Question

我正在尝试运行一个SQL查询来查找数据库中每个错误代码的计数。我有两张桌子

sw_sms_events，其中存储了已发送的事务ID和短信。
sw_events其中的事务ID和错误原因如果失败则存储，否则原因始终是＆＃34;成功发送TariffText＆＃34;。

总错误计数： - 从sw_sms_events中选择count（*），其中sms_text喜欢＆＃39;％Welkom in het buitenland％＆＃39;

每个错误原因的总错误数： -

select distinct count(*) over (partition by b.reason) , b.reason 
from sw_sms_events a, sw_events b 
where a.transaction_id= b.transaction_id  
and a.sms_text like '%Welkom in het buitenland%' 
and b.reason !='Successfully Sent TariffText'  
order by (count(*) over (partition by b.reason)) desc

通常这些查询给出相同的结果，即单个错误计数的总和=错误总数。但是在最坏情况下多次重试相同事务的情况下，结果不相同。我们在表中有多行具有相同的事务ID。

以下是最坏情况下的结果之一：

Name    24-07-2015
Total Number of SMSWelcome Sent 156788
Total Number of Error SMSWelcome    1738
Total Number of SMSWelcome Sent with null Tariffs   286



Error Reason    Error Count

Unknown error received :BEA-380000 , ErrorMessage : BSL-99999   1829
Backend system not available , ErrorMessage : BSL-50002 641
Remote Error    527

NativeQuery.executeQuery failed , ErrorMessage : BSL-11009  41
This service is available only for active products , ErrorMessage : BSL-15024   30

Unknown error received :BEA-382556 , ErrorMessage : BSL-99999   18

Customer information: Not retrieved. This action cannot continue without customer information. Please try later or contact your system administrator. , ErrorMessage : BSL-10004    13

OMS login failure: Problem in OMS UAMS login - Nested Exception/Error: java.net.ConnectException: Tried all: '1' addresses, but could not connect over HTTP to server: '195.233.102.177', port: '40123' ,   12

t3://195.233.102.171:30101: Bootstrap to: 195.233.102.171/195.233.102.171:30101' over: 't3' got an error or timed out , ErrorMessage : BSL-11000    5
getTariffsAndAddOns, status: Failure , ErrorCode : An internal error occured , ErrorMessage : BSL-14005 3

Authorization failed of dealer market restrictions , ErrorMessage : BSL-50005   2

com.amdocs.cih.exception.InvalidUsageException: The input parameter AssignedProductRef is invalid. , ErrorMessage : BSL-10004   1

我的问题是我如何修改当前的sql，使得当我们遇到表中同一事务多次的错误情况时，错误总数应始终等于单个错误计数的总和

Answer 1

我真的不明白为什么要使用分析查询。是不是更简单group by？

select count(*), b.reason 
from sw_sms_events a, sw_events b 
where a.transaction_id= b.transaction_id  
and a.sms_text like '%Welkom in het buitenland%' 
and b.reason !='Successfully Sent TariffText'  
group by b.reason
order by count(*) desc

当您说我们在表中有多个具有相同事务ID 的行时，您的意思是仅在sw_events表中，还是在sw_sms_events和sw_events表中？

如果是这样，事件会被计算多次，因为您在具有相同transaction_id的所有原始数据上执行笛卡尔积。您应该使用更严格的连接子句。

你也可以做一些事情（非常难看），如：

select count(distinct b.ROWID), b.reason 
from sw_sms_events a, sw_events b 
where a.transaction_id= b.transaction_id  
and a.sms_text like '%Welkom in het buitenland%' 
and b.reason !='Successfully Sent TariffText'  
group by b.reason
order by count(distinct b.ROWID) desc

确保每个事件只计算一次。

Answer 2

select distinct count(distinct b.ROWID) over (partition by b.reason) , b.reason 
from sw_sms_events a, sw_events b 
where a.transaction_id= b.transaction_id  
and a.sms_text like '%Welkom in het buitenland%' 
and b.reason !='Successfully Sent TariffText'  
order by (count(distinct b.ROWID) over (partition by b.reason)) desc

Oracle SQL：每个错误代码错误计数错误

2 个答案: