[jira] [Commented] (HIVE-20342) Create table from CTAS with location fails for managed tables

["Pablo Idiaquez (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-20342?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16590372#comment-16590372 ] 

Pablo Idiaquez commented on HIVE-20342:
---------------------------------------

 

LOCATION in this case requires to include FULL HDFS URI Path to work without giving error
{code:java}
CREATE TABLE test4 ROW FORMAT SERDE 
'org.apache.hadoop.hive.serde2.columnar.ColumnarSerDe' STORED AS RCFILE 
LOCATION 'hdfs://NAMENODE:8020/tmp/test4' 
 AS SELECT * FROM test;
{code}
However If LOCATION is passed with the short format '/folder/subfolder'   ( no hdfs://NN prefix) . it will still fail

 

> Create table from CTAS with location fails for managed tables
> -------------------------------------------------------------
>
>                 Key: HIVE-20342
>                 URL: https://issues.apache.org/jira/browse/HIVE-20342
>             Project: Hive
>          Issue Type: Bug
>          Components: Beeline, HiveServer2
>    Affects Versions: 3.0.0
>            Reporter: Pablo Idiaquez
>            Priority: Blocker
>
> When attempting to create table from SELECT with LOCATION  (managed tables)
>  
> To reproduce CREATE source table CLASS with some data
> then use the code below to CREATE table TABLE42 from select on CLASS and LOCATION /tmp/test1
>  
> {code:java}
> CREATE TABLE TABLE42 ROW FORMAT SERDE
> 'org.apache.hadoop.hive.serde2.columnar.ColumnarSerDe' STORED AS RCFILE
> LOCATION '/tmp/test1'
> AS SELECT * FROM
> CLASS;
> {code}
>  
> it fails with ERROR:
> {code:java}
> ERROR : Job Commit failed with exception 'org.apache.hadoop.hive.ql.metadata.HiveException(The following files were committed but not found: [/tmp/test1/delta_0000001_0000001_0000/000000_0])' 
> org.apache.hadoop.hive.ql.metadata.HiveException: The following files were committed but not found: [/tmp/test1/delta_0000001_0000001_0000/000000_0] 
> at org.apache.hadoop.hive.ql.exec.Utilities.handleMmTableFinalPath(Utilities.java:4329) 
> at org.apache.hadoop.hive.ql.exec.FileSinkOperator.jobCloseOp(FileSinkOperator.java:1393) 
> {code}
>  
> If the same is attempted for an EXTERNAL table will work OK
> {code:java}
> CREATE EXTERNAL TABLE test3 ROW FORMAT SERDE 'org.apache.hadoop.hive.serde2.columnar.ColumnarSerDe' STORED AS RCFILE LOCATION '/tmp/test2' AS SELECT * FROM test;
> {code}
> if we CREATE table on statement 1, and INSERT from SELECT on statement 2 , will work OK.
> (Step1 CREATE TABLE)
> {code:java}
> CREATE TABLE `TABLE42`( 
> `COL1` double, 
> `COL2` varchar(8), 
> `COL3` varchar(1), 
> `COL4` double, 
> `COL5` double) 
> ROW FORMAT SERDE 
> 'org.apache.hadoop.hive.serde2.columnar.ColumnarSerDe' 
> STORED AS INPUTFORMAT 
> 'org.apache.hadoop.hive.ql.io.RCFileInputFormat' 
> OUTPUTFORMAT 
> 'org.apache.hadoop.hive.ql.io.RCFileOutputFormat' 
> LOCATION '/tmp/test1';
> {code}
> (Step2 INSERT from SELECT)
> {code:java}
> INSERT into TABLE42 select * FROM CLASS;
> {code}
> will work as expected.
>  
> Thanks
> Pablo
>  
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)