flink 并行度设置及优先级

1 优先级2 设置3 说明
版本：flink-1.10.1

算子级别 > 全局级别 > job提交时 > 配置文件

(1) 算子级别

DataStream> resultStream = inputDataStream.flatMap(new WordCount.MyFlatMapper())
    .keyBy(0)
    .sum(1).setParallelism(2);

(2) 全局级别

StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
env.setParallelism(4);

(3) job提交时

命令行

./flink run -c com.atguigu.wc.StreamWordCount -p 3
FlinkTutorial-1.0-SNAPSHOT-jar-with-dependencies.jar --host lcoalhost –port 7777

web

(4) 配置文件

[root@bigdata101 flink-1.10.1]# vi conf/flink-conf.yaml 
parallelism.default: 1

socketTextStream 读取数据流并行度为 1

Flat Map 采用 job 提交时设置的并行度 3

Keyed Aggregation 采用在算子设置的并行度 2

Sink: Print to Std.Out 采用在配置文件中设置的并行度 1