流式插入概述

kafka （映射为一个flink table source_table）-> flink （insert into target_table select * from source_table） -> hudi （映射为一个 flink table target_table）

一. Hudi流式插入案例1(datagen)

1.1 准备工作

# 启动yarn session(非root账户)
/home/flink-1.14.5/bin/yarn-session.sh -d  2>&1 &

# 在yarn session模式下启动Flink SQL
/home/flink-1.14.5/bin/sql-client.sh embedded -s yarn-session

1.2 源端准备

这里我们使用了数据生成器，datagen，下面有参数控制数据生成的频率。

-- sets up the result mode to tableau to show the results directly in the CLI
set execution.result-mode=tableau;

-- 设置checkpoint，不然会一直卡住
set execution.checkpointing.interval=10sec;

create table my_sourceT_12 (
   uuid varchar(200),
   name varchar(100),
   age  int,
   ts   timestamp(3)
) with (
   'connector' = 'datagen',
   'rows-per-second' =  '1'
)
;

image.png

1.3 目标端表准备

CREATE TABLE my_targetT_12(
  uuid VARCHAR(20) PRIMARY KEY NOT ENFORCED,
  name VARCHAR(10),
  age INT,
  ts TIMESTAMP(3)
)
WITH (
  'connector' = 'hudi',
  'path' = 'hdfs://hp5:8020/user/hudi_data/my_targetT_12',
  'table.type' = 'MERGE_ON_READ' -- this creates a MERGE_ON_READ table, by default is COPY_ON_WRITE
);

1.4 ETL准备

insert into my_targetT_12 (uuid, name, age, ts)
select uuid, name, age, ts
  from my_sourceT_12 ;

select * from my_targetT_12;

image.png

1.5 数据验证

在页面运行这个，依旧会有一个报错

select * from my_targetT_12;

报错:
这个报错是我测试环境的CPU资源不够导致的

org.apache.hadoop.yarn.server.nodemanager.containermanager.monitor.ContainersMonitorImpl: Skipping monitoring container  since CPU usage is not yet available.

HDFS查看数据:

image.png

1.6 通过SPARK SQL查看数据

连接Spark SQL:

# Spark 3.3
spark-sql --packages org.apache.hudi:hudi-spark3.3-bundle_2.12:0.12.0 \
--conf 'spark.serializer=org.apache.spark.serializer.KryoSerializer' \
--conf 'spark.sql.extensions=org.apache.spark.sql.hudi.HoodieSparkSessionExtension' \
--conf 'spark.sql.catalog.spark_catalog=org.apache.spark.sql.hudi.catalog.HoodieCatalog'

创建Hudi表:
建表的语法存在差异，需要进行调整，有的字段类型都不对应

CREATE TABLE my_targetT_12(
  uuid VARCHAR(20) ,
  name VARCHAR(10),
  age INT,
  ts TIMESTAMP
)
using hudi
location 'hdfs://hp5:8020/user/hudi_data/my_targetT_12';

查询Hudi表数据:

select * from my_targetT_12 limit 10;

测试记录:

image.png image.png

二. Hudi流式插入案例2(Kafka)

2.1 准备工作

# 启动yarn session(非root账户)
/home/flink-1.15.2/bin/yarn-session.sh -d  2>&1 &

# 在yarn session模式下启动Flink SQL
/home/flink-1.15.2/bin/sql-client.sh embedded -s yarn-session

2.2 源端准备

这里我们使用Kafka作为源端

2.2.1 创建kafka的topic (hudi_flink)

cd /opt/cloudera/parcels/CDH-6.3.1-1.cdh6.3.1.p0.1470567/lib/kafka/bin/
./kafka-topics.sh --zookeeper hp2:2181 --create --replication-factor 3 --partitions 3 --topic hudi_flink4

2.2.2 Flink SQL Client消费kafka数据

1. 将Flink连接Kafka的jar包放到Flink的lib目录
   https://nightlies.apache.org/flink/flink-docs-release-1.15/docs/connectors/table/kafka/

2. 创建kafka表

-- sets up the result mode to tableau to show the results directly in the CLI
set execution.result-mode=tableau;

-- 设置checkpoint，不然会一直卡住
set execution.checkpointing.interval=10sec;

CREATE TABLE hudi_flink_kafka_source4 (
  orderId STRING,
  userId STRING,
  orderTime STRING,
  ip STRING,
  orderMoney DOUBLE,
  orderStatus INT
) WITH (
  'connector' = 'kafka',
  'topic' = 'hudi_flink4',
  'properties.bootstrap.servers' = 'hp2:9092',
  'properties.group.id' = 'zqs-1004',
  'scan.startup.mode' = 'latest-offset',
  'format' = 'json',
  'json.fail-on-missing-field' = 'false',
  'json.ignore-parse-errors' = 'true'
);

3. 往kafka的topic插入数据

/opt/cloudera/parcels/CDH-6.3.1-1.cdh6.3.1.p0.1470567/lib/kafka/bin/kafka-console-producer.sh --broker-list hp2:9092 --topic hudi_flink4

{"orderId": "20211122103434136000001","userId": "300000971","orderTime": "2021-11-22 10:34:34.136","ip": "123.232.118.98","orderMoney": 485.48,"orderStatus": 0}
{"orderId": "20211122103434136000002","userId": "300000972","orderTime": "2021-11-22 10:34:34.136","ip": "123.232.118.98","orderMoney": 485.48,"orderStatus": 0}
{"orderId": "20211122103434136000003","userId": "300000973","orderTime": "2021-11-22 10:34:34.136","ip": "123.232.118.98","orderMoney": 485.48,"orderStatus": 0}

4. 在flink sql客户端查看数据消费

select * from hudi_flink_kafka_source4 ;

image.png

2.3 目标端表准备

CREATE TABLE hudi_flink_kafka_sink4 (
  orderId STRING PRIMARY KEY NOT ENFORCED,
  userId STRING,
  orderTime STRING,
  ip STRING,
  orderMoney DOUBLE,
  orderStatus INT,
  ts STRING,
  partition_day STRING
)
PARTITIONED BY (partition_day) 
WITH (
  'connector' = 'hudi',
  'path' = 'hdfs://hp5:8020/user/hudi_data/hudi_flink_kafka_sink4',
  'table.type' = 'MERGE_ON_READ',
  'write.operation' = 'upsert',
  'hoodie.datasource.write.recordkey.field'= 'orderId',
  'write.precombine.field' = 'ts',
  'write.tasks'= '1',
  'compaction.tasks' = '1', 
  'compaction.async.enabled' = 'true', 
  'compaction.trigger.strategy' = 'num_commits', 
  'compaction.delta_commits' = '1'
);

2.4 ETL准备

INSERT INTO hudi_flink_kafka_sink4 
SELECT
  orderId, userId, orderTime, ip, orderMoney, orderStatus,
  substring(orderId, 0, 17) AS ts, substring(orderTime, 0, 10) AS partition_day 
FROM hudi_flink_kafka_source4 ;

2.5 验证数据

HDFS:

image.png

参考:

1. https://blog.csdn.net/NC_NE/article/details/125705845