Question

我正在学习PostgreSQL并且我有一个包含以下列的数据库日志：路径文本，ip inet，方法文本，状态文本，time timestamptz，id int

使用此数据库，我的目标是查找错误百分比（状态！='200 OK'/状态）超过2.5％的任何日期。我有一个查询，如下所示，以十进制形式获得错误百分比，但我无法弄清楚如何将它们乘以10以使它们成为'百分比'，以及如何检查它们是否超过2.5。

我已经在多个位置尝试了HAVING和WHERE，但是无论我进入HAVING还是WHERE，我都会在group by或者语法错误中遇到编程错误。将其更改为百分比并检查其是否超过2.5的正确方法是什么？提前致谢

模拟数据：

CREATE TABLE log (
    path text,
    ip inet,
    method text,
    status text,
    "time" timestamp with time zone DEFAULT now(),
    id integer NOT NULL
);

INSERT INTO log VALUES ('/article/bears-love-berries', '198.51.100.76' , 'GET', '200 OK', '2016-07-1 12:54:22+00', 3355597);
INSERT INTO log VALUES ('/article/candidate-is-jerk', '198.51.100.76' , 'GET', '200 OK', '2016-07-1 19:54:22+00', 3355598);
INSERT INTO log VALUES ('/article/bears-love-berries', '198.51.100.76' , 'GET', '404 NOT FOUND', '2016-07-1 22:54:22+00', 3355599);
INSERT INTO log VALUES ('/article/goats-eat-googles', '203.0.113.42' , 'GET', '200 OK', '2016-07-11 12:54:22+00', 3355600);
INSERT INTO log VALUES ('/article/candidate-is-jerk', '203.0.113.42' , 'GET', '200 OK', '2016-07-11 17:54:22+00', 3355601);
INSERT INTO log VALUES ('/article/bears-love-berries', '203.0.113.42' , 'GET', '200 OK', '2016-07-11 19:54:22+00', 3355602);
INSERT INTO log VALUES ('/article/goats-eat-googles', '203.0.113.42' , 'GET', '404 NOT FOUND', '2016-07-15 13:54:22+00', 3355603);
INSERT INTO log VALUES ('/article/bears-love-berries', '198.51.100.34' , 'GET', '200 OK', '2016-07-15 20:54:22+00', 3355604);
INSERT INTO log VALUES ('/article/candidate-is-jerk', '198.51.100.34' , 'GET', '200 OK', '2016-07-15 23:54:22+00', 3355605);
INSERT INTO log VALUES ('/article/candidate-is-jerk', '203.0.113.12' , 'GET', '200 OK', '2016-07-21 19:54:22+00', 3355606);
INSERT INTO log VALUES ('/article/bears-love-berries', '203.0.113.12' , 'GET', '200 OK', '2016-07-21 12:54:22+00', 3355607);
INSERT INTO log VALUES ('/article/goats-eat-googles', '198.51.100.76' , 'GET', '200 OK', '2016-07-21 01:54:22+00', 3355608);
INSERT INTO log VALUES ('/article/goats-eat-googles', '198.51.100.76' , 'GET', '404 NOT FOUND', '2016-07-29 07:54:22+00', 3355609);
INSERT INTO log VALUES ('/article/candidate-is-jerk', '198.51.100.76' , 'GET', '404 NOT FOUND', '2016-07-29 20:54:22+00', 3355610);
INSERT INTO log VALUES ('/article/goats-eat-googles', '198.51.100.34' , 'GET', '200 OK', '2016-07-29 19:54:22+00', 3355611);

查询：

SELECT DATE(time) AS day,
(
  (SELECT CAST(COUNT(status) AS FLOAT)
  FROM log WHERE status != '200 OK') /
  (SELECT CAST(COUNT(status) AS FLOAT))
) AS views
FROM log GROUP BY day
ORDER BY views

Answer 1

在sqlfiddle中共享一个测试用例之后，我们可以开始使用正在运行的代码。

小提琴在这里：http://sqlfiddle.com/#!15/b3ed9/39

不是将结果乘以100，而是可以检查已经乘以的0.025，不是吗？我的意思是，如果你的值在0-1之间，相当于2.5％的值只有0.025。

因此该命令应该：

select views.day, views.cnt from
 (select datelog.day,
  (
   cast(count(case when datelog.status != '200 OK' then 1 else null end) as float)
   /
   cast(count(*) as float)
  ) as cnt
  from (select DATE(time) as day, status from log) as datelog
  group by datelog.day
 ) as views
 where views.cnt > 0.025

最后一次修订在这里的回复中有深刻的基础：SQL use column from subselect in where clause

在SQL中编写子查询

1 个答案: