Question

在我的数据库中，我创建了表：

| ID |FAILURE | FAMILY_ID | FAILURE_DATE | 1 | AAA | 1 | 2016-06-16 17:22:09.823 | 2 | BBB | 2 | 2016-06-16 17:17:09.853 | 3 | CCC | 3 | 2016-06-16 17:12:09.853 | 4 | DDD | 4 | 2016-06-16 17:07:09.853 | 5 | AAA | 2 | 2016-06-16 17:22:09.857 | 6 | BBB | 1 | 2016-06-16 17:17:09.857 | 7 | CCC | 2 | 2016-06-16 17:12:09.857 | 8 | DDD | 3 | 2016-06-16 17:07:09.857 | 9 | AAA | 4 | 2016-06-16 17:22:09.857 | 10 | BBB | 3 | 2016-06-16 17:17:09.857 | 11 | CCC | 2 | 2016-06-16 17:12:09.857 | 12 | DDD | 1 | 2016-06-16 17:07:09.857 | 13 | AAA | 2 | 2016-06-16 17:22:09.857 | 14 | BBB | 1 | 2016-06-16 17:17:09.857 | 15 | CCC | 4 | 2016-06-16 17:12:09.857 | 16 | DDD | 3 | 2016-06-16 17:07:09.860 | 17 | AAA | 4 | 2016-06-16 17:22:09.860 | 18 | BBB | 1 | 2016-06-16 17:17:09.860 | 19 | CCC | 2 | 2016-06-16 17:12:09.860 | 20 | DDD | 3 | 2016-06-16 17:07:09.860

使用此表格，我需要选择每个家庭在确定的时间段内出现的失败次数。

我在下面创建SQL代码：

SELECT
    distinct main.id,
    main.FAILURE,
    main.FAMILY,
    main.FAILURE_DATE
FROM
    dbo.tblTest Main
iNNER JOIN
(
    SELECT 
        tbl.FAILURE,
        tbl.FAMILY,
        COUNT( 1 ) as CNT
    FROM
        DBO.tblTest tbl
    WHERE
        tbl.FAILURE_DATE >= (SELECT DATEADD(MINUTE, -45, ( SELECT GETDATE())))
    GROUP BY
        tbl.FAILURE,
        tbl.FAMILY
    HAVING
        COUNT ( 1 ) > 1
) result on result.FAMILY = Main.FAMILY AND result.FAILURE = Main.FAILURE

结果是：

|ID | FAILURE | FAMILY | FAILURE_DATE |6 | BBB | 1 | 2016-06-16 17:17:09.857 |14 | BBB | 1 | 2016-06-16 17:17:09.857 |18 | BBB | 1 | 2016-06-16 17:17:09.860 |13 | AAA | 2 | 2016-06-16 17:22:09.857 |5 | AAA | 2 | 2016-06-16 17:22:09.857 |17 | AAA | 4 | 2016-06-16 17:22:09.860 |9 | AAA | 4 | 2016-06-16 17:22:09.857

那么，如何才能获得FAILURE和FAMILY的时间？

greate结果应为：

|ID | FAILURE | FAMILY | FAILURE_DATE |6 | BBB | 1 | 2016-06-16 17:17:09.857 |13 | AAA | 2 | 2016-06-16 17:22:09.857 |9 | AAA | 4 | 2016-06-16 17:22:09.857

如何更改我的SQL查询以获得这样的结果？

Answer 1

使用DATEADD(MI,-45,GETDATE())时很难重新创建精确的结果，但我认为17:17应该根据您当前的时间，数据和发布时间而有效 - 您可以根据需要更换

这是一个通过CTE使用窗口函数的查询。

DECLARE @TEST TABLE (ID INT, 
                     FAILURE VARCHAR(5),  
                     FAMILY_ID INT,
                     FAILURE_DATE DATETIME)

INSERT INTO @TEST (ID,FAILURE,FAMILY_ID,FAILURE_DATE)
VALUES (1,'AAA',1,'2016-06-16 17:22:09.823'),
       (2,'BBB',2,'2016-06-16 17:17:09.853'),
       (3,'CCC',3,'2016-06-16 17:12:09.853'),
       (4,'DDD',4,'2016-06-16 17:07:09.853'),
       (5,'AAA',2,'2016-06-16 17:22:09.857'),
       (6,'BBB',1,'2016-06-16 17:17:09.857'),
       (7,'CCC',2,'2016-06-16 17:12:09.857'),
       (8,'DDD',3 ,'2016-06-16 17:07:09.857'),
       (9,'AAA',4,'2016-06-16 17:22:09.857'),
       (10,'BBB',3,'2016-06-16 17:17:09.857'),
       (11,'CCC',2,'2016-06-16 17:12:09.857'),
       (12,'DDD',1,'2016-06-16 17:07:09.857'),
       (13,'AAA',2,'2016-06-16 17:22:09.857'),
       (14,'BBB',1,'2016-06-16 17:17:09.857'),
       (15,'CCC',4,'2016-06-16 17:12:09.857'),
       (16,'DDD',3,'2016-06-16 17:07:09.860'),
       (17,'AAA',4,'2016-06-16 17:22:09.860'),
       (18,'BBB',1,'2016-06-16 17:17:09.860'),
       (19,'CCC',2,'2016-06-16 17:12:09.860'),
       (20,'DDD',3,'2016-06-16 17:07:09.860')

;WITH CTE AS
(
    SELECT *
           ,COUNT(2) OVER(PARTITION BY FAILURE,FAMILY_ID) AS recordCount
           ,ROW_NUMBER() OVER (PARTITION BY FAILURE,FAMILY_ID ORDER BY FAILURE_DATE,ID) AS rID
    FROM @TEST
    WHERE FAILURE_DATE >= '2016-06-16 17:17' --DATEADD(MI,-45,GETDATE())
)

SELECT ID, FAILURE, FAMILY_ID, FAILURE_DATE
FROM CTE
WHERE recordCount>1
AND rID=1

Answer 2

由于我现在没有Microsoft SQL Server 2012，请尝试以下代码：

SELECT X.id, X.FAILURE, X.FAMILY, X.FAILURE_DATE 
FROM (
    SELECT
        main.id,
        main.FAILURE,
        main.FAMILY,
        main.FAILURE_DATE,
        ROW_NUMBER()  OVER(PARTITION BY main.FAILURE,main.FAMILY ORDER BY main.FAILURE_DATE) AS RM 
    FROM
        dbo.tblTest Main 
) X    
WHERE X.RM=1

这是partition by的链接，可以帮助您

在表中选择重复的行

2 个答案: