Re: 转发: issue: the same cube in diffent project response different result.

[Xiaoxiang Yu <xi...@kyligence.io>]

Hi liang gang,
Using you sample data, sql query and hive ddl, Kylin gave the same result as Hive in my sandbox env. So I think I think maybe you should double check your codebase or env. If you confirm the result is different from the result return by hive in the future, please report a JIRA and attach required files.

----------------
Best wishes,
Xiaoxiang Yu


发件人: "lianggang@qutoutiao.net" <li...@qutoutiao.net>
日期: 2018年12月15日 星期六 18:43
收件人: Xiaoxiang Yu <xi...@kyligence.io>
抄送: "dev@kylin.apache.org" <de...@kylin.apache.org>
主题: 转发: issue: the same cube in diffent project response different result.

Hi  xiaoxiang,

Thanks for your response.  I provide attaching data file in case you can check my issue. Thanks.
“GRAPH_DETAIL_CUBE_ VIEW” is the fact table, and “DIM_METRIC_INFO” is the dimension table, it uses “inner join” in kylin Model.
When I use the 1.sql  to query ,it responses one data. Like below snapshot:
[cid:image005.jpg@01D4949A.40AAB230]

When I add ‘7464’ in “exp_id” filter like 2.sql to query,  it responses no data. Like below snapshot:
[cid:image008.jpg@01D4949A.40AAB230]

Below columns are measure columns. All of others are dimension columns.
SAMPLE_SIZE
SUM_X
SUM_SQUARE_X
ACT_DEV
[cid:image001.jpg@01D494A1.F2122D30]

I have checked it in kylin version 2.3.2 and 2.5.0, it has the same issue.

So I think it’s a bug of kylin. Please help me check, thanks!

发件人: Xiaoxiang Yu [mailto:xiaoxiang.yu@kyligence.io]
发送时间: 2018年12月13日 11:13
收件人: user@kylin.apache.org<ma...@kylin.apache.org>
抄送: lianggang@qutoutiao.net<ma...@qutoutiao.net>
主题: Re: issue: the same cube in diffent project response different result.

Hi, lianggang

I haven’t see such problem. Could you please provide following detail for deeper research:

1.       Sample data which could can reproduce inconsistent result.(Maybe a csv/json file contains data.)

2.       Your model and cube metadata (some json file ).

3.       Your sql query which result is not correct.(in a sql file)

4.       Your Kylin version,  Spark/Hbase/MapReduce version

If you confirm inconsistent result can be reproduced on sample data, please open a JIRA ticket on https://issues.apache.org/jira/projects/KYLIN and attach your files.

----------------
Best wishes,
Xiaoxiang Yu


发件人: "lianggang@qutoutiao.net<ma...@qutoutiao.net>" <li...@qutoutiao.net>>
答复: "user@kylin.apache.org<ma...@kylin.apache.org>" <us...@kylin.apache.org>>
日期: 2018年12月12日 星期三 17:43
收件人: "user@kylin.apache.org<ma...@kylin.apache.org>" <us...@kylin.apache.org>>
抄送: "lianggang@qutoutiao.net<ma...@qutoutiao.net>" <li...@qutoutiao.net>>
主题: issue: the same cube in diffent project response different result.

Hi All,

I encountered one strange problem for my cube data. Please help me check what’s wrong with Kylin.

I created two same cube in “ABtest” and “ABtest_prod” project, the cube structure is the same. built the same time range’s data. The data size is the same. But Using the same SQL to query, the response data is different. The snapshot like below:

ABtest project’s snapshot:
[cid:image001.png@01D4923F.A7D44B90]

ABtest_prod project’s snapshot:
[cid:image002.png@01D4923F.A7D44B90]

One other strange thing, when I use “trim(a.os)” in the query filter of ABtest_prod project, the response rows is 20. And data is correct. The snapshot like below:
[cid:image003.png@01D49241.0DC91B00]



I have checked it for a long time. I confirm the cube structure is the same, the time range and data size is also the same. I use the version 2.3.2 of Kylin.
Currently I don’t know what’s the reason. Please help me. Thank you very much!