kudu表数据的导入导出,处理null值问题

kudu to csv to hive to kudu

1 建表kudu

create table default.kudu_table(
id string not null,
ts bigint,
count bigint,
primary key(id)
)
PARTITION BY HASH (id) PARTITIONS 6 STORED AS KUDU;

2 插入数据

insert into default.kudu_table values (“1”, 11, 11), (“2”, 22, 22)

3 导出到linux

impala-shell -q ‘select * from default.kudu_table’ -B --output_delimiter="*" -o /data/tmp/test.csv

4 建表hive，临时表

create table default.kudu_table2(
id string,
ts bigint,
count bigint)row format delimited fields terminated by ‘*’;

处理null值问题

alter table default.kudu_table2 SET SERDEPROPERTIES(‘serialization.null.format’ = ‘NULL’);

5 csv导入hive

load data local inpath ‘/data/tmp/test.csv’ into table default.kudu_table2;

6 建表kudu，最终表

create table default.kudu_table3(
id string not null,
ts bigint,
count bigint,
primary key(id)
)
PARTITION BY HASH (id) PARTITIONS 6 STORED AS KUDU;

7 hive to kudu，刷新impala数据

insert into default.kudu_table3 select * from default.kudu_table2;