openGauss每日一练第19天 | 学习笔记

原创手机用户2634 2021-12-30

572

第十九课 openGauss收集统计信息、打印执行计划、垃圾收集和checkpoint

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

学习内容如下

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

2.收集表统计信息

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

4.更新表数据，并做垃圾收集

5.清理数据


create schema tpcds;
CREATE TABLE tpcds.ttt1
(
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.ttt1 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');

insert into tpcds.ttt1 values(generate_series(10, 10000));


select relname, relpages, reltuples from pg_class where relname = 'ttt1';
analyze VERBOSE tpcds.ttt1;
select relname, relpages, reltuples from pg_class where relname = 'ttt1';

EXPLAIN SELECT * FROM tpcds.ttt1;
SET explain_perf_mode=normal;
EXPLAIN SELECT * FROM tpcds.ttt1;

create index idx_ttt1 on tpcds.ttt1(ca_address_sk);
EXPLAIN SELECT * FROM tpcds.ttt1 WHERE ca_address_sk<100;

update tpcds.ttt1 set ca_address_sk = ca_address_sk + 1 where ca_address_sk <100;
VACUUM (VERBOSE, ANALYZE) tpcds.ttt1;

drop schema tpcds cascade;

复制

课后作业记录

omm=# 
omm=# Create schema tpcds;
CREATE SCHEMA
omm=# CREATE TABLE tpcds.customer_address
omm-# (
omm(# ca_address_sk integer NOT NULL ,
omm(# ca_address_id character(16),
omm(# ca_street_number character(10) ,
omm(# ca_street_name character varying(60) ,
omm(# ca_street_type character(15) ,
omm(# ca_suite_number character(10) ,
omm(# ca_city character varying(60) ,
omm(# omm(# ca_county character varying(30) ,
ca_state character(2) ,
omm(# omm(# ca_zip character(10) ,
ca_country character varying(20) ,
omm(# ca_gmt_offset numeric(5,2) ,
omm(# ca_location_type character(20)
omm(# );
CREATE TABLE
omm=# insert into tpcds.customer_address values
omm-# (1, 'AAAAAAAABAAAAAAA', '18', 'Jackson', 'Parkway', 'Suite 280', 'Fairfield', 'Maricopa County', 'AZ', '86192' ,'United States', -7.00, 'condo'),
omm-# (2, 'AAAAAAAACAAAAAAA', '362', 'Washington 6th', 'RD', 'Suite 80', 'Fairview', 'Taos County', 'NM', '85709', 'United States', -7.00, 'condo'),
omm-# (3, 'AAAAAAAADAAAAAAA', '585', 'Dogwood Washington', 'Circle', 'Suite Q', 'Pleasant Valley', 'York County', 'PA', '12477', 'United States', -5.00, 'single family');
INSERT 0 3
omm=# 
omm=# insert into tpcds.customer_address values(generate_series(10, 10000));
INSERT 0 9991
omm=# 
omm=# 
omm=# select relname, relpages, reltuples from pg_class where relname = 'customer_address';
     relname      | relpages | reltuples 
------------------+----------+-----------
 customer_address |        0 |         0
(1 row)

omm=# omm=# 
analyze VERBOSE tpcds.customer_address;
INFO:  analyzing "tpcds.customer_address"(gaussdb pid=1)
INFO:  ANALYZE INFO : "customer_address": scanned 55 of 55 pages, containing 9994 live rows and 0 dead rows; 9994 rows in sample, 9994 estimated total rows(gaussdb pid=1)
omm=# ANALYZE

omm=# 
omm=# select relname, relpages, reltuples from pg_class where relname = 'customer_address';
     relname      | relpages | reltuples 
------------------+----------+-----------
 customer_address |       55 |      9994
(1 row)

omm=# 
omm=# SET explain_perf_mode=normal;
SET
omm=# 
omm=# EXPLAIN SELECT * FROM tpcds.customer_address;
                              QUERY PLAN                               
-----------------------------------------------------------------------
 Seq Scan on customer_address  (cost=0.00..154.94 rows=9994 width=151)
(1 row)

omm=# 
omm=# EXPLAIN(FORMAT JSON) SELECT * FROM tpcds.customer_address;
                 QUERY PLAN                 
--------------------------------------------
 [                                         +
   {                                       +
     "Plan": {                             +
       "Node Type": "Seq Scan",            +
       "Relation Name": "customer_address",+
       "Alias": "customer_address",        +
       "Startup Cost": 0.00,               +
       "Total Cost": 154.94,               +
       "Plan Rows": 9994,                  +
       "Plan Width": 151                   +
     }                                     +
   }                                       +
 ]
(1 row)

omm=# 
omm=# EXPLAIN(COSTS FALSE)SELECT * FROM tpcds.customer_address;
omm=#           QUERY PLAN          
------------------------------
 Seq Scan on customer_address
(1 row)


omm=# EXPLAIN SELECT SUM(ca_address_sk) FROM tpcds.customer_address WHERE ca_address_sk<100;
                               QUERY PLAN                                
-------------------------------------------------------------------------
 Aggregate  (cost=180.16..180.17 rows=1 width=12)
   ->  Seq Scan on customer_address  (cost=0.00..179.93 rows=94 width=4)
         Filter: (ca_address_sk < 100)
(3 rows)

omm=# 
omm=# create index customer_address_idx on tpcds.customer_address(ca_address_sk);
CREATE INDEX
omm=# EXPLAIN SELECT * FROM tpcds.customer_address WHERE ca_address_sk<100;
                                           QUERY PLAN                                           
------------------------------------------------------------------------------------------------
 [Bypass]
 Index Scan using customer_address_idx on customer_address  (cost=0.00..9.90 rows=94 width=151)
   Index Cond: (ca_address_sk < 100)
(3 rows)

omm=# omm=# 

omm=# update tpcds.customer_address set ca_address_sk = ca_address_sk + 1 where ca_address_sk <100;
UPDATE 93
omm=# VACUUM (VERBOSE, ANALYZE) tpcds.customer_address;
INFO:  vacuuming "tpcds.customer_address"(gaussdb pid=1)
INFO:  index "customer_address_idx" now contains 10087 row versions in 31 pages(gaussdb pid=1)
DETAIL:  0 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:  "customer_address": found 0 removable, 10087 nonremovable row versions in 55 out of 55 pages(gaussdb pid=1)
DETAIL:  93 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:  analyzing "tpcds.customer_address"(gaussdb pid=1)
INFO:  ANALYZE INFO : "customer_address": scanned 55 of 55 pages, containing 9994 live rows and 93 dead rows; 9994 rows in sample, 9994 estimated total rows(gaussdb pid=1)
VACUUM
omm=# 
omm=# 
omm=# 
omm=# CHECKPOINT;
CHECKPOINT
omm=# 
omm=# drop schema tpcds cascade;
NOTICE:  drop cascades to table tpcds.customer_address
DROP SCHEMA
omm=# 
omm=# 





omm=# 
omm=# create schema tpcds;
CREATE SCHEMA
omm=# CREATE TABLE tpcds.ttt1
omm-# (
omm(# ca_address_sk integer NOT NULL ,
omm(# ca_address_id character(16),
omm(# ca_street_number character(10) ,
omm(# ca_street_name character varying(60) ,
omm(# ca_street_type character(15) ,
omm(# ca_suite_number character(10) ,
omm(# ca_city character varying(60) ,
omm(# ca_county character varying(30) ,
omm(# ca_state character(2) ,
omm(# ca_zip character(10) ,
omm(# ca_country character varying(20) ,
omm(# ca_gmt_offset numeric(5,2) ,
omm(# ca_location_type character(20)
omm(# );
CREATE TABLE
omm=# insert into tpcds.ttt1 values
omm-# (1, 'AAAAAAAABAAAAAAA', '18', 'Jackson', 'Parkway', 'Suite 280', 'Fairfield', 'Maricopa County', 'AZ', '86192' ,'United States', -7.00, 'condo'),
omm-# (2, 'AAAAAAAACAAAAAAA', '362', 'Washington 6th', 'RD', 'Suite 80', 'Fairview', 'Taos County', 'NM', '85709', 'United States', -7.00, 'condo'),
omm-# (3, 'AAAAAAAADAAAAAAA', '585', 'Dogwood Washington', 'Circle', 'Suite Q', 'Pleasant Valley', 'York County', 'PA', '12477', 'United States', -5.00, 'single family');
INSERT 0 3
omm=# 
omm=# insert into tpcds.ttt1 values(generate_series(10, 10000));
INSERT 0 9991
omm=# 
omm=# 
omm=# 
omm=# omm=# 
select relname, relpages, reltuples from pg_class where relname = 'ttt1';


omm=#  relname | relpages | reltuples 
---------+----------+-----------
 ttt1    |        0 |         0
(1 row)analyze VERBOSE tpcds.ttt1;
INFO:  analyzing "tpcds.ttt1"(gaussdb pid=1)
INFO:  ANALYZE INFO : "ttt1": scanned 55 of 55 pages, containing 9994 live rows and 0 dead rows; 9994 rows in sample, 9994 estimated total rows(gaussdb pid=1)
ANALYZE
omm=# select relname, relpages, reltuples from pg_class where relname = 'ttt1';
 relname | relpages | reltuples 
---------+----------+-----------
 ttt1    |       55 |      9994
(1 row)

omm=# 
omm=# omm=# 
omm=# 
EXPLAIN SELECT * FROM tpcds.ttt1;
                        QUERY PLAN                         
-----------------------------------------------------------
 Seq Scan on ttt1  (cost=0.00..154.94 rows=9994 width=151)
(1 row)

omm=# 
omm=# SET explain_perf_mode=normal;
SET
omm=# EXPLAIN SELECT * FROM tpcds.ttt1;
                        QUERY PLAN                         
-----------------------------------------------------------
 Seq Scan on ttt1  (cost=0.00..154.94 rows=9994 width=151)
(1 row)

omm=# 
omm=# create index idx_ttt1 on tpcds.ttt1(ca_address_sk);
CREATE INDEX
omm=# EXPLAIN SELECT * FROM tpcds.ttt1 WHERE ca_address_sk<100;
omm=#                                QUERY PLAN                               
------------------------------------------------------------------------
 [Bypass]
 Index Scan using idx_ttt1 on ttt1  (cost=0.00..9.90 rows=94 width=151)
   Index Cond: (ca_address_sk < 100)
(3 rows)


omm=# 
omm=# 
omm=# update tpcds.ttt1 set ca_address_sk = ca_address_sk + 1 where ca_address_sk <100;
UPDATE 93
omm=# VACUUM (VERBOSE, ANALYZE) tpcds.ttt1;
INFO:  vacuuming "tpcds.ttt1"(gaussdb pid=1)
INFO:  index "idx_ttt1" now contains 10087 row versions in 31 pages(gaussdb pid=1)
DETAIL:  0 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:  "ttt1": found 0 removable, 10087 nonremovable row versions in 55 out of 55 pages(gaussdb pid=1)
DETAIL:  93 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:  analyzing "tpcds.ttt1"(gaussdb pid=1)
INFO:  ANALYZE INFO : "ttt1": scanned 55 of 55 pages, containing 9994 live rows and 93 dead rows; 9994 rows in sample, 9994 estimated total rows(gaussdb pid=1)
VACUUM
omm=# 
omm=# 
omm=# drop schema tpcds cascade;
NOTICE:  drop cascades to table tpcds.ttt1
DROP SCHEMA
omm=# 



复制

opengauss

「喜欢这篇文章，您的关注和赞赏是给作者最好的鼓励」

关注作者