Postgresql Automatic Vacuuming

VACUUM是PG日常回收磁盘空间的一个手段，通过参数设置，触发做vacuum回收磁盘空间

相关参数：

#autovacuum_work_mem = -1 # 最小1MB,如果是-1，那么使用maintenance_work_mem设置的值

vacuum_cost_delay = 10 # 0-100 milliseconds

#vacuum_cost_page_hit = 1 # 0-10000 credits

#vacuum_cost_page_miss = 10 # 0-10000 credits

#vacuum_cost_page_dirty = 20 # 0-10000 credits

#vacuum_cost_limit = 200 # 1-10000 credits

#vacuum_defer_cleanup_age = 0 # 该参数只能在postgresql.conf参数文件中设置，表示延迟多少个事物后再vacuum

autovacuum = on # 自动vacuum的开关

log_autovacuum_min_duration = 0 # -1 关闭，0记录所有vacuum日志,其他数字表是大于等于这个数值的vacuum操作才记录到日志

autovacuum_max_workers = 3 # 设置最大autovacuum进程个数，默认3

autovacuum_naptime = 1min # autovacuum运行的时间间隔

autovacuum_vacuum_threshold = 50 # vacuum触发的阀值条件之一 (修改，删除，的记录数)

autovacuum_analyze_threshold = 50 # analyze触发的阀值条件之一 (修改，删除，插入的记录数)

#触发vacuum计算公式：vacuum threshold = vacuum base threshold + vacuum scale factor * number of tuples

autovacuum_vacuum_scale_factor = 0.2 # vacuum触发的阀值条件之一 (修改，删除，插入表的20%)

autovacuum_analyze_scale_factor = 0.1 # analyze触发的阀值条件之一 (修改，删除，插入表的10%)

#触发analyze计算公式：analyze threshold = analyze base threshold + analyze scale factor * number of tuples

autovacuum_freeze_max_age = 200000000

#即使autovacuum未开启，也会vacuum，表示一个表中的xid如果大于设置的阀值，那么可能导致数据丢失，不可见。如果达到这个阀值，那么就会触发vacuum，防止wraparound，可以通过 pg_class.relfrozenxid 查询表的事务年龄。

autovacuum_multixact_freeze_max_age = 400000000 # 对于multixid冻结的阀值，pg_class.relminmxid查询，如果达到该阀值，就进行vaccum

#autovacuum_vacuum_cost_delay = 20ms

#autovacuum执行时，对vacuum的cost进程评估，如果超过autovacuum_vacuum_cost_limit的值，那么就延迟该参数的值20ms，如果设置为-1，那么使用vacuum_cost_delay设置的值

#autovacuum_vacuum_cost_limit = -1

# 默认autovacuum评估的阀值，设置为-1，表是使用vacuum_cost_limit的值，超过就会休眠autovacuum_vacuum_cost_delay这么久的时间，然后再唤醒继续执行vacuum

#vacuum_freeze_min_age = 50000000

#vacuum_freeze_table_age = 150000000

#vacuum_multixact_freeze_min_age = 5000000

#vacuum_multixact_freeze_table_age = 150000000

触发analyze：

hank=> create table test1 (a int,b text);
CREATE TABLE
hank=> select reltuples from pg_class where relname='test1';
reltuples
-----------
0

hank=> show autovacuum_analyze_threshold;
-[ RECORD 1 ]----------------+---
autovacuum_analyze_threshold | 50

hank=> show autovacuum_analyze_scale_factor;
-[ RECORD 1 ]-------------------+----
autovacuum_analyze_scale_factor | 0.1
根据之前触发analyze的条件计算所得
hank=> select 0.1*0+50;
?column? | 50.0 #超过50就进行analyze

hank=> select * from pg_stat_all_tables where relname ='test1';
-[ RECORD 1 ]-------+------
relid | 21913
schemaname | hank
relname | test1
seq_scan | 0
seq_tup_read | 0
idx_scan |
idx_tup_fetch |
n_tup_ins | 50
n_tup_upd | 0
n_tup_del | 0
n_tup_hot_upd | 0
n_live_tup | 50
n_dead_tup | 0 #
n_mod_since_analyze | 50
last_vacuum |
last_autovacuum |
last_analyze |
last_autoanalyze | #最后一次自动ananlyze的时间
vacuum_count | 0
autovacuum_count | 0
analyze_count | 0
autoanalyze_count | 0

hank=>insert into test1 values (51,'hank');

hank=> select * from pg_stat_all_tables where relname ='test1';
-[ RECORD 1 ]-------+------------------------------
relid | 21913
schemaname | hank
relname | test1
seq_scan | 0
seq_tup_read | 0
idx_scan |
idx_tup_fetch |
n_tup_ins | 51
n_tup_upd | 0
n_tup_del | 0
n_tup_hot_upd | 0
n_live_tup | 51
n_dead_tup | 0
n_mod_since_analyze | 0
last_vacuum |
last_autovacuum |
last_analyze |
last_autoanalyze | 2016-04-08 13:59:40.333561+08
vacuum_count | 0
autovacuum_count | 0
analyze_count | 0
autoanalyze_count | 1

触发vacuum：
hank=> select reltuples from pg_class where relname='test1';
-[ RECORD 1 ]-
reltuples | 51

hank=> show autovacuum_vacuum_threshold;
-[ RECORD 1 ]---------------+---
autovacuum_vacuum_threshold | 50

hank=> show autovacuum_vacuum_scale_factor;
-[ RECORD 1 ]------------------+----
autovacuum_vacuum_scale_factor | 0.2

hank=> select 0.2*51+50;
-[ RECORD 1 ]--
?column? | 60.2 #需要61条触发vacuum

hank=> update test1 set b='oooo' where a<50;
UPDATE 49
hank=> delete from test1 where a<12;
DELETE 11

hank=> select * from pg_stat_all_tables where relname ='test1';
-[ RECORD 1 ]-------+------------------------------
relid | 21913
schemaname | hank
relname | test1
seq_scan | 2
seq_tup_read | 102
idx_scan |
idx_tup_fetch |
n_tup_ins | 51
n_tup_upd | 49
n_tup_del | 11
n_tup_hot_upd | 49
n_live_tup | 40
n_dead_tup | 60 #可以看到有60条数据是dead rows
n_mod_since_analyze | 60
last_vacuum |
last_autovacuum |
last_analyze |
last_autoanalyze | 2016-04-08 13:59:40.333561+08
vacuum_count | 0
autovacuum_count | 0
analyze_count | 0
autoanalyze_count | 1

hank=> delete from test1 where a<13;
DELETE 1

hank=> select * from pg_stat_all_tables where relname ='test1';
-[ RECORD 1 ]-------+------------------------------
relid | 21913
schemaname | hank
relname | test1
seq_scan | 3
seq_tup_read | 142
idx_scan |
idx_tup_fetch |
n_tup_ins | 51
n_tup_upd | 49
n_tup_del | 12
n_tup_hot_upd | 49
n_live_tup | 39
n_dead_tup | 0
n_mod_since_analyze | 0
last_vacuum |
last_autovacuum | 2016-04-08 14:10:51.113524+08 #最后autovacuum的时间
last_analyze |
last_autoanalyze | 2016-04-08 14:10:51.143719+08
vacuum_count | 0
autovacuum_count | 1
analyze_count | 0
autoanalyze_count | 2

修改表级autovacuum触发条件：

修改表级参数：
hank=> alter table test1 set (autovacuum_vacuum_threshold = 100,autovacuum_vacuum_scale_factor = 0.5);
ALTER TABLE
hank=> select reltuples from pg_class where relname='test1';
-[ RECORD 1 ]-
reltuples | 39

hank=> select 39*0.5+100; #计算触发autovacuum条数
-[ RECORD 1 ]---
?column? | 119.5

hank=> update test1 set b='abc' where a<51;
UPDATE 38
hank=> update test1 set b='aaa' where a<51;
UPDATE 38
hank=> update test1 set b='bbb' where a<51;
UPDATE 38
hank=> select 38*3; #已经update114条
-[ RECORD 1 ]-
?column? | 114
hank=> select min(a) from test1 ; #因为之前实验删除过表看下a的最小值然后根据a的值继续更新到我们需要的条数
-[ RECORD 1 ]
min | 13
hank=> update test1 set b='ccc' where a<18; #更新5条，目前是114+5=119
UPDATE 5
hank=> select * from pg_stat_all_tables where relname ='test1'; #观察last_autovacuum可见并没有触发autovacuum
-[ RECORD 1 ]-------+------------------------------
relid | 21913
schemaname | hank
relname | test1
seq_scan | 9
seq_tup_read | 376
idx_scan |
idx_tup_fetch |
n_tup_ins | 51
n_tup_upd | 168
n_tup_del | 12
n_tup_hot_upd | 168
n_live_tup | 39
n_dead_tup | 119
n_mod_since_analyze | 5
last_vacuum |
last_autovacuum | 2016-04-08 14:10:51.113524+08
last_analyze |
last_autoanalyze | 2016-04-08 14:32:51.394712+08
vacuum_count | 0
autovacuum_count | 1
analyze_count | 0
autoanalyze_count | 3

hank=> update test1 set b='ddd' where a<14; #更新一条，现在是120条，刚好触发autovacuum的条件
UPDATE 1

hank=> select * from pg_stat_all_tables where relname ='test1'; #可见last_autovacuum已经变化
-[ RECORD 1 ]-------+------------------------------
relid | 21913
schemaname | hank
relname | test1
seq_scan | 10
seq_tup_read | 415
idx_scan |
idx_tup_fetch |
n_tup_ins | 51
n_tup_upd | 169
n_tup_del | 12
n_tup_hot_upd | 169
n_live_tup | 39
n_dead_tup | 0
n_mod_since_analyze | 6
last_vacuum |
last_autovacuum | 2016-04-08 14:35:51.513553+08
last_analyze |
last_autoanalyze | 2016-04-08 14:32:51.394712+08
vacuum_count | 0
autovacuum_count | 2
analyze_count | 0
autoanalyze_count | 3

analyze同上的道理，就不再实验

pg_stat_all_tables视图见下表

Column	Type	Description
relid	oid	table的oid
schemaname	name	表所属的schema名称
relname	name	表名
seq_scan	bigint	Number of sequential scans initiated on this table
seq_tup_read	bigint	Number of live rows fetched by sequential scans
idx_scan	bigint	Number of index scans initiated on this table
idx_tup_fetch	bigint	Number of live rows fetched by index scans
n_tup_ins	bigint	Number of rows inserted
n_tup_upd	bigint	Number of rows updated (includes HOT updated rows)
n_tup_del	bigint	Number of rows deleted
n_tup_hot_upd	bigint	Number of rows HOT updated (i.e., with no separate index update required)
n_live_tup	bigint	live rows 的数量
n_dead_tup	bigint	dead rows的数量
n_mod_since_analyze	bigint	Estimated number of rows modified since this table was last analyzed
last_vacuum	timestamp with time zone	Last time at which this table was manually vacuumed (not counting VACUUM FULL)
last_autovacuum	timestamp with time zone	最后一次autovacuum的时间
last_analyze	timestamp with time zone	Last time at which this table was manually analyzed
last_autoanalyze	timestamp with time zone	最后一次autoanalyze的时间
vacuum_count	bigint	Number of times this table has been manually vacuumed (not counting VACUUM FULL)
autovacuum_count	bigint	Number of times this table has been vacuumed by the autovacuum daemon
analyze_count	bigint	Number of times this table has been manually analyzed
autoanalyze_count	bigint	Number of times this table has been analyzed by the autovacuum daemon

参考：http://www.postgresql.org/docs/current/static/runtime-config-autovacuum.html