AWS 时间流中的聚合计数错误导致错误

Question

我将遥测数据推送到 AWS 时间流：

measure_value::varchar	IP	time	measure_name
test.html	192.168.1.100	2021-05-25 14:27:45	hits
blah.html	192.168.1.101	2021-05-25 14:27:45	hits
test.html	192.168.1.102	2021-05-25 14:27:46	hits

我希望在时间流中显示数据的聚合，显示每小时每个 uri 的点击次数。

measure_value::varchar	Count	time
test.html	2	2021-05-25 14:00
blah.html	1	2021-05-25 14:00

我正在尝试使用：

SELECT measure_value::varchar as URIs, CREATE_TIME_SERIES(time, measure_value::varchar) AS served FROM $__database.$__table WHERE $__timeFilter group by measure_value::varchar

但我收到错误消息：

ValidationException: Duplicate timestamps are not allowed in a timeseries.

是我用错了函数还是我的数据有误？

===================

尝试@berto99 的解决方案...我得到：

SELECT measure_value::varchar AS URIs, date_trunc('hour', time) AS hour, count(measure_value::varchar) as queries
FROM $__database.$__table
WHERE $__timeFilter
GROUP BY measure_value::varchar, date_trunc('hour', time)

=====================

更新#2：

到达那里，仍然不是 100% 那里。

SELECT measure_value::varchar AS URIs, bin(time, 15m) AS hour, count(measure_value::varchar) as queries
FROM $__database.$__table
WHERE $__timeFilter
GROUP BY measure_value::varchar, bin(time, 15m) order by hour

Answer 1

我知道可能有更好的方法来做到这一点，但像这样：

SELECT measure_value::varchar AS URIs, date_trunc('hour', time) AS time
FROM $__database.$__table
WHERE $__timeFilter
GROUP BY measure_value::varchar, date_trunc('hour', time)
ORDER BY date_trunc('hour', time)

也许您还需要使用 date_trunc('hour', time at time zone '-X') 调整时区，其中 X 是您的时区

Answer 2

结合@Berto99 的建议和对 Whosebug 的更多挖掘，最终得到了这一切 - TimeStream + Grafana: not recognizing series in data.

您必须将 Berto99 的建议放入子查询，然后运行通过 CREATE_TIME_SERIES。最终查询结果为：

WITH binned_query AS (
SELECT measure_value::varchar AS URIs, bin(time, 15m) AS bin_time, count(measure_value::varchar) as queries
FROM $__database.$__table
WHERE $__timeFilter
GROUP BY measure_value::varchar, bin(time, 15m) order by bin_time
)
SELECT URIs, CREATE_TIME_SERIES(bin_time,queries) as Endpoint
FROM binned_query 
GROUP BY URIs

已从使用 date_trunc 更改为 bin，因为它可以让您更灵活地执行 15 分钟的间隔。

漂亮的图：

AWS 时间流中的聚合计数错误导致错误

Aggregating Counts in AWS Timestream Error Causes Errors

grafana

amazon-timestream