文档首页> 数据湖探索 DLI> 常见问题> Flink作业相关问题> 运维指导> Kafka Sink配置发送失败重试机制
更新时间：2022-05-05 GMT+08:00
查看PDF
分享

Kafka Sink配置发送失败重试机制

  • 问题描述

    用户执行Flink Opensource SQL, 采用Flink 1.10版本。Flink Sink写Kafka报错后作业失败：

    Caused by: org.apache.kafka.common.errors.NetworkException: The server disconnected before a response was received.
  • 问题原因

    由于CPU使用率过高，导致网络闪断。

  • 解决方案

    在SQL语句中配置发送失败重试：connector.properties.retries=5

    create table kafka_sink(
     car_type string
    , car_name string
    , primary key (union_id) not enforced
) with (
    "connector.type" = "upsert-kafka",
    "connector.version" = "0.11",
    "connector.properties.bootstrap.servers" = "xxxx:9092",
    "connector.topic" = "kafka_car_topic ",
    "connector.sink.ignore-retraction" = "true",
    "connector.properties.retries" = "5",
    "format.type" = "json"
);
父主题： 运维指导
分享：

运维指导所有常见问题

more

close