clickhouse PARTITION操作

关注微信公共号：小程在线

关注CSDN博客：程志伟的博客

ClickHouse server version 21.6.6

创建分区表

示例一：函数式分区

dblab-VirtualBox :) create table visits(userid int,visitdate date,website String) engine=MergeTree() PARTITION BY toYYYYMM(visitdate) order by userid;

CREATE TABLE visits
(
`userid` int,
`visitdate` date,
`website` String
)
ENGINE = MergeTree
PARTITION BY toYYYYMM(visitdate)
ORDER BY userid

Query id: 96c69dce-dcdf-4a19-8faf-ffdbff0fe8ac

Ok.

0 rows in set. Elapsed: 0.111 sec.
示例二：字段直接分区

dblab-VirtualBox :) create table test.user_action_log (　 `event_time` DateTime,　 `user_id` String,　 `user_name` String,　 `school_id` String,　 `school_name` String,　 `stay_time` Float64,　 `create_time` DateTime,　 `student_id` String,　 `grade` String,　 `date_key` String　)　ENGINE = MergeTree()　PARTITION BY (date_key)　ORDER BY (user_id,student_id,school_id)　PRIMARY KEY (user_id)　SETTINGS index_granularity = 8192;

CREATE TABLE test.user_action_log
(
`event_time` DateTime,
`user_id` String,
`user_name` String,
`school_id` String,
`school_name` String,
`stay_time` Float64,
`create_time` DateTime,
`student_id` String,
`grade` String,
`date_key` String
)
ENGINE = MergeTree
PARTITION BY date_key
PRIMARY KEY user_id
ORDER BY (user_id, student_id, school_id)
SETTINGS index_granularity = 8192

Query id: 41fa5885-328c-4c64-bf21-42c87d353766

Ok.

0 rows in set. Elapsed: 0.008 sec.

注：在创建表时一定要指定排序键，这一点非常重要，在批量插入数据时也可以对数据提前进行排序

表插入数据

insert into visits(userid,website,visitdate) values(100,'www.baidu.com','2020-06-01'),(100,'www.google.com','2020-07-02'),(100,'www.toutiao.com','2020-08-03');

select * from visits where visitdate='2020-07-02';

clickhouse想对于hive没有show partitions xxx来展示分区的操作我们可以通过查询数据库中的system.parts表来查看想要的信息。

dblab-VirtualBox :) select database,table,partition,partition_id,name ,path from system.parts where table='visits';

SELECT
database,
table,
partition,
partition_id,
name,
path
FROM system.parts
WHERE table = 'visits'

删除分区：

alter table visits drop partition 202007 ;

重置分区：

alter table visits CLEAR column website in partition 202007;

卸载和装载分区|数据块：

alter table visits DETACH partition 202007; --将分区数据移动到 detached ，并且忘记它

alter table visits ATTACH partition 202007; --将detached 目录中的分区重新添加到表中.