MySQL 基于具有多个 return 值的子 select 删除语句，同时为每个唯一值保留 500 条记录

Question

MySQL 删除基于具有多个 return 值的子 select 的语句。

这是我现在正在做的事情：

DELETE FROM `dnsstats` WHERE id NOT IN
(
 SELECT id FROM
    (
    SELECT id FROM `dnsstats` WHERE peerhost = 'x.x.x.243' ORDER BY id DESC LIMIT 500
    ) foo 
)
 AND id NOT IN
(
 SELECT id FROM
    (
    SELECT id FROM `dnsstats` WHERE peerhost = 'x.x.x.40' ORDER BY id DESC LIMIT 500
  ) foo2
)
 AND id NOT IN
(
 SELECT id FROM
    (
    SELECT id FROM `dnsstats` WHERE peerhost = 'x.x.x.50' ORDER BY id DESC LIMIT 500
  ) foo3
);

效果很好，但我想 select 动态地 select 个人 IP 地址，如下所示：

SELECT peerhost FROM `dnsstats` GROUP BY peerhost;

..并根据那些 returned 值删除。

这是我从上一个问题中得到的：

DELETE FROM `dnsstats` WHERE id NOT IN
(
 SELECT id FROM
    (
    SELECT id FROM `dnsstats` WHERE peerhost in (
            SELECT peerhost FROM `dnsstats` GROUP BY peerhost
        ) ORDER BY id DESC LIMIT 500
    ) foo 
);

..但这给我留下了 500 条记录。我想为每个 peerhost

保留 500 条记录

知道我可以做些什么来完成这项工作吗？谢谢！

原问题：

Answer 1

我会使用变量来做到这一点。下面枚举每个peerhost:

select s.*,
       (@rn := if(@ph = peerhost, @rn + 1,
                  if(@ph := peerhost, 1, 1)
                 )
       ) as rn
from dnstats s cross join
     (select @ph := '', @rn := 0) params
order by peerhost, id desc;

您可以在使用子查询的 delete 中使用它：

delete s
    from dnstats s join
         (select s.*,
                 (@rn := if(@ph = peerhost, @rn + 1,
                            if(@ph := peerhost, 1, 1)
                           )
                 ) as rn
          from dnstats s cross join
               (select @ph := '', @rn := 0) params
          order by peerhost, id desc
         ) es
         on es.id = s.id
    where rn > 500;

这将删除每个 peerhost 的前 500 行以外的所有行。

MySQL 基于具有多个 return 值的子 select 删除语句，同时为每个唯一值保留 500 条记录

MySQL delete statement based on sub-select with multiple return values while retaining 500 records for each unique value

mysql

sql

subquery

limit

sql-delete