最近有朋友問如下問題：

       一個表中的資料經過增刪改後並執行vacuum table操作後，發現後續插入的記錄的ctid比老的資料還要小，這是為什麼。

原因：這是空間重複利用的結果，是正常的。由於資料的記錄是儲存在某個塊中的，當資料被刪除後該記錄的地方在塊中就會形成空洞，如果沒有任何事物再需要這個記錄的時候，經過vacuum操作後，就會標識這條記錄的空間可以被重複利用了，也就是說新插入的記錄是可以放到該位置了，這個地方的ctid比老記錄的ctid小就是正常的了。

     測試結果：

     1. 建立一張測試表，並向表中插入100w條測試記錄。

sde=# create table t4 (id int,info text);
CREATE TABLE
sde=# insert into t4 select generate_series(1,1000000),md5(random()::text);
INSERT 0 1000000
 

    2. 查詢記錄id從1000到1010這幾條記錄的ctid。

sde=# select ctid from t4 where id>10000 and id<10010;
  ctid
---------
 (83,41)
 (83,42)
 (83,43)
 (83,44)
 (83,45)
 (83,46)
 (83,47)
 (83,48)
 (83,49)
(9 行記錄)

   3. 刪除這幾條記錄，並vacuum。

sde=# delete  from t4 where id>10000 and id<10010;
DELETE 9
sde=# vacuum verbose t4;
INFO:  vacuuming "sde.t4"
INFO:  "t4": removed 9 row versions in 1 pages
INFO:  "t4": found 9 removable, 999991 nonremovable row versions in 8334 out of 8334 pages
描述:  0 dead row versions cannot be removed yet, oldest xmin: 873827
There were 0 unused item pointers.
Skipped 0 pages due to buffer pins, 0 frozen pages.
0 pages are entirely empty.
CPU: user: 0.25 s, system: 0.00 s, elapsed: 0.25 s.
INFO:  vacuuming "pg_toast.pg_toast_52633"
INFO:  index "pg_toast_52633_index" now contains 0 row versions in 1 pages
描述:  0 index row versions were removed.
0 index pages have been deleted, 0 are currently reusable.
CPU: user: 0.00 s, system: 0.00 s, elapsed: 0.00 s.
INFO:  "pg_toast_52633": found 0 removable, 0 nonremovable row versions in 0 out of 0 pages
描述:  0 dead row versions cannot be removed yet, oldest xmin: 873827
There were 0 unused item pointers.
Skipped 0 pages due to buffer pins, 0 frozen pages.
0 pages are entirely empty.
CPU: user: 0.00 s, system: 0.00 s, elapsed: 0.00 s.
VACUUM
 

  發現清除了9條記錄。

  5. 繼續插入幾條記錄

sde=# insert into t4 values (1000001,md5(random()::text));
INSERT 0 1
sde=# insert into t4 values (1000002,md5(random()::text));
INSERT 0 1
sde=# insert into t4 values (1000003,md5(random()::text));
INSERT 0 1
sde=# insert into t4 values (1000004,md5(random()::text));
INSERT 0 1
sde=# insert into t4 values (1000005,md5(random()::text));
INSERT 0 1
sde=# select ctid from t4 where id>=1000001 ;
  ctid
---------
 (83,41)
 (83,42)
 (83,43)
 (83,44)
 (83,45)
(5 行記錄)
 

  可以看到新插入的記錄放到了剛才刪除的那幾條記錄的位置上，繼續插入

sde=# insert into t4 values (1000006,md5(random()::text));
INSERT 0 1
sde=# insert into t4 values (1000007,md5(random()::text));
INSERT 0 1
sde=# insert into t4 values (1000008,md5(random()::text));
INSERT 0 1
sde=# insert into t4 values (1000009,md5(random()::text));
INSERT 0 1
sde=# insert into t4 values (10000010,md5(random()::text));
INSERT 0 1
sde=# select ctid from t4 where id>=1000001 ;
   ctid
-----------
 (83,41)
 (83,42)
 (83,43)
 (83,44)
 (83,45)
 (83,46)
 (83,47)
 (83,48)
 (83,49)
 (8333,41)
(10 行記錄)

   當把塊內的空洞沾滿後，新插入的記錄會放到最後一個塊上了。