Sql 无法完成查询（运行时间太长）

我试图得到一些测试结果的总体通过率

序列号代表一种仪器。一台仪器要通过107项不同的测试。试验分为两部分：

Linux测试（21）

其他（共86个，如电压测试、灯光测试）

在两种情况下，仪器将被视为通行证

标准方式：它在一次运行中通过107个不同的测试。代码如下：

选择序列号 从桌子上 按序号、条目分组 有计数（不同的测试标准）>=107

另一种方法：它在一次运行中通过107个不同的测试，或者 它在一轮中通过了86项其他测试，在一轮中通过了21项Linux测试 晚些时候

表中的“overalltestrunstatus”列仅表示仪器通过了列表中的所有测试，但列表本身可能不完整。（假设一个仪器获得“overalltestrunstatus”，但它通过的测试<107，它仍然是失败的）。有时，如果仪器失败或通过测试的次数少于某个数字，则仪器将重新运行。所以“entry”代表该乐器所代表的回合

我的目标是，找到那些通过测试但不符合标准的仪器

我的想法是找到所有那些“总体测试运行状态”等于通过但通过测试少于107的仪器。然后遍历结果和仪器的每个条目。如果它通过了86个其他测试，我会给它一个检查，如果通过了其他21个Linux，我会给它另一个检查。稍后我将计算所有有两张支票的仪器

我想要的是：

+----------------+------------+---------------+------------+
|  serial number |   entries  | others tests  | Linux tests|
+----------------+------------+---------------+------------+
|  170119904     |     1      |       yes     |    no      |
+----------------+------------+---------------+------------+
|  170119904     |     2      |       no      |    yes     |
+----------------+------------+---------------+------------+
|  180117000     |     1      |       no      |    no      |
+----------------+------------+---------------+------------+
|  180117000     |     2      |       no      |    no      |
+----------------+------------+---------------+------------+

因此，通过检查特定序列号是否通过了其他测试和Linux测试。我会知道它通过与否

我的问题是，每次查询运行大约20个小时，然后给出一个错误，如“断开连接”或“出现一些内部错误”。我永远也无法真正地度过难关。有什么想法吗？我真的很感谢你的帮助

这是表格：

+----------------------+--------------+----------+----------------+------------+---------+
| overalltestrunstatus | serialnumber | passfail |  testcriteria  |  testname  | Entries |
+----------------------+--------------+----------+----------------+------------+---------+
| PASS                 |    170119904 | Pass     | Backlight test | Functional |       1 |
+----------------------+--------------+----------+----------------+------------+---------++----------------------+--------------+----------+----------------+------------+---------+
| PASS                 |    170119904 | Pass     | Linux set test | Functional |       1 |
+----------------------+--------------+----------+----------------+------------+---------++----------------------+--------------+----------+----------------+------------+---------+
| PASS                 |    170119904 | Pass     | Factor    test | Functional |       1 |
+----------------------+--------------+----------+----------------+------------+---------++----------------------+--------------+----------+----------------+------------+---------+
| PASS                 |    170119904 | Pass     | Voltage   test | Functional |       1 |
+----------------------+--------------+----------+----------------+------------+---------++----------------------+--------------+----------+----------------+------------+---------+
| PASS                 |    170119904 | Pass     | Digital   test | Functional |       1 |
+----------------------+--------------+----------+----------------+------------+---------++----------------------+--------------+----------+----------------+------------+---------+
| PASS                 |    170119904 | Pass     | Detactor  test | Functional |       1 |
+----------------------+--------------+----------+----------------+------------+---------+

这是我的代码：

/* -serialnumber |-- entries--|--Pass1 --|--Pass2---|
   --------------|------------|----------|----------|
*/
DECLARE @checklist TABLE
(
  serialnumber varchar(100), 
  entries int,
  passother int default  0,
  passlinux int default  0
)

DECLARE @temp TABLE
(
  serialnumber int, 
  entries int,
  passother int default  0,
  passlinux int default  0
)

-- select serial number into the table

-- PCBA 214 ge to be counted
INSERT INTO @checklist (serialnumber, entries)
    SELECT serialnumber, entries 
    FROM dbo.boardtestresults1_full_view  t2
    WHERE partnumber = 'PCBA-10214-0001' 
        AND t2.serialnumber IN 
        (SELECT DISTINCT serialnumber
        FROM dbo.boardtestresults1_full_view  
        WHERE partnumber = 'PCBA-10214-0001' AND overalltestrunstatus = 'Pass' AND serialnumber NOT IN (
        SELECT DISTINCT serialnumber
        FROM dbo.boardtestresults1_full_view  
        WHERE partnumber = 'PCBA-10214-0001' AND overalltestrunstatus = 'Pass'
        GROUP BY serialnumber, Entries 
        HAVING COUNT(DISTINCT combined) >= 107))



-- check all for PCBA 10214

DECLARE @MyCursor1 CURSOR;
DECLARE @snumber VARCHAR(100); -- serailnumber
DECLARE @entries VARCHAR(100); -- entries
DECLARE @others int;
DECLARE @linux int;
DECLARE @pass1 int;
DECLARE @pass2 int;

BEGIN
    -- get serial numbers for all PCBA 214 from checklist
    SET @MyCursor1 = CURSOR FOR
    SELECT serialnumber, Entries
    FROM @checklist
    WHERE serialnumber like 'PCBA-10214-0001%' 

    -- get the roll
    OPEN @MyCursor1 
    FETCH NEXT FROM @MyCursor1 
    INTO @snumber, @entries

    -- loop through
    WHILE @@FETCH_STATUS = 0
    BEGIN
      /*
         YOUR ALGORITHM GOES HERE   
      */
    -- if pass the most or if pass linux
    IF ((SELECT COUNT(DISTINCT combined)FROM dbo.boardtestresults1_full_view WHERE  serialnumber = @snumber AND entries = @entries) >= 86)
        UPDATE @checklist SET passother = 1 WHERE serialnumber = @snumber AND entries = @entries
    ELSE IF ((SELECT COUNT(DISTINCT combined)FROM dbo.boardtestresults1_full_view WHERE  serialnumber = @snumber AND entries = @entries AND testname = 'LINUX Test' AND overalltestrunstatus = 'Pass') >= 21) 
        UPDATE @checklist SET passlinux = 1 WHERE serialnumber = @snumber AND entries = @entries
    END; 

    CLOSE @MyCursor1 ; 
    DEALLOCATE @MyCursor1;
END;

---

下面是一个工作示例，

PassFail

列未被计算，但在那里它对查询没有限制。为了简化示例，我添加了一个名为

TestType

的专栏，在您的用例中不太清楚Linux测试是如何从示例集派生出来的，可能是TestCriteria，但这使它不那么模棱两可。您应该能够获得测试的二进制文件，如果您需要进一步的帮助，您需要定义所有数据，我们可以提供帮助

declare @example as table (
    ExampleID     int identity(1,1) not null primary key clustered
,   OverAllStatus nvarchar(255) not null
,   SN            int not null
,   PassFail      nvarchar(255) not null
,   TestType      bit not null
,   TestCriteria  nvarchar(255) not null
,   Entry_        int not null
);

insert into @example (OverAllStatus, SN, PassFail, TestType, TestCriteria, Entry_)
select 'PASS', 170119904, 'Pass', 0, 'Backlight test', 1 union all
select 'PASS', 170119904, 'Pass', 0, 'Factor test', 1 union all
select 'PASS', 170119904, 'Pass', 0, 'Voltage test', 1 union all
select 'PASS', 170119904, 'Pass', 0, 'Detactor test', 1 union all

select 'PASS', 170119904, 'Pass', 0, 'Backlight test', 2 union all
select 'PASS', 170119904, 'Pass', 1, 'Linux set test', 2 union all
select 'PASS', 170119904, 'Pass', 0, 'Factor test', 2 union all
select 'PASS', 170119904, 'Pass', 0, 'Voltage test', 2 union all
select 'PASS', 170119904, 'Fail', 0, 'Digital test', 2 union all
select 'PASS', 170119904, 'Pass', 0, 'Detactor test', 2 union all

select 'PASS', 180117000, 'Pass', 0, 'Backlight test', 1 union all
select 'PASS', 180117000, 'Pass', 1, 'Linux set test', 1 union all
select 'PASS', 180117000, 'Pass', 0, 'Factor test', 1 union all
select 'PASS', 180117000, 'Pass', 0, 'Digital test', 2 union all
select 'PASS', 180117000, 'Pass', 0, 'Detactor test', 1;


  --| Control Group, only those that pass but does not have 107 records (OverAllStatus)
  ;with cte as (
     select SN
          , Entry_
          , count(*) cnt
       from @example
      where OverAllStatus = 'Pass'
   group by SN, Entry_
   --| Control count - this is your 107 or the numbers of tests the SN has passed
     having count(*) < 7
            )
   --| We are setting the count of records to the Test Type or the grouping
    , cte2 as (
    select b.SN
         , a.Entry_
         , iif(TestType = 1, 'Linux', 'Other') Test
      from cte a
      join @example b
        on a.SN = b.SN
       and a.Entry_ = b.Entry_
            )
    --| Roll up the counts
    , cte3 as ( 
    select distinct a.SN
         , a.Entry_
         , count(*) over(partition by a.SN, a.Test, a.Entry_) cnt
         , a.Test
      from cte2 a
            )

  --| Pivot the result set
    select SN
         , Entry_
         , [Other]
         , [Linux]
      from (
                select SN
                     , Entry_
                     , cnt
                     , test
                  from cte3
                  ) as x
            pivot 
                (
                 max(cnt)
                 for Test in ([Other], [Linux])
                ) as p;

---

你能编辑你的问题并提供一些（简化的）样本数据和期望的结果吗？你也能在没有所有差异的情况下尝试吗？

光标

可能对问题没有帮助。看起来可以用数据集方法编写。while循环中没有fetch next。看起来像是一个无限循环问题。循环所能做的就是减慢一个本来很快的过程。在你的情况下，无限期。这应该是一个基于集合的update语句，带有大小写表达式。非常感谢。我是SQL新手，这可能需要一段时间才能理解和应用。无论如何，我真的很感谢你的帮助。np，一旦你评估了它，如果它对你有帮助，请选择它作为答案。如果您对代码或如何进行更改有任何疑问，请发布，如果我没有回复，我相信会有人回复；我会努力跟进的，我很乐意，但我不能，因为我的声誉低于15。

SN          Entry_  Other   Linux
170119904   1       4       NULL
180117000   1       3       1
170119904   2       5       1
180117000   2       1       NULL