Influxdb 总和小于XDB测量中标签和的和？

或者我做错了什么。我有一个服务，统计它收到的请求。请求具有平台、执行请求的客户端应用程序版本和其他标记。当服务重新启动时（很少在更新时发生，指标会重置）

因此，我想计算最近时间范围内来自每个平台的查询的百分比，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

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

并在不分组的情况下执行相同操作：

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)
);

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

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平台更多的请求。但我甚至不知道哪一个是不正确的，总的还是按平台还是两者都不正确

---

如何正确计算请求总数和按平台计算请求总数

好的，问题是因为我的测量有其他标签，比如服务器和应用程序版本，每个标签都有单独的计数器，然后它们都变成交错的，可以在图表上看到，这应该是平滑的和累积的，但非常尖锐：

但是当我们添加

groupby*

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

它分为多个单独的平滑系列：

现在，这是可微的，从中我们可以创建子查询来进行聚合

总数：

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，我想故事应该是这样的：每次你想区分有一些标签的计数器时，你都需要

GROUP BY*

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