表数据计数最大值

Question

我在MySQL中有两个表

+---------+-----------+
| machine | status    |
+---------+-----------+
| 40001   | Completed | 
| 40001   | Completed | 
| 40001   | Completed | 
| 40001   | Completed | 
| 40001   | Pending   | 
| 40001   | Pending   | 
| 40001   | Pending   | 
| 40001   | Pending   | 
| 40001   | Pending   | 
| 40001   | Pending   | 
+---------+-----------+
And the other one as
+---------+---------+
| machine | packets | 
+---------+---------+
| 40001   |   527   |
| 40001   |   1497  |
| 40002   |   1414  | 
| 40002   |   2796  |
| 40003   |  392    | 
| 40003   |  1663   | 
| 40004   |   500   | 
| 40004   |   1277  |
+-------+----------+

我想编写一个选择查询，该查询为我提供了该机器的机器，完成计数，未决计数和最大数据包数。 所以我尝试了

SELECT machine,max(packets) AS sync,
      sum(if(laststatus='completed', 1, 0)) AS generation,
      sum(if(laststatus != 'completed', 1, 0)) AS pending
FROM machine_status
right join machine_packets on machine_packets.machine=machine_status.machine
GROUP BY machine

但是我得到了：

+---------+------+------------+---------+
| machine | sync | generation | pending |
+---------+------+------------+---------+
| 40001   | 1497 |          8 |       2 | 
| 40002   | 2796 |          4 |       2 | 
| 40003   | 1663 |          6 |       0 | 
| 40004   | 1277 |          0 |       2 | 
| 40005   | 2755 |          0 |       0 | 
| 40006   |  927 |          0 |       0 | 
| 40007   |  306 |          0 |       0 | 
+---------+------+------------+---------+

如我们所见，值在“生成和待处理”列中加倍。 我哪里做错了 ？

Answer 1

SELECT machine,sync,
      sum(if(laststatus='completed', 1, 0)) AS generation,
      sum(if(laststatus != 'completed', 1, 0)) AS pending
FROM machine_status
right join (select machine, 
                   max(packets) AS sync 
            from machine_packets 
            group by machine) mp on mp.machine=machine_status.machine
GROUP BY machine

它们加倍，因为machine_packets每个ID具有2条记录。 为了避免这种情况，您可以在子查询中移动它

Answer 2

一种安全的方法是使用union all ：

select machine, sum(status = 'completed') then generation,
       sum(status <> 'completed') then pending,
       max(packets) as packets
from ((select machine, status, 0 as packets
       from machine_status
      ) union all
      (select machine, 0, packets
       from machine_packets
      )
     ) m
group by machine;

这将在任一表中包括所有计算机的行。