SELECT DISTINCT

Question

我有一个MS SQL查询

SELECT DISTINCT Date_of_Record, Computer_Name, IP_Address, Agent_Version, Mac_Address 
FROM [SEP_Versions].dbo.Computers 
WHERE Computer_Name in (SELECT Computer_Name 
         FROM [SEP_Versions].dbo.Computers 
         GROUP BY Computer_Name 
         HAVING COUNT(DISTINCT Agent_Version) > 1) 
AND [SEP_Versions].dbo.Computers.COMPUTER_NAME LIKE '%s001' 
AND [SEP_Versions].dbo.Computers.IP_Address LIKE '%.100' 
ORDER BY Computer_Name, Date_of_Record 

输出

Date  Computer IP    Version   Mac 
2014-10-24 COMP001 10.10.10.10  12.1.3001.165 NULL 
2014-10-25 COMP001 10.10.10.10  12.1.3001.165 NULL 
2014-10-26 COMP001 10.10.10.10  12.1.3001.165 NULL 
2014-10-27 COMP001 10.10.10.10  12.1.4013.4013 NULL 
2014-10-28 COMP001 10.10.10.10  12.1.4013.4013 NULL 
2014-10-29 COMP001 10.10.10.10  12.1.5773.2276 NULL 

如何改变SELECT DISTINCT语句，因此它最早输出具有鲜明版本记录日期，即

Date  Computer IP    Version   Mac 
2014-10-24 COMP001 10.10.10.10  12.1.3001.165 NULL 
2014-10-27 COMP001 10.10.10.10  12.1.4013.4013 NULL 
2014-10-29 COMP001 10.10.10.10  12.1.5773.2276 NULL 

@Pradeep

您的解决方案输出

Date  Computer IP    Version   Mac 
2014-10-17 COMP001 10.10.10.10  12.1.3001.165 NULL 
2014-10-17 COMP002 10.10.10.11  12.1.4013.4013 NULL 
2014-10-17 COMP003 10.10.10.12  12.1.5773.2276 NULL 
2014-10-17 COMP004 10.10.10.14  11.0.5003.2276 NULL 

@ 1010和@AWinkle

如果我运行

SELECT DISTINCT Computer_Name, IP_Address, Agent_Version 
FROM dbo.Computers 
WHERE Computer_Name in (SELECT Computer_Name 
         FROM dbo.Computers 
         GROUP BY Computer_Name 
         HAVING COUNT(DISTINCT Agent_Version) > 1) 
AND dbo.Computers.COMPUTER_NAME LIKE '%s001' 
AND dbo.Computers.IP_Address LIKE '%.100' 
ORDER BY Computer_Name 

然后我得到

Computer IP    Version   Mac 
COMP001 10.10.10.10  12.1.3001.165 NULL 
COMP001 10.10.10.10  12.1.4013.4013 NULL 
COMP001 10.10.10.10  12.1.5773.2276 NULL 

如何修改SQL语句，所以它增加了相应Date_of_Record列

Answer 1

简单的答案是通过使用窗口函数。

;WITH cte 
    AS (SELECT DISTINCT Row_number() 
          OVER (
          partition BY Computer_Name,Agent_Version 
          ORDER BY Date_of_Record) RN, 
         Date_of_Record, 
         Computer_Name, 
         IP_Address, 
         Agent_Version, 
         Mac_Address 
     FROM [SEP_Versions].dbo.Computers 
     WHERE Computer_Name IN (SELECT Computer_Name 
            FROM [SEP_Versions].dbo.Computers 
            GROUP BY Computer_Name 
            HAVING Count(DISTINCT Agent_Version) > 1) 
       AND [SEP_Versions].dbo.Computers.COMPUTER_NAME LIKE '%s001' 
       AND [SEP_Versions].dbo.Computers.IP_Address LIKE '%.100') 
SELECT Date_of_Record, 
        Computer_Name, 
        IP_Address, 
        Agent_Version, 
        Mac_Address 
FROM cte 
WHERE RN = 1 
ORDER BY Computer_Name, 
      Date_of_Record 

Answer 2

我觉得这是你想要

SELECT MIN(Date_of_Record), Computer_Name, IP_Address, Agent_Version, Mac_Address 
FROM ... 
WHERE ... 
GROUP BY Computer_Name, IP_Address, Agent_Version, Mac_Address 
ORDER BY ... 

group by子句将汇总行的计算机名，IP_ADDRESS AGENT_VERSION，MAC_ADDRESS的每个组合的东西。 MIN（Date_of_Record）将为每个组计算最早的日期。

Answer 3

在SQL Server中，您不能仅通过选定数量的列来选择不同列，而需要对它们进行分组。使用MIN函数将在最早的日期提供不同的版本（通过计算机/ ip/mac）。

SELECT MIN(Date_of_Record) Earliest_Date_of_Record, Computer_Name, IP_Address, Agent_Version, Mac_Address 
FROM [SEP_Versions].dbo.Computers 
WHERE Computer_Name in (SELECT Computer_Name 
        FROM [SEP_Versions].dbo.Computers 
        GROUP BY Computer_Name 
        HAVING COUNT(DISTINCT Agent_Version) > 1) 
AND [SEP_Versions].dbo.Computers.COMPUTER_NAME LIKE '%s001' 
AND [SEP_Versions].dbo.Computers.IP_Address LIKE '%.100' 
GROUP BY Computer_Name, IP_Address, Agent_Version, Mac_Address 
ORDER BY Computer_Name, Date_of_Record