incubator-doris Data loaded into wrong partition

lmyy7pcs  于 2022-04-22  发布在  Java
关注(0)|答案(0)|浏览(222)

Describe the bug

There is a partitioned table, partition by stat_date.
show create table like this:

CREATE TABLE t (
p_id_ bigint(20) NOT NULL ,
v_id_mod__ bigint(20) NOT NULL ,
v_id__ bigint(20) NOT NULL ,
s_id_ bigint(20) NOT NULL ,
stat_date_ bigint(20) NOT NULL ,
) ENGINE=OLAP
DUPLICATE KEY(p_id_, v_id_mod__, v_id__, s_id_, stat_date_)
PARTITION BY RANGE(stat_date_) (
PARTITION p20181201 VALUES LESS THAN ("1543680000"),
PARTITION p20181202 VALUES LESS THAN ("1543766400"),
PARTITION p20181203 VALUES LESS THAN ("1543852800"),
PARTITION p20181204 VALUES LESS THAN ("1543939200"),
PARTITION p20181205 VALUES LESS THAN ("1544025600"),
PARTITION p20190805 VALUES LESS THAN ("1565020800"),
PARTITION p20190806 VALUES LESS THAN ("1565107200"),
PARTITION p20190807 VALUES LESS THAN ("1565193600"),
PARTITION p20190808 VALUES LESS THAN ("1565280000"),
PARTITION p20190809 VALUES LESS THAN ("1565366400"),
PARTITION p20190810 VALUES LESS THAN ("1565452800"),
PARTITION p20190811 VALUES LESS THAN ("1565539200"))
DISTRIBUTED BY HASH(v_id__) BUCKETS 32
PROPERTIES (
"storage_type" = "COLUMN"
);

there is two sql result in inconsistent results:
select stat_date_,sum(1) from f group by stat_date_;
+------------+-----------+
| stat_date_ | sum(1) |
+------------+-----------+
| 1565366400 | 232581496 |
+------------+-----------+

select stat_date_,sum(1) from f where stat_date_=1565366400 group by stat_date_;
+------------+-----------+
| stat_date_ | sum(1) |
+------------+-----------+
| 1565366400 | 203781984 |
+------------+-----------+

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

explain the two sql, result shows that:

  1. the first sql read data from two partition
  2. the second sql read data from one partition

and more, select count(*) from two different partition equals to the first sql result.
So, the probably reason for the inconsistency is the data is loaded into wrong partition.

暂无答案!

目前还没有任何答案,快来回答吧!

相关问题