PostgreSQL DBA(39) - PG 12 Functions for partition

In previous versions of PG 12, to obtain information such as the partition and sub-partition in the partition table you need to use a recursive CTE query to get the script, unintuitive and cumbersome, add the PG 12 in the pg_partition_tree and pg_partition_root system functions are used to get a partition tree and the root partition of relation.

Here a simple example will be described.

Test Script

-- Hash Partitiondrop table if exists t_hash1;create table t_hash1 (c1 int not null,c2  varchar(40),c3 varchar(40)) partition by hash(c1);-- Level 1create table t_hash1_1 partition of t_hash1 for values with (modulus 6,remainder 0) partition by hash(c1);create table t_hash1_2 partition of t_hash1 for values with (modulus 6,remainder 1) partition by hash(c1);create table t_hash1_3 partition of t_hash1 for values with (modulus 6,remainder 2);create table t_hash1_4 partition of t_hash1 for values with (modulus 6,remainder 3);create table t_hash1_5 partition of t_hash1 for values with (modulus 6,remainder 4);create table t_hash1_6 partition of t_hash1 for values with (modulus 6,remainder 5);-- Level 2create table t_hash1_1_1 partition of t_hash1_1 for values with (modulus 2,remainder 0);create table t_hash1_1_2 partition of t_hash1_1 for values with (modulus 2,remainder 1);create table t_hash1_2_1 partition of t_hash1_2 for values with (modulus 2,remainder 0);create table t_hash1_2_2 partition of t_hash1_2 for values with (modulus 2,remainder 1);

t_hash1 Hash is a partition table, there are sub-partition 6, wherein the sub-partition is the partition table t_hash1_2 t_hash1_1 and, respectively, two partitions.

Zhengzhou infertility hospital: http: //byby.zztjyy.com/yiyuanzaixian/zztjyy//

In PG 11, you need to use a recursive CTE query for information about the partition:

-- PG11WITH RECURSIVE partition_info      (relid,             -- oid       relname,            -- 名称       relsize,            -- 大小       relispartition,     -- 是否分区表       relkind) AS (    SELECT oid AS relid,           relname,           pg_relation_size(oid) AS relsize,           relispartition,           relkind    FROM pg_catalog.pg_classWHERE relname = 't_hash1' AND -- 最顶层的分区表      relkind = 'p'   UNION ALL    SELECT         c.oid AS relid,         c.relname AS relname,         pg_relation_size(c.oid) AS relsize,         c.relispartition AS relispartition,         c.relkind AS relkind    FROM partition_info AS p,         pg_catalog.pg_inherits AS i,         pg_catalog.pg_class AS c    WHERE p.relid = i.inhparent AND -- 从最顶层的分区表(即t_hash1)开始递归         c.oid = i.inhrelid AND -- 寻找子分区         c.relispartition -- 分区表标记  )SELECT * FROM partition_info; relid |   relname   | relsize | relispartition | relkind -------+-------------+---------+----------------+--------- 57457 | t_hash1     |       0 | f              | p 57466 | t_hash1_3   |       0 | t              | r 57469 | t_hash1_4   |       0 | t              | r 57472 | t_hash1_5   |       0 | t              | r 57475 | t_hash1_6   |       0 | t              | r 57460 | t_hash1_1   |       0 | t              | p 57463 | t_hash1_2   |       0 | t              | p 57487 | t_hash1_2_2 |       0 | t              | r 57478 | t_hash1_1_1 |       0 | t              | r 57481 | t_hash1_1_2 |       0 | t              | r 57484 | t_hash1_2_1 |       0 | t              | r(11 rows)

In the PG 12, the system functions can be used directly to obtain information:

testdb=# \sf pg_partition_treeCREATE OR REPLACE FUNCTION pg_catalog.pg_partition_tree(rootrelid regclass, OUT relid regclass, OUT parentrelid regclass, OUT isleaf boolean, OUT level integer) RETURNS SETOF record LANGUAGE internal PARALLEL SAFE STRICTAS $function$pg_partition_tree$function$testdb=# select pg_partition_tree('t_hash1');      pg_partition_tree      ----------------------------- (t_hash1,,f,0) (t_hash1_1,t_hash1,f,1) (t_hash1_2,t_hash1,f,1) (t_hash1_3,t_hash1,t,1) (t_hash1_4,t_hash1,t,1) (t_hash1_5,t_hash1,t,1) (t_hash1_6,t_hash1,t,1) (t_hash1_1_1,t_hash1_1,t,2) (t_hash1_1_2,t_hash1_1,t,2) (t_hash1_2_1,t_hash1_2,t,2) (t_hash1_2_2,t_hash1_2,t,2)(11 rows)

Information returned includes: 
RELID -> the partition RELID 
parentrelid -> parent partition 
isleaf -> if the leaf node 
level -> level

通过pg_partition_root可以获取分区表的root节点

testdb=# \sf pg_partition_rootCREATE OR REPLACE FUNCTION pg_catalog.pg_partition_root(regclass) RETURNS regclass LANGUAGE internal IMMUTABLE PARALLEL SAFE STRICTAS $function$pg_partition_root$function$testdb=# select pg_partition_root('t_hash1_2_2'); pg_partition_root ------------------- t_hash1(1 row)

参考资料 
Postgres 12 highlight - Functions for partitions


Guess you like

Origin blog.51cto.com/14339005/2412111