Flink 对接 kafka
- 2022 年 3 月 27 日
本文字数:6197 字
阅读完需:约 20 分钟
本篇文章我们用 Flink Kafka Connector 对接 Kafka,实现一个简单的报警业务。我们暂时不去谈论理论,先上手实现这个简单的需求。
flink-connector-kafka 是 flink 内置的 Kafka 连接器,包含了从 topic 读取数据的 Flink Kafka Consumer 和 向 topic 写入数据的 flink kafka producer,除了基本功能外还提供了基于 checkpoint 机制提供了完美的容错能力。
本文基于 flink 1.10.1 和 flink-connector-kafka-0.10_2.11 版本,pom 如下:
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-connector-kafka-0.10_2.11</artifactId>
<version>1.10.0</versio>
</dependency>
以企业常见的预警业务为例,本文要实现的业务逻辑很简单,当设备上报的油桶余量不足 10%时,便生成一个报警,这里我们将报警写入 MySQL,以供 web 业务端展示报警报表。
首先我们用网络数据调试器向网关模拟发送数据,网关会将数据解析后写入 kafka
kafka-console-consumer --bootstrap-server cdh1.macro.com:9092,cdh2.macro.com:9092,cdh3.macro.com:9092 --from-beginning --topic fill
{"addTime":1593147840000,"currentAmount":0.3,"devId":"XT365-000170","devStatus":"1","ifOffline":"1","ip":"127.0.0.1","leftTankAmount":5,"realTotalAmount":2377.39,"registerTime":1606658457000,"settingAmount":0.3,"tankCapacity":1000,"totalAmount":2017.9315}
{"addTime":1593147840000,"currentAmount":0.3,"devId":"XT365-000170","devStatus":"1","ifOffline":"1","ip":"127.0.0.1","leftTankAmount":5,"realTotalAmount":2377.69,"registerTime":1606658458000,"settingAmount":0.3,"tankCapacity":1000,"totalAmount":2017.9315}
^C20/11/29 23:26:55 INFO internals.ConsumerCoordinator: [Consumer clientId=consumer-console-consumer-82199-1, groupId=console-consumer-82199] Revoke previously assigned partitions fill-0
20/11/29 23:26:55 INFO internals.AbstractCoordinator: [Consumer clientId=consumer-console-consumer-82199-1, groupId=console-consumer-82199] Member consumer-console-consumer-82199-1-aa5fc2e6-1f06-4714-9d89-fe080a9400e2 sending LeaveGroup request to coordinator cdh2.macro.com:9092 (id: 2147483598 rack: null) due to the consumer is being closed
Processed a total of 1200 messages
可以看到我们已经向 kafka 生产了 1200 条数据了
接下来我们写一段代码来消费 kafka 数据,并将报警结果写入 MySQL
import com.alibaba.fastjson.JSONObject;
import com.iiot.bean.InSufficient;
import com.iiot.commCommon.Fill;
import com.iiot.jdbc.MySQLSinks;
import org.apache.flink.api.common.functions.MapFunction;
import org.apache.flink.api.common.serialization.SimpleStringSchema;
import org.apache.flink.streaming.api.datastream.DataStream;
import org.apache.flink.streaming.api.datastream.DataStreamSource;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.windowing.windows.TimeWindow;
import org.apache.flink.streaming.api.windowing.time.Time;
import java.util.List;
import org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumer010;
import org.apache.flink.util.Collector;
import org.apache.flink.shaded.guava18.com.google.common.collect.Lists;
import org.apache.flink.streaming.api.functions.windowing.AllWindowFunction;
import java.util.Properties;
public class InSufficientOilAlarms {
public static void main(String[] args) throws Exception{
//构建流执行环境
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
//kafka
Properties prop = new Properties();
prop.put("bootstrap.servers", "cdh1.macro.com:9092,cdh2.macro.com:9092,cdh3.macro.com:9092");
// prop.put("zookeeper.connect", "localhost:2181");
prop.put("group.id", "fill6");
prop.put("key.serializer", "org.apache.kafka.common.serialization.StringDeserializer");
prop.put("value.serializer", "org.apache.kafka.common.serialization.StringDeserializer");
prop.put("auto.offset.reset", "earliest");
DataStreamSource<String> stream = env
.addSource(new FlinkKafkaConsumer010<String>(
"fill",
new SimpleStringSchema(), prop)).
//单线程打印,控制台不乱序,不影响结果
setParallelism(1);
//从kafka里读取数据,转换成Person对象
DataStream<Fill> dataStream = stream.map(value ->
JSONObject.parseObject(value, Fill.class)
);
SingleOutputStreamOperator<InSufficient> result = dataStream.map(new MapFunction<Fill, InSufficient>() {
@Override
public InSufficient map(Fill fill) throws Exception {
InSufficient inSufficient = new InSufficient();
Float leftTankAmount = fill.getLeftTankAmount();
Float tankCapacity = fill.getTankCapacity();
String devCode = fill.getDevId();
long timeBegin = fill.getAddTime().getTime();
System.out.println("devCode:-------------------------------------------------" + devCode);
String alarmType = "";
if ((leftTankAmount / tankCapacity) < 0.1 ) {
alarmType = "inSufficientOil";
inSufficient.setDev_code(devCode);
inSufficient.setCreateTime(System.currentTimeMillis());
inSufficient.setTimeBegin(timeBegin);
inSufficient.setAlarmType(alarmType);
inSufficient.setRemainAmount(leftTankAmount);
}
return inSufficient;
}
}
);
//收集5秒钟的总数
result.timeWindowAll(Time.seconds(5L)).
apply(new AllWindowFunction<InSufficient, List<InSufficient>, TimeWindow>() {
@Override
public void apply(TimeWindow timeWindow, Iterable<InSufficient> iterable, Collector<List<InSufficient>> out) throws Exception {
List<InSufficient> inSufficients = Lists.newArrayList(iterable);
if(inSufficients.size() > 0) {
System.out.println("5秒的总共收到的条数:" + inSufficients.size());
out.collect(inSufficients);
}
}
})
//sink 到数据库
.addSink(new MySQLSinks());
//打印到控制台
//.print();
env.execute("kafka 消费任务开始");
}
}
将项目打包,传到集群中,用 Flink on YARN 的方式运行作业
[root@cdh3 bin]# flink run -m yarn-cluster -c com.iiot.alarm.InSufficientOilAlarms /data0/flinkdemo/stream-1.0-SNAPSHOT-jar-with-dependencies.jar
20/11/30 01:40:15 INFO cli.CliFrontend: --------------------------------------------------------------------------------
20/11/30 01:40:15 INFO cli.CliFrontend: Starting Command Line Client (Version: 1.10.0-csa1.2.0.0, Rev:04dddd1, Date:29.05.2020 @ 14:54:45 UTC)
20/11/30 01:40:15 INFO cli.CliFrontend: OS current user: root
20/11/30 01:40:16 INFO cli.CliFrontend: Current Hadoop/Kerberos user: hdfs
20/11/30 01:40:16 INFO cli.CliFrontend: JVM: Java HotSpot(TM) 64-Bit Server VM - Oracle Corporation - 1.8/25.171-b11
20/11/30 01:40:16 INFO cli.CliFrontend: Maximum heap size: 3531 MiBytes
20/11/30 01:40:16 INFO cli.CliFrontend: JAVA_HOME: /usr/java/latest
20/11/30 01:40:16 INFO cli.CliFrontend: Hadoop version: 2.7.5
20/11/30 01:40:16 INFO cli.CliFrontend: JVM Options:
20/11/30 01:40:16 INFO cli.CliFrontend: -Datlas.conf=/etc/atlas/conf/
20/11/30 01:40:16 INFO cli.CliFrontend: -Dlog.file=/var/log/flink/flink-root-client-cdh3.macro.com.log
20/11/30 01:40:16 INFO cli.CliFrontend: -Dlog4j.configuration=file:/etc/flink/conf/log4j-cli.properties
20/11/30 01:40:16 INFO cli.CliFrontend: -Dlogback.configurationFile=file:/etc/flink/conf/logback.xml
20/11/30 01:40:16 INFO cli.CliFrontend: Program Arguments:
20/11/30 01:40:16 INFO cli.CliFrontend: run
20/11/30 01:40:16 INFO cli.CliFrontend: -m
20/11/30 01:40:16 INFO cli.CliFrontend: yarn-cluster
20/11/30 01:40:16 INFO cli.CliFrontend: -c
20/11/30 01:40:16 INFO cli.CliFrontend: com.iiot.alarm.InSufficientOilAlarms
20/11/30 01:40:16 INFO cli.CliFrontend: /data0/flinkdemo/stream-1.0-SNAPSHOT-jar-with-dependencies.jar
20/11/30 01:40:51 INFO zookeeper.ZooKeeper: Client environment:java.io.tmpdir=/tmp
20/11/30 01:40:51 INFO zookeeper.ZooKeeper: Client environment:java.compiler=<NA>
20/11/30 01:40:51 INFO zookeeper.ZooKeeper: Client environment:os.name=Linux
20/11/30 01:40:51 INFO zookeeper.ZooKeeper: Client environment:os.arch=amd64
20/11/30 01:40:51 INFO zookeeper.ZooKeeper: Client environment:os.version=3.10.0-327.el7.x86_64
20/11/30 01:40:51 INFO zookeeper.ZooKeeper: Client environment:user.name=root
20/11/30 01:40:51 INFO zookeeper.ZooKeeper: Client environment:user.home=/root
20/11/30 01:40:51 INFO zookeeper.ZooKeeper: Client environment:user.dir=/opt/cloudera/parcels/FLINK-1.10.0-csa1.2.0.0-cdh7.1.1.0-565-3454809/bin
20/11/30 01:40:51 INFO zookeeper.ZooKeeper: Client environment:os.memory.free=134MB
20/11/30 01:40:51 INFO zookeeper.ZooKeeper: Client environment:os.memory.max=3531MB
20/11/30 01:40:51 INFO zookeeper.ZooKeeper: Client environment:os.memory.total=359MB
20/11/30 01:40:51 INFO utils.Compatibility: Using emulated InjectSessionExpiration
20/11/30 01:40:51 INFO imps.CuratorFrameworkImpl: Starting
20/11/30 01:40:51 INFO zookeeper.ZooKeeper: Initiating client connection, connectString=cdh1.macro.com:2181,cdh2.macro.com:2181,cdh3.macro.com:2181 sessionTimeout=60000 watcher=org.apache.flink.shaded.curator.org.apache.curator.ConnectionState@1460c81d
20/11/30 01:40:51 INFO common.X509Util: Setting -D jdk.tls.rejectClientInitiatedRenegotiation=true to disable client-initiated TLS renegotiation
20/11/30 01:40:51 INFO zookeeper.ClientCnxnSocket: jute.maxbuffer value is 4194304 Bytes
20/11/30 01:40:51 INFO zookeeper.ClientCnxn: zookeeper.request.timeout value is 0. feature enabled=
20/11/30 01:40:51 WARN zookeeper.ClientCnxn: SASL configuration failed: javax.security.auth.login.LoginException: No JAAS configuration section named 'Client' was found in specified JAAS configuration file: '/tmp/jaas-8202592158525653501.conf'. Will continue connection to Zookeeper server without SASL authentication, if Zookeeper server allows it.
20/11/30 01:40:51 INFO zookeeper.ClientCnxn: Opening socket connection to server cdh1.macro.com/192.168.0.171:2181
20/11/30 01:40:51 INFO zookeeper.ClientCnxn: Socket connection established, initiating session, client: /192.168.0.208:38183, server: cdh1.macro.com/192.168.0.171:2181
20/11/30 01:40:51 ERROR curator.ConnectionState: Authentication failed
20/11/30 01:40:51 INFO imps.CuratorFrameworkImpl: Default schema
20/11/30 01:40:51 INFO zookeeper.ClientCnxn: Session establishment complete on server cdh1.macro.com/192.168.0.171:2181, sessionid = 0x3008be9995512b4, negotiated timeout = 60000
20/11/30 01:40:51 INFO state.ConnectionStateManager: State change: CONNECTED
20/11/30 01:40:51 INFO imps.EnsembleTracker: New config event received: {server.1=cdh2.macro.com:3181:4181:participant, version=0, server.3=cdh1.macro.com:3181:4181:participant, server.2=cdh3.macro.com:3181:4181:participant}
20/11/30 01:40:51 ERROR imps.EnsembleTracker: Invalid config event received: {server.1=cdh2.macro.com:3181:4181:participant, version=0, server.3=cdh1.macro.com:3181:4181:participant, server.2=cdh3.macro.com:3181:4181:participant}
20/11/30 01:40:51 INFO imps.EnsembleTracker: New config event received: {server.1=cdh2.macro.com:3181:4181:participant, version=0, server.3=cdh1.macro.com:3181:4181:participant, server.2=cdh3.macro.com:3181:4181:participant}
20/11/30 01:40:51 ERROR imps.EnsembleTracker: Invalid config event received: {server.1=cdh2.macro.com:3181:4181:participant, version=0, server.3=cdh1.macro.com:3181:4181:participant, server.2=cdh3.macro.com:3181:4181:participant}
20/11/30 01:40:52 INFO leaderretrieval.ZooKeeperLeaderRetrievalService: Starting ZooKeeperLeaderRetrievalService /leader/rest_server_lock.
可以在 YARN 作业中看到 Flink 的做作业一直在运行。
flink dashboard 也可以看到作业一直在运行:
进入 YARN reourcemanager 里面查看作业运行日志:
可以看到 MySQL 已经插入数据了。
版权声明: 本文为 InfoQ 作者【云原生】的原创文章。
原文链接:【http://xie.infoq.cn/article/770c33113bddb55abd71e1766】。文章转载请联系作者。
云原生
云原生专家 2018.03.23 加入
云原生专家,专注云原生,个人公众号——云原生
评论