mysql 多个分组在一个 table

mysql multiple grouping on one table

我有一个tablejob_result

CREATE TABLE job_result  (
  node varchar(20) DEFAULT NULL,
  jobId int(10) DEFAULT NULL,
  subResult int(1) DEFAULT NULL
) 

table 包含很多节点。

每个节点有多个jobId。

每个jobId有多个subResults(通常少于10个)

提供了一些示例数据

insert into job_result values ('A', 14, 0);
insert into job_result values ('A', 15, 0);
insert into job_result values ('A', 16, 1);
insert into job_result values ('A', 17, 0);
insert into job_result values ('A', 18, 1);
insert into job_result values ('A', 19, 1);
insert into job_result values ('A', 20, 0);


insert into job_result values ('B', 1, 0);
insert into job_result values ('B', 2, 0);
insert into job_result values ('B', 3, 1);
insert into job_result values ('B', 4, 1);
insert into job_result values ('B', 5, 1);
insert into job_result values ('B', 6, 1);
insert into job_result values ('B', 7, 1);


insert into job_result values ('C', 10, 0);
insert into job_result values ('C', 11, 0);
insert into job_result values ('C', 12, 0);
insert into job_result values ('C', 13, 0);
insert into job_result values ('C', 14, 0);
insert into job_result values ('C', 15, 0);
insert into job_result values ('C', 16, 0);

我想要一个查询在每个节点上给我一个 "health check" 以查看某个节点是否有 X 个连续失败的作业。失败的作业是至少有一个非零结果的 jobId。

SELECT node, jobId, SUM(subResult) res
FROM job_result WHERE node = 'A' 
GROUP BY jobId 
ORDER BY jobId desc 
LIMIT 5

这为我提供了节点 A 上最后 5 个作业的结果,其中 0 表示作业很好。

+------+---------+------+
| node | jobId   | res  |
+------+---------+------+
| A    |   20    |   0  |
| A    |   19    |   1  |
| A    |   18    |   1  |
| A    |   17    |   0  |
| A    |   16    |   1  |

然后我需要对每个节点的结果求和并分组到一行中,我做到了这一点:

SELECT node, SUM(I.res) res
FROM
(SELECT node, sum(subResult) res
FROM job_result WHERE node = 'A' 
GROUP BY jobId 
ORDER BY jobId desc 
LIMIT 5) I

+------+------+
| node | res  |
+------+------+
| A    |  3   |

我的问题是:如何将此查询扩展为 return 每个节点一行? 我已经尝试了几天相关的子查询和连接,但我总是失败,因为内部 select 看不到要 select 的哪个节点。 到位后,我将努力将 LIMIT 5 更改为动态值。

想要的结果就像

+------+------+
| node | res  |
+------+------+
| A    |  3   |
| B    |  0   |
| C    |  1   |
| E    |  0   |

每个节点一行,右列显示失败次数

请指教!

在 MySQL 中,在大型 table 中,获得最后 5 个结果的最佳方法是使用变量来枚举结果。剩下的只是聚合:

SELECT node, SUM(subr > 0) as numfails
FROM (SELECT node, jobid, subr,
             (@rn := if(@n = node, @rn + 1,
                        if(@n := node, 1, 1)
                       )
             ) as rn
      FROM (SELECT node, jobid,
                   SUM(subresult) as subr
            FROM job_result jr
            GROUP BY node, jobid
           ) jr CROSS JOIN
           (SELECT @n := '', @rn := 0) init
      ORDER BY node, jobid desc
     ) nr
WHERE rn <= 5
GROUP BY node;