqctl stop_app
Stopping rabbit application on node mq02@mq02 ...
[root@mq02 rabbitmq]# rabbitmqctl reset
Resetting node mq02@mq02 ...
The same command executed on the same node171
Mirroring the queue for the cluster, this is feasible