Druid doesn't support aggregation on top of the aggregated data

[beepupa]

versions:presto:0.237,druid:0.17 conf in presto in druid

[mbasmanova]

CC: @zhenxiao @luohao

[beepupa]

why add DATE_TRUNC('MONTH',"time") "time" is work

[beinan]

why add DATE_TRUNC('MONTH',"time") "time" is work

For this query, druid connector is not pushing down the aggregation group by to druid, because there is 'unsupported' function in your group-by clause. In this case, druid connector will fetching all the rows from druid and perform the aggregation on the presto side.

Why your original is not working: Because druid connector is trying to push down the 'group by' clause to druid, but looks like we thought druid cannot do aggregation on aggregated data ( as well as do aggregation on the data with limit). Anyway, I think it might be something we can improve. At least, we should not make the original query fail I think.

[beepupa]

为什么添加DATE_TRUNC（'MONTH'，“ time”）“ time”是可行的

对于此查询，group by由于您的group-by子句中存在“不受支持”的功能，因此druid连接器不会将聚合向下推至druid。在这种情况下，德鲁伊连接器将从德鲁伊获取所有行，并在presto一侧执行聚合。

为什么您的原始 文档不起作用：因为druid连接器试图将'group by'子句下推到druid，但看起来我们认为druid无法对聚合数据进行聚合（以及对具有限制的数据进行聚合）。无论如何，我认为这可能是我们可以改进的。至少，我认为我们不应使原始查询失败。

Thank you for your reply. I found that many statements can be executed in Druid but not in Presto. Looking forward to the optimized version

[beinan]

Hey @beepupa many thanks for reporting these to us! I would keep this issue updated once we made any progress.

I really appreciate if you can post the problematic queries statements you found under this issue. It would be very helpful, thanks!

[beepupa]

嘿@beepupa非常感谢您向我们报告这些！一旦取得任何进展，我将保持此问题的更新。

如果您可以发布在此问题下发现的有问题的查询语句，我将不胜感激。这将非常有帮助，谢谢！