Lianggang created KYLIN-3733: -------------------------------- Summary: kylin can't response correct data when using "in" filter Key: KYLIN-3733 URL: https://issues.apache.org/jira/browse/KYLIN-3733 Project: Kylin Issue Type: Bug Affects Versions: v2.5.0, v2.3.2 Reporter: Lianggang Attachments: demo.txt
Hi guys, I used the attachment data to create one cube in kylin. I encounter a strange issue. when I use sql_1 to query, the data is correct. sql_1: _select exp_id,os,new_user_flag_ _,sum(sample_size)_ _,sum(sum_x)_ _,sum(sum_square_x)_ _,sum(act_dev)_ _from RPT_QUKAN.ABTEST_GRAPH_DETAIL_CUBE_MIDU_VIEW5_ _where refresh_dt ='2018-12-13'_ _and exp_id *in* ('8382')_ _group by exp_id,os,new_user_flag;_ when I use sql_2 to query, the data is *not* correct. sql_2: _select exp_id,os,new_user_flag_ _,sum(sample_size)_ _,sum(sum_x)_ _,sum(sum_square_x)_ _,sum(act_dev)_ _from RPT_QUKAN.ABTEST_GRAPH_DETAIL_CUBE_MIDU_VIEW5_ _where refresh_dt ='2018-12-13'_ _and exp_id *in* ('8382','8383')_ _group by exp_id,os,new_user_flag;_ when I add trim() for exp_id, the data is correct. like sql_3: _select exp_id,os,new_user_flag_ _,sum(sample_size)_ _,sum(sum_x)_ _,sum(sum_square_x)_ _,sum(act_dev)_ _from RPT_QUKAN.ABTEST_GRAPH_DETAIL_CUBE_MIDU_VIEW5_ _where refresh_dt ='2018-12-13'_ _and *trim*(exp_id) in ('8382','8383')_ _group by exp_id,os,new_user_flag;_ I am not sure whether the rowkey is too long for this result. please help check the issue. thanks! the table structure like below: CREATE TABLE `RPT_QUKAN.ABTEST_GRAPH_DETAIL_CUBE_MIDU_VIEW5` ( `exp_id` string ,`new_user_flag` string ,`os` string ,`sample_size` double ,`sum_x` double ,`sum_square_x` double ,`act_dev` bigint ,`data_dt` string ,`refresh_dt` string) the dimension columns: exp_id,new_user_flag,os,data_dt,refresh_dt the *sum* measure columns: sum_x,sample_size,sum_square_x,act_dev -- This message was sent by Atlassian JIRA (v7.6.3#76005)