PG创建or修改索引
- 2019 年 10 月 4 日
- 筆記
pgsql中的索引不能重名，重名的创建失败。
创建二级索引的命令：create index CONCURRENTLY idx_abc on tb1(a,b);
注意：reindex 重建索引的过程是阻塞的，一般大表不建议使用这个命令，可以重建一个索引，然后删除老的索引。
下面看一个《PostgreSQL实战》书上page202的例子：
由于PG的MVCC机制，当运行大量的更新操作后，会有索引膨胀的现象。这时候 可以通过 create index concurrently 不阻塞查询和更新的情况下，在线重新重建索引，创建好索引之后，再删除原先的索引，减少索引的尺寸，提高查询速度。对于主键也可以使用这种方式进行，例如：
db1=# d testdata Table "public.testdata" Column | Type | Collation | Nullable | Default ----------+--------------+-----------+----------+--------- id | integer | | not null | course | integer | | | grade | numeric(4,2) | | | testtime | date | | | Indexes: "testdata_pkey" PRIMARY KEY, btree (id) db1=# create unique index concurrently on testdata using btree(id); db1=# select schemaname, relname, indexrelname, pg_relation_size (indexrelid) as index_size, idx_scan, idx_tup_read, idx_tup_fetch from pg_stat_user_indexes where indexrelname in (select indexname from pg_indexes where schemaname ='public' and tablename='testdata'); schemaname | relname | indexrelname | index_size | idx_scan | idx_tup_read | idx_tup_fetch ------------+----------+-----------------+------------+----------+--------------+--------------- public | testdata | testdata_pkey | 16384 | 2 | 5999998 | 5999998 public | testdata | testdata_id_idx | 16384 | 0 | 0 | 0 (2 rows) db1=# begin; BEGIN db1=# alter table testdata drop constraint testdata_pkey; ALTER TABLE db1=# alter table testdata add constraint testdata_id_idx primary key using index testdata_id_idx ; ALTER TABLE db1=# end; COMMIT db1=# d testdata Table "public.testdata" Column | Type | Collation | Nullable | Default ----------+--------------+-----------+----------+--------- id | integer | | not null | course | integer | | | grade | numeric(4,2) | | | testtime | date | | | Indexes: "testdata_id_idx" PRIMARY KEY, btree (id) db1=# select schemaname, relname, indexrelname, pg_relation_size (indexrelid) as index_size, idx_scan, idx_tup_read, idx_tup_fetch from pg_stat_user_indexes where indexrelname in (select indexname from pg_indexes where schemaname ='public' and tablename='testdata'); schemaname | relname | indexrelname | index_size | idx_scan | idx_tup_read | idx_tup_fetch ------------+----------+-----------------+------------+----------+--------------+--------------- public | testdata | testdata_id_idx | 16384 | 0 | 0 | 0 (1 row)
这样就完成了主键索引的重建，对于大规模的数据库集群，可以通过 pg_repack 工具进行定时的索引重建。
