本文是《Flink的sink實戰》系列的第二篇,《Flink的sink實戰之一:初探》對sink有了基本的了解,本章來體驗将資料sink到kafka的操作;
版本和環境準備
本次實戰的環境和版本如下:
- JDK:1.8.0_211
- Flink:1.9.2
- Maven:3.6.0
- 作業系統:macOS Catalina 10.15.3 (MacBook Pro 13-inch, 2018)
- IDEA:2018.3.5 (Ultimate Edition)
- Kafka:2.4.0
-
Zookeeper:3.5.5
請確定上述環境和服務已經就緒;
源碼下載下傳
如果您不想寫代碼,整個系列的源碼可在GitHub下載下傳到,位址和連結資訊如下表所示:
這個git項目中有多個檔案夾,本章的應用在flinksinkdemo檔案夾下,如下圖紅框所示:
準備工作
正式編碼前,先去官網檢視相關資料了解基本情況:
位址:https://ci.apache.org/projects/flink/flink-docs-release-1.9/dev/connectors/kafka.html我這裡用的kafka是2.4.0版本,在官方文檔查找對應的庫和類,如下圖紅框所示:
kafka準備
- 建立名為test006的topic,有四個分區,參考指令:
./kafka-topics.sh --create --bootstrap-server 127.0.0.1:9092 --replication-factor 1 --partitions 4 --topic test006
- 在控制台消費test006的消息,參考指令:
./kafka-console-consumer.sh --bootstrap-server 127.0.0.1:9092 --topic test006
- 此時如果該topic有消息進來,就會在控制台輸出;
- 接下來開始編碼;
建立工程
- 用maven指令建立flink工程:
mvn archetype:generate -DarchetypeGroupId=org.apache.flink -DarchetypeArtifactId=flink-quickstart-java -DarchetypeVersion=1.9.2
- 根據提示,groupid輸入com.bolingcavalry,artifactid輸入flinksinkdemo,即可建立一個maven工程;
- 在pom.xml中增加kafka依賴庫:
org.apache.flink flink-connector-kafka_2.11 1.9.0
- 工程建立完成,開始編寫flink任務的代碼;
發送字元串消息的sink
先嘗試發送字元串類型的消息:
- 建立KafkaSerializationSchema接口的實作類,後面這個類要作為建立sink對象的參數使用:
package com.bolingcavalry.addsink;import org.apache.flink.streaming.connectors.kafka.KafkaSerializationSchema;import org.apache.kafka.clients.producer.ProducerRecord;import java.nio.charset.StandardCharsets;public class ProducerStringSerializationSchema implements KafkaSerializationSchema { private String topic; public ProducerStringSerializationSchema(String topic) { super(); this.topic = topic; } @Override public ProducerRecord serialize(String element, Long timestamp) { return new ProducerRecord(topic, element.getBytes(StandardCharsets.UTF_8)); }}
- 建立任務類KafkaStrSink,請注意FlinkKafkaProducer對象的參數,FlinkKafkaProducer.Semantic.EXACTLY_ONCE表示嚴格一次:
package com.bolingcavalry.addsink;import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;import org.apache.flink.streaming.connectors.kafka.FlinkKafkaProducer;import java.util.ArrayList;import java.util.List;import java.util.Properties;public class KafkaStrSink { public static void main(String[] args) throws Exception { final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment(); //并行度為1 env.setParallelism(1); Properties properties = new Properties(); properties.setProperty("bootstrap.servers", "192.168.50.43:9092"); String topic = "test006"; FlinkKafkaProducer producer = new FlinkKafkaProducer<>(topic, new ProducerStringSerializationSchema(topic), properties, FlinkKafkaProducer.Semantic.EXACTLY_ONCE); //建立一個List,裡面有兩個Tuple2元素 List list = new ArrayList<>(); list.add("aaa"); list.add("bbb"); list.add("ccc"); list.add("ddd"); list.add("eee"); list.add("fff"); list.add("aaa"); //統計每個單詞的數量 env.fromCollection(list) .addSink(producer) .setParallelism(4); env.execute("sink demo : kafka str"); }}
使用mvn指令編譯建構,在target目錄得到檔案 flinksinkdemo-1.0-SNAPSHOT.jar;在flink的web頁面送出 flinksinkdemo-1.0-SNAPSHOT.jar,并制定執行類,如下圖:
- 送出成功後,如果flink有四個可用slot,任務會立即執行,會在消費kafak消息的終端收到消息,如下圖:
- 任務執行情況如下圖:
發送對象消息的sink
再來嘗試如何發送對象類型的消息,這裡的對象選擇常用的Tuple2對象:
- 建立KafkaSerializationSchema接口的實作類,該類後面要用作sink對象的入參,請注意代碼中捕獲異常的那段注釋:生産環境慎用printStackTrace()!!!
package com.bolingcavalry.addsink;import org.apache.flink.api.java.tuple.Tuple2;import org.apache.flink.shaded.jackson2.com.fasterxml.jackson.core.JsonProcessingException;import org.apache.flink.shaded.jackson2.com.fasterxml.jackson.databind.ObjectMapper;import org.apache.flink.streaming.connectors.kafka.KafkaSerializationSchema;import org.apache.kafka.clients.producer.ProducerRecord;import javax.annotation.Nullable;public class ObjSerializationSchema implements KafkaSerializationSchema> { private String topic; private ObjectMapper mapper; public ObjSerializationSchema(String topic) { super(); this.topic = topic; } @Override public ProducerRecord serialize(Tuple2 stringIntegerTuple2, @Nullable Long timestamp) { byte[] b = null; if (mapper == null) { mapper = new ObjectMapper(); } try { b= mapper.writeValueAsBytes(stringIntegerTuple2); } catch (JsonProcessingException e) { // 注意,在生産環境這是個非常危險的操作, // 過多的錯誤列印會嚴重影響系統性能,請根據生産環境情況做調整 e.printStackTrace(); } return new ProducerRecord(topic, b); }}
- 建立flink任務類:
package com.bolingcavalry.addsink;import org.apache.flink.api.java.tuple.Tuple2;import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;import org.apache.flink.streaming.connectors.kafka.FlinkKafkaProducer;import java.util.ArrayList;import java.util.List;import java.util.Properties;public class KafkaObjSink { public static void main(String[] args) throws Exception { final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment(); //并行度為1 env.setParallelism(1); Properties properties = new Properties(); //kafka的broker位址 properties.setProperty("bootstrap.servers", "192.168.50.43:9092"); String topic = "test006"; FlinkKafkaProducer> producer = new FlinkKafkaProducer<>(topic, new ObjSerializationSchema(topic), properties, FlinkKafkaProducer.Semantic.EXACTLY_ONCE); //建立一個List,裡面有兩個Tuple2元素 List> list = new ArrayList<>(); list.add(new Tuple2("aaa", 1)); list.add(new Tuple2("bbb", 1)); list.add(new Tuple2("ccc", 1)); list.add(new Tuple2("ddd", 1)); list.add(new Tuple2("eee", 1)); list.add(new Tuple2("fff", 1)); list.add(new Tuple2("aaa", 1)); //統計每個單詞的數量 env.fromCollection(list) .keyBy(0) .sum(1) .addSink(producer) .setParallelism(4); env.execute("sink demo : kafka obj"); }}
- 像前一個任務那樣編譯建構,把jar送出到flink,并指定執行類是com.bolingcavalry.addsink.KafkaObjSink;
- 消費kafka消息的控制台輸出如下:
- 在web頁面可見執行情況如下:
至此,flink将計算結果作為kafka消息發送出去的實戰就完成了,希望能給您提供參考,接下來的章節,我們會繼續體驗官方提供的sink能力;