用seatunnel替代logstash，把数据从kafka抽取到ES

seatunnel(2.1.3)调用spark-sql(2.4)、flink-sql(1.14)对结构化数据进行处理；能够通过配置，在一个任务里调度多个source和sink

一、为spark structured streaming任务添加对ES7的支持

在seatunnel源码里升级elasticsearch-spark组件，添加spark-catalyst的依赖后，重新打包

    <properties>
        <elasticsearch7.client.version>7.16.3elasticsearch7.client.version>
        <elasticsearch-spark.version>7.16.3elasticsearch-spark.version>
    properties>
    <dependencyManagement>
        <dependencies>
            <dependency>
                <groupId>org.apache.sparkgroupId>
                <artifactId>spark-catalyst_${scala.binary.version}artifactId>
                <version>${spark.version}version>
                <scope>${spark.scope}scope>
            dependency>
        dependencies>
    dependencyManagement>

二、配置任务

1、配置env

指定spark任务启动参数

env {
  spark.app.name = "seatunnel"
  spark.yarn.am.memory = "2g"
  spark.executor.memory = "3g"
  spark.executor.instances = 3
  spark.streaming.batchDuration = 2
}

2、配置source

定义临时动态表，kafka consumer官方配置参数都添加在consumer.XXX

source {
    kafkaStream {        
        topics = "monitor-log"
        consumer.bootstrap.servers = "10.0.0.1:9092,10.0.0.2:9092"
        consumer.group.id = "seatunnel"
        consumer.auto.offset.reset = "latest"
        consumer.heartbeat.interval.ms = 2000
        consumer.session.timeout.ms = 6000
        consumer.reconnect.backoff.ms = 2000
        consumer.request.timeout.ms = 30000
        consumer.fetch.min.bytes = 10485760
        consumer.fetch.max.wait.ms = 1000
        consumer.max.partition.fetch.bytes = 20971520
        consumer.max.poll.interval.ms = 30000
        consumer.max.poll.records = 1000
        # 指定临时动态表的名字
        result_table_name = "monitor_log_kafka"
    }
}

3、配置transform

把数据从一个临时表转换到另一个临时表

transform {
    json {
        source_table_name = "monitor_log_kafka"
        # kafka动态数据表里有topic和raw_message两个字段
        source_field = "raw_message"

        # 文件放在plugins/json/files/schemas目录下
        # 如果采集到的数据，和样例数据里的字段格式不一致，则转化为样例数据里的格式
        schema_file = "monitor-log.json"

        result_table_name = "monitor_log"
        # target_field为root，会舍弃原有的raw_message、topic字段；
        # 要保留raw_message，需指定一个非root的target_field，来存储json反序列化结果
        target_field = "__root__"
    }
    sql {
        # sql里可直接引用前面定义的各个临时表
        sql = "SELECT * FROM monitor_log WHERE app is not null AND type = 'svc_link'"
        result_table_name = "svc_link"
    }
    sql {
        sql = "SELECT * FROM monitor_log WHERE app is not null AND type <> 'svc_link'"
        result_table_name = "app_log"
    }
}

4、配置sink

把数据从临时表存入目标数据库

sink {
    elasticsearch {
        source_table_name = "svc_link"

        hosts = ["10.0.1.1:9200", "10.0.1.2:9200"]
        index = "wzp_svc_link-${now}"
        index_time_format = "yyyy.MM.dd"
        es.net.ssl = false
        es.net.http.auth.user = "elastic"
        es.net.http.auth.pass = "elastic"
        es.batch.size.bytes = "100mb"
        es.batch.size.entries = 10000
        es.batch.write.refresh = false
        es.write.rest.error.handlers = "log" 
        es.write.rest.error.handler.log.logger.name = "BulkErrors"
        es.write.rest.error.handler.log.logger.level = "WARN"
    }
    elasticsearch {
        source_table_name = "app_log"
        # 存到另一个ES库或index
    }
}

三、启动任务

export JAVA_HOME=/etc/alternatives/jre
export HADOOP_HOME=/apps/svr/hadoop-2.9.2
export SPARK_HOME=/apps/svr/spark-2.4.8-bin-hadoop2.7

cd /apps/svr/apache-seatunnel-incubating-2.1.3
./bin/start-seatunnel-spark.sh --master yarn --deploy-mode cluster --config ./config/wzp.spark.streaming.conf