[
https://issues.apache.org/jira/browse/DRILL-2408?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14506031#comment-14506031
]
Aman Sinha commented on DRILL-2408:
-----------------------------------
Can you test your patch with a CTAS that has parallelism where some parquet
writer minor fragments don't have any input data ? Those writers would have
normally created 0 length files. Here's an example test case:
- create a table T1 (a1) with 100 rows where a1 has only 2 unique values
- alter session set `planner.slice_target` = 1
- CREATE TABLE T2 AS SELECT a1, count(*) FROM T1 GROUP BY a1
> CTAS should not create empty folders when underlying query returns no results
> -----------------------------------------------------------------------------
>
> Key: DRILL-2408
> URL: https://issues.apache.org/jira/browse/DRILL-2408
> Project: Apache Drill
> Issue Type: Bug
> Components: Storage - Writer
> Affects Versions: 0.8.0
> Reporter: Aman Sinha
> Assignee: Aman Sinha
> Priority: Critical
> Fix For: 1.0.0
>
> Attachments: DRILL-2408.1.patch.txt, DRILL-2408.2.patch.txt,
> DRILL-2408.3.patch.txt, DRILL-2408.4.patch.txt
>
>
> {noformat}
> 0: jdbc:drill:schema=dfs> select c_integer, c_bigint, c_date, c_time,
> c_varchar from j4 where c_bigint is null;
> +------------+------------+------------+------------+------------+
> | c_integer | c_bigint | c_date | c_time | c_varchar |
> +------------+------------+------------+------------+------------+
> +------------+------------+------------+------------+------------+
> No rows selected (0.126 seconds)
> 0: jdbc:drill:schema=dfs> create table ctas_t6(c1,c2,c3,c4,c5) as select
> c_integer, c_bigint, c_date, c_time, c_varchar from j4 where c_bigint is null;
> +------------+---------------------------+
> | Fragment | Number of records written |
> +------------+---------------------------+
> | 0_0 | 0 |
> +------------+---------------------------+
> 1 row selected (0.214 seconds)
> 0: jdbc:drill:schema=dfs> select * from ctas_t6;
> Query failed: IndexOutOfBoundsException: Index: 0, Size: 0
> Error: exception while executing query: Failure while executing query.
> (state=,code=0)
> {noformat}
> parquet file was not created, but directory was:
> {noformat}
> [Mon Apr 06 09:03:41
> root@/mapr/vmarkman.cluster.com/drill/testdata/joins/ctas_t6 ] # pwd
> /mapr/vmarkman.cluster.com/drill/testdata/joins/ctas_t6
> [Mon Apr 06 09:03:45
> root@/mapr/vmarkman.cluster.com/drill/testdata/joins/ctas_t6 ] # ls -l
> total 0
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
