/rocketmq-connect

A tool for scalable and reliably streaming data between Apache RocketMQ and other systems.

Primary LanguageJavaApache License 2.0Apache-2.0

rocketmq-connect

License

文档中心

RocketMQ Connect 文档

快速开始

单机模式下rocketmq-connect-sample作为 demo

rocketmq-connect-sample的主要作用是从源文件中读取数据发送到RocketMQ集群 然后从Topic中读取消息，写入到目标文件

1.准备

  1. Linux/Unix/Mac
  2. 64bit JDK 1.8+;
  3. Maven 3.2.x或以上版本;
  4. 启动 RocketMQ;
  5. 创建测试Topic

sh ${ROCKETMQ_HOME}/bin/mqadmin updateTopic -t fileTopic -n localhost:9876 -c DefaultCluster -r 8 -w 8

tips : ${ROCKETMQ_HOME} 位置说明

bin-release.zip 版本：/rocketmq-all-4.9.4-bin-release

source-release.zip 版本：/rocketmq-all-4.9.4-source-release/distribution

2.构建Connect

git clone https://github.com/apache/rocketmq-connect.git

cd  rocketmq-connect

mvn -Prelease-connect -DskipTests clean install -U

3.运行Worker

cd distribution/target/rocketmq-connect-0.0.1-SNAPSHOT/rocketmq-connect-0.0.1-SNAPSHOT

sh bin/connect-standalone.sh -c conf/connect-standalone.conf &

tips: 可修改 /bin/runconnect.sh 适当调整 JVM Parameters Configuration

JAVA_OPT="${JAVA_OPT} -server -Xms256m -Xmx256m"

runtime启动成功：

The standalone worker boot success.

查看启动日志文件：

tail -100f ~/logs/rocketmqconnect/connect_runtime.log

ctrl + c 退出日志

4.启动source connector

当前目录创建测试文件 test-source-file.txt

touch test-source-file.txt

echo "Hello \r\nRocketMQ\r\n Connect" >> test-source-file.txt

curl -X POST -H "Content-Type: application/json" http://127.0.0.1:8082/connectors/fileSourceConnector -d '{"connector.class":"org.apache.rocketmq.connect.file.FileSourceConnector","filename":"test-source-file.txt","connect.topicname":"fileTopic"}'

看到以下日志说明 file source connector 启动成功了

tail -100f ~/logs/rocketmqconnect/connect_runtime.log

2019-07-16 11:18:39 INFO pool-7-thread-1 - Source task start, config:{"properties":{"source-record-...

source connector配置说明

key nullable default description
connector.class false 实现 Connector接口的类名称（包含包名）
filename false 数据源文件名称
connect.topicname false 同步文件数据所需topic

5.启动sink connector

curl -X POST -H "Content-Type: application/json" http://127.0.0.1:8082/connectors/fileSinkConnector -d '{"connector.class":"org.apache.rocketmq.connect.file.FileSinkConnector","filename":"test-sink-file.txt","connect.topicnames":"fileTopic"}'

cat test-sink-file.txt

tail -100f ~/logs/rocketmqconnect/connect_runtime.log

看到以下日志说明file sink connector 启动成功了

2019-07-16 11:24:58 INFO pool-7-thread-2 - Sink task start, config:{"properties":{"source-record-...

如果 test-sink-file.txt 生成并且与 source-file.txt 内容一样，说明整个流程正常运行。 文件内容可能顺序不一样，这主要是因为RocketMQ发到不同queue时，接收不同queue消息顺序可能也不一致导致的，是正常的。

sink connector配置说明

key nullable default description
connector.class false 实现Connector接口的类名称（包含包名）
filename false sink拉去的数据保存到文件
connect.topicnames false sink需要处理数据消息topics 
注：source/sink配置文件说明是以rocketmq-connect-sample为demo，不同source/sink connector配置有差异，请以具体source/sink connector 为准

6.停止connector

GET请求  
http://(your worker ip):(port)/connectors/(connector name)/stop

停止demo中的两个connector
curl     http://127.0.0.1:8082/connectors/fileSinkConnector/stop
curl     http://127.0.0.1:8082/connectors/fileSourceConnector/stop

看到以下日志说明connector停止成功了

Source task stop, config:{"properties":{"source-record-converter":"org.apache.rocketmq.connect.runtime.converter.JsonConverter","filename":"/home/zhoubo/IdeaProjects/my-new3-rocketmq-externals/rocketmq-connect/rocketmq-connect-runtime/source-file.txt","task-class":"org.apache.rocketmq.connect.file.FileSourceTask","topic":"fileTopic","connector-class":"org.apache.rocketmq.connect.file.FileSourceConnector","update-timestamp":"1564765189322"}}

7.停止Worker进程

sh bin/connectshutdown.sh

8.日志目录

${user.home}/logs/rocketmqconnect

9.配置文件

持久化配置文件默认目录 /tmp/storeRoot

key description
connectorConfig.json connector配置持久化文件
position.json source connect数据处理进度持久化文件
taskConfig.json task配置持久化文件
offset.json sink connect数据消费进度持久化文件
connectorStatus.json connector 状态持久化文件
taskStatus.json task 状态持久化文件

10.配置说明

可根据使用情况修改 RESTful 端口，storeRoot 路径，Nameserver 地址等信息

文件位置：work 启动目录下 conf/connect-standalone.conf

#current cluster node uniquely identifies
workerId=DEFAULT_WORKER_1

# Http prot for user to access REST API
httpPort=8082

# Local file dir for config store
storePathRootDir=/home/connect/storeRoot

#需要修改为自己的rocketmq nameserver 接入点
# RocketMQ namesrvAddr
namesrvAddr=127.0.0.1:9876  

#用于加载Connector插件，类似于jvm启动加载jar包或者class类，这里目录目录用于放Connector相关的实现插件，
支持文件和目录
# Source or sink connector jar file dir
pluginPaths=rocketmq-connect-sample/target/rocketmq-connect-sample-0.0.1-SNAPSHOT.jar

# 补充：将 Connector 相关实现插件保存到指定文件夹 
# pluginPaths=/usr/local/connector-plugins/*

11.其它restful接口

集群信息

  • 查看集群节点信息：
curl -X GET http://(your worker ip):(port)/getClusterInfo
  • 重新加载Connector插件目录下的Connector包：
curl -X GET http://(your worker ip):(port)/plugin/reload

Connector/Task管理

  • 创建或更新connector（存在且配置不同会更新，不存在创建）
curl -X GET http://(your worker ip):(port)/connectors/{connectorName}
  • Pause(暂停)指定的connector
curl -X GET http://(your worker ip):(port)/connectors/{connectorName}/pause
  • Resume(重启)指定的connector
curl -X GET http://(your worker ip):(port)/connectors/{connectorName}/resume
  • Pause(暂停)所有的connector
curl -X GET http://(your worker ip):(port)/connectors/pause/all
  • Resume(重启)所有的connector
curl -X GET http://(your worker ip):(port)/connectors/resume/all
  • 停止并删除指定的connector(谨慎使用)
curl -X GET http://(your worker ip):(port)/connectors/{connectorName}/stop
  • 停止并删除所有的connector(谨慎使用)
curl -X GET http://(your worker ip):(port)/connectors/stop/all
  • 列举集群中所有connector信息
curl -X GET http://(your worker ip):(port)/connectors/list
  • 根据connectorName获取connector的配置信息
curl -X GET http://(your worker ip):(port)/connectors/{connectorName}/config
  • 根据connectorName获取connector的状态
curl -X GET http://(your worker ip):(port)/connectors/{connectorName}/status
  • 根据connectorName获取connector下所有task信息
curl -X GET http://(your worker ip):(port)/connectors/{connectorName}/tasks
  • 根据connectorName和task id获取task状态
curl -X GET http://(your worker ip):(port)/connectors/{connectorName}/tasks/{task}/status
  • 获取当前worker分配的connector信息
curl -X GET http://(your worker ip):(port)/allocated/connectors
  • 获取当前worker分配的task信息
curl -X GET http://(your worker ip):(port)/allocated/tasks

从内存删除Connector配置信息（谨慎使用）：

curl -X GET http://(your worker ip):(port)/connectors/(connector name)/delete

12.Connector通用配置参数说明

key nullable default description
connector.class false 指定插件的class信息
max.tasks false 1 connector下运行的task的数量，根据情况而定
value.converter false org.apache.rocketmq.connect.runtime.converter.record.StringConverter ConnectRecord key的转换器
key.converter false org.apache.rocketmq.connect.runtime.converter.record.StringConverter ConnectRecord value的转换器
transforms true 配置的数据转换器，多个需要用 ","分割
errors.log.include.messages true false 是否将导致故障的connector异常信息纪录在日志中
errors.log.enable true false 如果为true，请将每个错误以及失败操作和问题记录的详细信息写入Connect应用程序日志。默认情况下，这是“false”，因此只报告不可容忍的错误。
errors.retry.timeout true 0 重试超时时间
errors.retry.delay.max.ms true 60000 重试延迟时间
errors.tolerance true ToleranceType.NONE 在Connector操作期间容忍错误的行为。“NONE”是默认值，表示任何错误都将导致连接器任务立即失败；"ALL“更改行为以跳过有问题的记录。

Transform配置案例[单个配置]

"transforms": "Replace",
"transforms.Replace.field.pattern": "company",
"transforms.Replace.field.replacement": "company02",
"transforms.Replace.class": "org.apache.rocketmq.connect.transforms.PatternRename$Value",

Transform配置案例[多个配置],会按照配置顺序依次执行

"transforms": "Replace, Replace02",
"transforms.Replace.field.pattern": "company",
"transforms.Replace.field.replacement": "company02",
"transforms.Replace.class": "org.apache.rocketmq.connect.transforms.PatternRename$Value",
"transforms.Replace02.field.pattern": "company02",
"transforms.Replace02.field.replacement": "company03",
"transforms.Replace02.class": "org.apache.rocketmq.connect.transforms.PatternRename$Value",

1.transforms: 为固定配置,不可变

2.Replace, Replace02: 为配置名称，可自定义, 多个用","分割，若为多个，下面配置均需重复配置

3.transforms.${transform-name}.class: 用此配置来表示transform的class

4.transforms.${transform-name}.{config.key}: transform中使用的多个配置项；

Transform接口

package io.openmessaging.connector.api.component;

import io.openmessaging.KeyValue;
import io.openmessaging.connector.api.data.ConnectRecord;

public interface Transform<R extends ConnectRecord> extends Component {
    default void validate(KeyValue config) {
    }

    R doTransform(R var1);
}

  1. 自定义Transform只需要实现以上接口即可，将自定义包按照到 [pluginPaths]指定的插件目录下面即可;

  2. 平台已经实现的transform在 transforms module下面，欢迎大家定义和扩展更加丰富的transform库

org.apache.rocketmq.connect.transforms.ChangeCase$Key org.apache.rocketmq.connect.transforms.ChangeCase$Value org.apache.rocketmq.connect.transforms.ExtractNestedField$Value org.apache.rocketmq.connect.transforms.ExtractNestedField$Key org.apache.rocketmq.connect.transforms.PatternRename$Key org.apache.rocketmq.connect.transforms.PatternRename$Value org.apache.rocketmq.connect.transforms.PatternFilter$Key org.apache.rocketmq.connect.transforms.PatternFilter$Value org.apache.rocketmq.connect.transforms.SetMaximumPrecision$Key org.apache.rocketmq.connect.transforms.SetMaximumPrecision$Value

Source Connector特殊配置

key nullable default description
connect.topicname true 指定数据写入的topic，若不配置则直接取position中key为topic的值，若取不到则抛出异常
ordering.msg.enable true false 当目标集群不为rocketmq 5.x时，顺序消息会乱序，若设置为true，才能支持顺序，但会降低connector性能

Sink Connector特殊配置

key nullable default description
connect.topicnames false sink订阅的topic配置多个用","分割
task.group.id true connect-{connectName} sink订阅topic的group id
errors.deadletterqueue.topic.name true 错误队列的topic配置
errors.deadletterqueue.read.queue.nums true 8 错误队列创建时指定读的队列数
errors.deadletterqueue.write.queue.nums true 8 错误队列创建时指定写的队列数
errors.deadletterqueue.context.properties.enable true false 上报错误数据是否包含错误的详细的属性信息

13.Worker配置参数说明

❌ 表示该模式下不需要配置此选项 ✅ 表示该模式下配置生效

key nullable standalone distributed default description
workerId false ✅ ️ ✅ DEFAULT_WORKER_1 集群节点唯一标识
namesrvAddr false ✅ ✅ localhost:9876 RocketMQ Name Server地址列表，多个NameServer地址用分号隔开
httpPort false ✅ ✅ 8082 runtime提供restful接口服务端口
pluginPaths false ✅ ✅ source或者sink目录，启动runtime时加载
storePathRootDir true ✅ ✅ /tmp/connectorStore 持久化文件保存目录
positionStoreTopic true ❌ ✅ connector-position-topic source端position变更通知topic
positionPersistInterval true ✅ ✅ 20s source端持久化position数据间隔
configStoreTopic true ❌ ✅ connector-config-topic 集群connector配置变更通知topic
configPersistInterval true ❌ ✅ 20s 集群中配置信息持久化间隔
connectStatusTopic true ❌ ✅ connect-status-topic connector和task状态变更通知
statePersistInterval true ❌ ✅ 20s connector及task状态持久化间隔
rmqProducerGroup true ✅ ✅ defaultProducerGroup Producer组名，多个Producer如果属于一个应用，发送同样的消息，则应该将它们归为同一组
rmqConsumerGroup true ✅ ✅ defaultConsumerGroup Consumer组名，多个Consumer如果属于一个应用，发送同样的消息，则应该将它们归为同一组
maxMessageSize true ✅ ✅ 4MB RocketMQ最大消息大小
operationTimeout true ✅ ✅ 3s Producer发送消息超时时间
rmqMaxRedeliveryTimes true ✅ ✅ 最大重新消费次数
rmqMessageConsumeTimeout true ✅ ✅ 3s Consumer超时时间
rmqMaxConsumeThreadNums true ✅ ✅ 32 Consumer客户端最大线程数
rmqMinConsumeThreadNums true ✅ ✅ 1 Consumer客户端最小线程数
allocTaskStrategy true ✅ ✅ org.apache.rocketmq.connect.
runtime.service.strategy.
DefaultAllocateConnAndTaskStrategy		 负载均衡策略类
offsetCommitTimeoutMsConfig true ✅ ✅ 5000L source和sink offset提交超时时间
offsetCommitIntervalMsConfig true ✅ ✅ 60000L source和sink offset提交间隔时间配置
keyConverter true ✅ ✅ org.apache.rocketmq.connect.runtime.converter.record.StringConverter 集群配置默认 key 转换器
valueConverter true ✅ ✅ org.apache.rocketmq.connect.runtime.converter.record.StringConverter 集群配置默认 Value 转换器

allocTaskStrategy说明

该参数默认可省略，这是一个可选参数，目前选项如下：

  • 默认值
org.apache.rocketmq.connect.runtime.service.strategy.DefaultAllocateConnAndTaskStrategy
  • 一致性Hash
org.apache.rocketmq.connect.runtime.service.strategy.AllocateConnAndTaskStrategyByConsistentHash

更多集群和负载均衡文档

负载均衡

14.runtime支持JVM参数说明

key nullable default description
rocketmq.runtime.cluster.rebalance.waitInterval true 20s 负载均衡间隔
rocketmq.runtime.max.message.size true 4M Runtime限制最大消息大小
virtualNode true 1 一致性hash负载均衡的虚拟节点数
consistentHashFunc true MD5Hash 一致性hash负载均衡算法实现类

VirtualNode

一致性hash中虚拟节点数

consistentHashFunc

hash算法具体实现类,可以自己实现，在后续版本中会增加更多策略，该类应该实现

org.apache.rocketmq.common.consistenthash.HashFunction;

package org.apache.rocketmq.common.consistenthash;

public interface HashFunction {
    long hash(String var1);
}

默认情况下采用的是MD5Hash算法

//default hash function
private static class MD5Hash implements HashFunction {
    MessageDigest instance;

    public MD5Hash() {
        try {
            instance = MessageDigest.getInstance("MD5");
        } catch (NoSuchAlgorithmException e) {
        }
    }

    @Override
    public long hash(String key) {
        instance.reset();
        instance.update(key.getBytes());
        byte[] digest = instance.digest();

        long h = 0;
        for (int i = 0; i < 4; i++) {
            h <<= 8;
            h |= ((int) digest[i]) & 0xFF;
        }
        return h;
    }
}

开发指南

如何在IDE中启动Connect Worker ?

单机模式启动Connect Worker

Main Class配置:

org.apache.rocketmq.connect.runtime.StandaloneConnectStartup

img_2.png

Program arguments配置

-c ${user path}/rocketmq-connect/distribution/conf/connect-standalone.conf

Environment variables配置

CONNECT_HOME=${user path}/rocketmq-connect/distribution

###集群模式启动Connect Worker

Main Class配置

org.apache.rocketmq.connect.runtime.DistributedConnectStartup

Program arguments配置

-c ${user path}/rocketmq-connect/distribution/conf/connect-distributed.conf

Environment variables配置

CONNECT_HOME=${user path}/rocketmq-connect/distribution