openGauss每日一练第19天 | 统计信息执行计划垃圾回收
学习目标
学习openGauss收集统计信息、打印执行计划、垃圾收集和checkpoint
课程学习
连接数据库
#第一次进入等待15秒#数据库启动中...
su - omm
gsql -r
1.准备数据
Create schema tpcds;
CREATE TABLE tpcds.customer_address
(
ca_address_sk integer NOT NULL ,
ca_address_id character(16),
ca_street_number character(10) ,
ca_street_name character varying(60) ,
ca_street_type character(15) ,
ca_suite_number character(10) ,
ca_city character varying(60) ,
ca_county character varying(30) ,
ca_state character(2) ,
ca_zip character(10) ,
ca_country character varying(20) ,
ca_gmt_offset numeric(5,2) ,
ca_location_type character(20)
);
insert into tpcds.customer_address values
(1, 'AAAAAAAABAAAAAAA', '18', 'Jackson', 'Parkway', 'Suite 280', 'Fairfield', 'Maricopa County', 'AZ', '86192' ,'United States', -7.00, 'condo'),
(2, 'AAAAAAAACAAAAAAA', '362', 'Washington 6th', 'RD', 'Suite 80', 'Fairview', 'Taos County', 'NM', '85709', 'United States', -7.00, 'condo'),
(3, 'AAAAAAAADAAAAAAA', '585', 'Dogwood Washington', 'Circle', 'Suite Q', 'Pleasant Valley', 'York County', 'PA', '12477', 'United States', -5.00, 'single family');
–使用序列的generate_series(1,N)函数对表插入数据
insert into tpcds.customer_address values(generate_series(10, 10000));
2.收集统计信息
–查看系统表中表的统计信息
select relname, relpages, reltuples from pg_class where relname = 'customer_address';
—使用ANALYZE VERBOSE语句更新统计信息，并输出表的相关信息
analyze VERBOSE tpcds.customer_address;
–查看系统表中表的统计信息
select relname, relpages, reltuples from pg_class where relname = 'customer_address';
3.打印执行计划
–使用默认的打印格式
SET explain_perf_mode=normal;
–显示表简单查询的执行计划
EXPLAIN SELECT * FROM tpcds.customer_address;
–以JSON格式输出的执行计划（explain_perf_mode为normal时）
EXPLAIN(FORMAT JSON) SELECT * FROM tpcds.customer_address;
–禁止开销估计的执行计划
EXPLAIN(COSTS FALSE)SELECT * FROM tpcds.customer_address;
–带有聚集函数查询的执行计划
EXPLAIN SELECT SUM(ca_address_sk) FROM tpcds.customer_address WHERE ca_address_sk<100;
–有索引条件的执行计划
create index customer_address_idx on tpcds.customer_address(ca_address_sk);
EXPLAIN SELECT * FROM tpcds.customer_address WHERE ca_address_sk<100;
4.垃圾收集
–VACUUM回收表或B-Tree索引中已经删除的行所占据的存储空间
update tpcds.customer_address set ca_address_sk = ca_address_sk + 1 where ca_address_sk <100;
VACUUM (VERBOSE, ANALYZE) tpcds.customer_address;
5.事务日志检查点
–检查点（CHECKPOINT）是一个事务日志中的点，所有数据文件都在该点被更新以反映日志中的信息，所有数据文件都将被刷新到磁盘
CHECKPOINT;
6.清理数据
drop schema tpcds cascade;
课后作业
1. 创建分区表，并用generate_series(1,N)函数对表插入数据
omm=# create table t1
omm-# (id int,
omm(# name char(30),
omm(# salary decimal(9,2),
omm(# did int)omm-#
partition by range(did)
omm-# (partition p1_t1 values less than(100),
omm(# partition p2_t1 values less than(200),
omm(# partition p3_t1 values less than(300));
CREATE TABLE
omm=# insert into t1 (did) values (generate_series(1,299));
INSERT 0 299
2. 收集表统计信息
omm=# select relname, relpages,reltuples from pg_class where relname='t1';
relname | relpages | reltuples
---------+----------+-----------
t1 | 0 | 0
(1 row)
omm=# analyze verbose t1;
INFO: analyzing "public.t1"(gaussdb pid=1)
INFO: ANALYZE INFO : "t1": scanned 1 of 1 pages, containing 99 live rows and 99 dead rows; 99 rows in sample, 99 estimated total rows(gaussdb pid=1)
INFO: ANALYZE INFO : "t1": scanned 1 of 1 pages, containing 100 live rows and 100 dead rows; 100 rows in sample, 100 estimated total rows(gaussdb pid=1)
INFO: ANALYZE INFO : "t1": scanned 1 of 1 pages, containing 100 live rows and 100 dead rows; 100 rows in sample, 100 estimated total rows(gaussdb pid=1)
ANALYZE
omm=# select relname, relpages, reltuples from pg_class where relname='t1';
relname | relpages | reltuples
---------+----------+-----------
t1 | 3 | 299
(1 row)
3. 显示简单查询的执行计划；建立索引并显示有索引条件的执行计划
omm=# explain select * from t1 ;
QUERY PLAN
------------------------------------------------------------------------
Partition Iterator (cost=0.00..5.99 rows=299 width=146)
Iterations: 3
-> Partitioned Seq Scan on t1 (cost=0.00..5.99 rows=299 width=146)
Selected Partitions: 1..3
(4 rows)
omm=# explain select * from t1 where id=7;
QUERY PLAN
----------------------------------------------------------------------
Partition Iterator (cost=0.00..6.74 rows=1 width=146)
Iterations: 3
-> Partitioned Seq Scan on t1 (cost=0.00..6.74 rows=1 width=146)
Filter: (id = 7)
omm=# Selected Partitions: 1..3
(5 rows)
omm=# explain select * from t1 where did=7;
QUERY PLAN
----------------------------------------------------------------------
Partition Iterator (cost=0.00..4.74 rows=1 width=146)
Iterations: 1
-> Partitioned Seq Scan on t1 (cost=0.00..4.74 rows=1 width=146)
Filter: (did = 7)
Selected Partitions: 1
(5 rows)
omm=# create index ind_t1_id on t1(id);
omm=# CREATE INDEX
omm=# explain select * from t1 where id=7;
QUERY PLAN
----------------------------------------------------------------------
Partition Iterator (cost=0.00..6.74 rows=1 width=146)
Iterations: 3
-> Partitioned Seq Scan on t1 (cost=0.00..6.74 rows=1 width=146)
Filter: (id = 7)
Selected Partitions: 1..3
(5 rows)
数据量太小了。。。不走索引。
4. 更新表数据，并做垃圾收集
omm=# update t1 set id=did;
UPDATE 299
omm=# vacuum(verbose, analyze) t1;
INFO: vacuuming "public.t1"(gaussdb pid=1)
INFO: scanned index "ind_t1_did" to remove 198 row versions(gaussdb pid=1)
DETAIL: CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: scanned index "ind_t1_id" to remove 198 row versions(gaussdb pid=1)
DETAIL: CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: index "ind_t1_did" now contains 198 row versions in 2 pages(gaussdb pid=1)
DETAIL: 99 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: index "ind_t1_id" now contains 198 row versions in 7 pages(gaussdb pid=1)
DETAIL: 99 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: "t1": found 198 removable, 198 nonremovable row versions in 3 out of 3 pages(gaussdb pid=1)
DETAIL: 99 dead row versions cannot be removed yet.
There were 0 unused item pointers.
0 pages are entirely empty.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: vacuuming "public.t1"(gaussdb pid=1)
INFO: scanned index "ind_t1_did" to remove 200 row versions(gaussdb pid=1)
DETAIL: CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: scanned index "ind_t1_id" to remove 200 row versions(gaussdb pid=1)
DETAIL: CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: index "ind_t1_did" now contains 200 row versions in 2 pages(gaussdb pid=1)
DETAIL: 100 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: index "ind_t1_id" now contains 200 row versions in 7 pages(gaussdb pid=1)
DETAIL: 100 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: "t1": found 200 removable, 200 nonremovable row versions in 3 out of 3 pages(gaussdb pid=1)
DETAIL: 100 dead row versions cannot be removed yet.
There were 0 unused item pointers.
0 pages are entirely empty.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: vacuuming "public.t1"(gaussdb pid=1)
INFO: scanned index "ind_t1_did" to remove 200 row versions(gaussdb pid=1)
DETAIL: CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: scanned index "ind_t1_id" to remove 200 row versions(gaussdb pid=1)
DETAIL: CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: index "ind_t1_did" now contains 200 row versions in 2 pages(gaussdb pid=1)
DETAIL: 100 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: index "ind_t1_id" now contains 200 row versions in 7 pages(gaussdb pid=1)
DETAIL: 100 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: "t1": found 200 removable, 200 nonremovable row versions in 3 out of 3 pages(gaussdb pid=1)
DETAIL: 100 dead row versions cannot be removed yet.
There were 0 unused item pointers.
0 pages are entirely empty.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: scanned index "ind_t1_id" to remove 0.000000 invisible rows(gaussdb pid=1)
DETAIL: CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: analyzing "public.t1"(gaussdb pid=1)
INFO: ANALYZE INFO : "t1": scanned 3 of 3 pages, containing 99 live rows and 99 dead rows; 99 rows in sample, 99 estimated total rows(gaussdb pid=1)
INFO: ANALYZE INFO : "t1": scanned 3 of 3 pages, containing 100 live rows and 100 dead rows; 100 rows in sample, 100 estimated total rows(gaussdb pid=1)
INFO: ANALYZE INFO : "t1": scanned 3 of 3 pages, containing 100 live rows and 100 dead rows; 100 rows in sample, 100 estimated total rows(gaussdb pid=1)
VACUUM
5. 清理数据
omm=# drop table t1;
DROP TABLE
