今天做一個備份看板的時候,無意中寫了如下的SQL,當時看到之後就有點疑惑了。
mysql> select backup_date ,count(*) piece_no from redis_backup_result;
+-------------+----------+
| backup_date | piece_no |
+-------------+----------+
| 2018-08-14 | 40906 |
+-------------+----------+
1 row in set (0.03 sec)
一天之內肯定沒有這麼多的記錄,明顯不對,到底是哪裏出了問題呢。
自己仔細看了下SQL,發現是沒有加group by
我們隨機查出10條數據。
mysql> select backup_date from redis_backup_result limit 10;
+-------------+
| backup_date |
+-------------+
| 2018-08-14 |
| 2018-08-14 |
| 2018-08-14 |
| 2018-08-15 |
| 2018-08-15 |
| 2018-08-15 |
| 2018-08-15 |
| 2018-08-15 |
| 2018-08-15 |
| 2018-08-15 |
+-------------+
10 rows in set (0.00 sec)
這個梗很多同學都知道,是在早期的版本中sql_mode默認爲null,不會校驗這個部分,從語法角度來說,是允許的,但是到了高版本,比如5.7之後是不支持的。
mysql> show variables like 'sql_mode%';
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| sql_mode | |
+---------------+-------+
1 row in set (0.01 sec)
添加group by之後,結果就符合預期了。
mysql> select backup_date ,count(*) piece_no from redis_backup_result group by backup_date;
+-------------+----------+
| backup_date | piece_no |
+-------------+----------+
| 2018-08-14 | 3 |
| 2018-08-15 | 121 |
| 2018-08-16 | 184 |
| 2018-08-17 | 3284 |
| 2018-08-18 | 7272 |
| 2018-08-19 | 7272 |
| 2018-08-20 | 7272 |
| 2018-08-21 | 7272 |
| 2018-08-22 | 8226 |
+-------------+----------+
9 rows in set (0.06 sec)
但是問題到了這裏,我比較奇怪上面的邏輯,到底是怎麼解析的,看起來是SQL解析了第一行,然後輸出了count(*)的操作。
顯然這個是從執行計劃中無法得到的信息。
mysql> explain extended select backup_date ,count(*) piece_no from redis_backup_result;
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
| 1 | SIMPLE | redis_backup_result | NULL | ALL | NULL | NULL | NULL | NULL | 38351 | 100.00 | NULL |
+----+-------------+---------------------+------------+------+---------------+------+---------+------+-------+----------+-------+
1 row in set, 2 warnings (0.00 sec)
mysql> show warnings;
| Level | Code | Message
| Warning | 1681 | 'EXTENDED' is deprecated and will be removed in a future release. |
| Note | 1003 | /* select#1 */ select `devopsdb`.`redis_backup_result`.`backup_date` AS `backup_date`,count(0) AS `piece_no` from `devopsdb`.`redis_backup_result` |
2 rows in set (0.00 sec)
我們換個思路。添加sql_mode的約束。
mysql> set session sql_mode='ONLY_FULL_GROUP_BY,STRICT_TRANS_TABLES,NO_ZERO_IN_DATE,NO_ZERO_DATE,ERROR_FOR_DIVISION_BY_ZERO,NO_AUTO_CREATE_USER,NO_ENGINE_SUBSTITUTION';
Query OK, 0 rows affected, 1 warning (0.00 sec)
可以看到這個表有4萬多的記錄。
mysql> select count(*)from redis_backup_result;
+----------+
| count(*) |
+----------+
| 40944 |
+----------+
1 row in set (0.01 sec)
爲了驗證,我們可以使用_rowid的方式來做初步的驗證。
mysql> select _rowid from redis_backup_result limit 5;
+--------+
| _rowid |
+--------+
| 117 |
| 118 |
| 119 |
| 120 |
| 121 |
+--------+
5 rows in set (0.00 sec)
然後可以實現一個初步的思路。
mysql> select _rowid,count(*)from redis_backup_result;
+--------+----------+
| _rowid | count(*) |
+--------+----------+
| 117 | 41036 |
+--------+----------+
1 row in set (0.03 sec)
然後藉助rownum來實現。
mysql> SELECT @rowno:=@rowno+1 as rowno,r._rowid from redis_backup_result r ,(select @rowno:=0) t limit 20;
+-------+--------+
| rowno | _rowid |
+-------+--------+
| 1 | 117 |
| 2 | 118 |
| 3 | 119 |
| 4 | 120 |
| 5 | 121 |
| 6 | 122 |
| 7 | 123 |
| 8 | 124 |
| 9 | 125 |
| 10 | 126 |
| 11 | 127 |
| 12 | 128 |
| 13 | 129 |
| 14 | 130 |
| 15 | 131 |
| 16 | 132 |
| 17 | 133 |
| 18 | 134 |
| 19 | 135 |
| 20 | 136 |
+-------+--------+
20 rows in set (0.00 sec)
寫一個完整的語句。
mysql> SELECT @rowno:=@rowno+1 as rowno,r._rowid ,backup_date,count(*) from redis_backup_result r ,(select @rowno:=0) t ;
+-------+--------+-------------+----------+
| rowno | _rowid | backup_date | count(*) |
+-------+--------+-------------+----------+
| 1 | 117 | 2018-08-14 | 41061 |
+-------+--------+-------------+----------+
1 row in set (0.02 sec)
很明顯是第1行的記錄,然後做了count(*)的操作。