如何按状态比较时间戳来计算正常运行时间

问题描述

我有一个记录 http 调用的表。这是我的桌子的样子：

callTimestamp	httpStatus	端点
2021-04-01 06:00:00	200	https://someserver/someapi/v1/endpoint1
2021-04-01 10:21:11	200	https://someserver/someapi/v1/endpoint1
2021-04-01 10:25:00	500	https://someserver/someapi/v1/endpoint1
2021-04-01 11:33:15	200	https://someserver/someapi/v1/endpoint1
2021-04-01 11:34:31	200	https://someserver/someapi/v1/endpoint1
2021-04-01 11:35:22	500	https://someserver/someapi/v1/endpoint1
2021-04-01 12:22:54	200	https://someserver/someapi/v1/endpoint1
2021-04-01 10:21:11	200	https://someserver/someapi/v1/endpoint2
2021-04-01 10:25:32	500	https://someserver/someapi/v1/endpoint2
2021-04-01 10:59:12	200	https://someserver/someapi/v1/endpoint2

我需要计算正常运行时间率，即：

1. 以秒为单位的总正常运行时间率 (100%) 为 TIME_TO_SEC(TIMEDIFF(Now(),MIN(callTimestamp)))

2. 需要用 httpStatus 500 计算记录之间的（停机时间）差异，然后用按端点分组的 httpStatus 200 计算。 从上面端点 1 的示例中，我有两次出现：

   2021-04-01 11:33:15 - 2021-04-01 10:25:00 = 4095 秒

   2021-04-01 12:22:54 - 2021-04-01 11:35:22 = 2852 秒

预期结果示例
考虑
MIN(callTimestamp) = 2021-04-01 06:00:00
现在() = 2021-04-01 22:00:00

TIME_TO_SEC(TIMEDIFF(Now(),MIN(callTimestamp))) = 57600 秒（这是我的 100%）

端点 1 = 57600 - 6947 = 50563 秒正常运行时间
端点 2 = 57600 - 2020 = 55580 秒正常运行时间

uptimeRate	端点
87.93	https://someserver/someapi/v1/endpoint1
96.49	https://someserver/someapi/v1/endpoint2

我不知道如何按端点计算停机时间。

对我如何做到这一点有任何想法吗？

P.S：我使用的是 MysqL 8.0.20

解决方法

您可以使用lead()over()计算下一个callTimestamp和min()over()窗口函数来选择最小值callTimestamp。然后通过分组和聚合，您可以获得所需的内容。

架构和插入语句：

 create table mytable(callTimestamp datetime,httpStatus int,endpoint varchar(100));
 insert into mytable values('2021-04-01 10:21:11',200,'https://someserver/someapi/v1/endpoint1');
 insert into mytable values('2021-04-01 10:25:00',500,'https://someserver/someapi/v1/endpoint1');
 insert into mytable values('2021-04-01 11:33:15','https://someserver/someapi/v1/endpoint1');
 insert into mytable values('2021-04-01 11:34:31','https://someserver/someapi/v1/endpoint1');
 insert into mytable values('2021-04-01 11:35:22','https://someserver/someapi/v1/endpoint1');
 insert into mytable values('2021-04-01 12:22:54','https://someserver/someapi/v1/endpoint1');
 insert into mytable values('2021-04-01 10:21:11','https://someserver/someapi/v1/endpoint2');
 insert into mytable values('2021-04-01 10:25:32','https://someserver/someapi/v1/endpoint2');
 insert into mytable values('2021-04-01 10:59:12','https://someserver/someapi/v1/endpoint2');

查询：

 select endpoint,100*(TIME_TO_SEC(TIMEDIFF(NOW(),max(mincalltime)))-sum(TIME_TO_SEC(TIMEDIFF(nexttime,calltimestamp))))/ TIME_TO_SEC(TIMEDIFF(NOW(),max(mincalltime))) uptimeRate
 
 
 from
 (select *,lead(calltimestamp)over (partition by endpoint order by calltimestamp)nexttime,min(calltimestamp)over(partition by endpoint order by calltimestamp) mincalltime
 from mytable 
 )t
 where httpstatus=500
 group by endpoint

输出：

端点	uptimeRate
https://someserver/someapi/v1/endpoint1	98.8895
https://someserver/someapi/v1/endpoint2	99.6771

dbhere