首页 > 解决方案 > 总和小于 InfluxDB 测量中标签的总和?

问题描述

或者我做错了什么。我有一项服务可以计算它收到的请求。请求具有平台、执行它们的客户端应用程序的版本以及其他标签。当服务重新启动时(在更新时很少发生这种情况,指标会被重置)。

因此,我想计算最近时间范围内来自每个平台的查询百分比,并执行以下操作:

SELECT SUM("received")  as "received"
FROM (
  SELECT NON_NEGATIVE_DIFFERENCE(MAX("received")) as "received"
  FROM  "http_metrics"
  WHERE time >= now() - 4h GROUP BY time(1s)
) GROUP BY "platform";

返回:

...
tags: platform=ios
time                 received
----                 --------
1970-01-01T00:00:00Z 581

tags: platform=unknown
time                 received
----                 --------
1970-01-01T00:00:00Z 12310

tags: platform=web
time                 received
----                 --------
1970-01-01T00:00:00Z 6196

在不分组的情况下做同样的事情:

SELECT SUM("received")  as "received"
FROM (
  SELECT NON_NEGATIVE_DIFFERENCE(MAX("received")) as "received"
  FROM  "http_metrics"
  WHERE time >= now() - 4h GROUP BY time(1s)
);

返回:

time                 received
----                 --------
1970-01-01T00:00:00Z 8274

这显然是不正确的,因为“未知”平台无法接收比所有请求更多的请求。但我什至不知道哪个是不正确的,完全的还是按平台的,或者两者兼而有之?

如何正确计算请求的总数和按平台的总和?

标签: influxdb

解决方案


好的,所以问题是,由于我的测量有其他标签,例如服务器和应用程序版本,以及每个标签的单独计数器,所以它们都变得交错,可以在图表上看到,这应该是平滑和累积的,但是非常尖锐: 接收事件测量

但是当我们添加GROUP BY *

SELECT "received" FROM  "http_metrics"  WHERE $timeFilter GROUP BY *;

它分成许多单独的平滑系列:

计数器增加缓慢,所以看水平

现在,这是可区分的,我们可以从中创建子查询来聚合。

全部的:

SELECT SUM("received") as "received" FROM (
    SELECT NON_NEGATIVE_DIFFERENCE(MAX("received")) as "received"
    FROM "http_metrics"
    WHERE time >= now() - 6h
    GROUP BY time(1s), *
) WHERE  time >= now() - 6h;

time                           received
----                           --------
2018-07-17T01:07:46.184292033Z 1367

分组:

SELECT SUM("received") as "received" FROM (
    SELECT NON_NEGATIVE_DIFFERENCE(MAX("received")) as "received"
    FROM "http_metrics"
    WHERE time >= now() - 6h
    GROUP BY time(1s), *
) WHERE  time >= now() - 6h
GROUP BY "platform";

... I'm not going to bore you with response, but sum matches total sum.

所以,我猜道德故事应该是:每次你想区分有标签的计数器时,你需要GROUP BY *


推荐阅读