[
https://issues.apache.org/jira/browse/HIVE-3939?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15342662#comment-15342662
]
Xiao Yu edited comment on HIVE-3939 at 6/21/16 9:06 PM:
--------------------------------------------------------
[~navis] It appears this was a regression I'm currently on version 1.1.0 and
can still reproduce this issue on both partitioned and non-partitioned tables.
Issuing a `USE database;` then using non database prefixed table names works.
was (Author: xyu):
[~navis] It appears this was fixed by HIVE-3465 for partitioned tables however
I can still reproduce the `database.table` overwriting problem with a
non-partitioned table. (Version 1.1.0)
> INSERT INTO behaves like INSERT OVERWRITE if the table name referred is not
> all lowercase
> -----------------------------------------------------------------------------------------
>
> Key: HIVE-3939
> URL: https://issues.apache.org/jira/browse/HIVE-3939
> Project: Hive
> Issue Type: Bug
> Components: Database/Schema
> Affects Versions: 0.9.0
> Environment: Windows 2012, HDInsight
> Reporter: mohan dharmarajan
>
> If table referred does not use all lowercase in INSERT INTO command, the data
> is not appended but overwritten.
> set hive.exec.dynamic.partition.mode=nonstrict;
> set hive.exec.dynamic.partition=true;
> CREATE TABLE test (key int, value string) PARTITIONED BY (ds string);
> SELECT * FROM test;
> INSERT INTO TABLE test PARTITION (ds) SELECT key, value, value FROM src;
> SELECT * FROM test;
> The following statement works as expected. The data from src is appended to
> test
> SELECT * FROM test;
> INSERT INTO TABLE test PARTITION (ds) SELECT key, value, value FROM src;
> SELECT * FROM test;
> The following is copied from the processing log
> Loading data to table default.test partition (ds=null)
> Loading partition {ds=1}
> Loading partition {ds=2}
> The following statement does not work. Note the table name referred as Test
> (not test). INSERT INTO behaves like INSERT OVERWRITE
> SELECT * FROM test;
> INSERT INTO TABLE Test PARTITION (ds) SELECT key, value, value FROM src;
> SELECT * FROM test;
> The following is copied from the processing log
> Loading data to table default.test partition (ds=null)
> Moved to trash: hdfs://localhost:8020/hive/warehouse/test/ds=1
> Moved to trash: hdfs://localhost:8020/hive/warehouse/test/ds=2
> Loading partition {ds=1}
> Loading partition {ds=2}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
