[
https://issues.apache.org/jira/browse/HIVE-14322?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15392865#comment-15392865
]
Thejas M Nair commented on HIVE-14322:
--------------------------------------
[~sershe] Sorry, I should have been more clear.
I didn't mean actually auto-creating the schema made the difference. What I
meant was using a schema created with schematool, but running metastore with
the -hiveconf datanucleus.schema.autoCreateColumns=true option made a
difference.
> Postgres db issues after Datanucleus 4.x upgrade
> ------------------------------------------------
>
> Key: HIVE-14322
> URL: https://issues.apache.org/jira/browse/HIVE-14322
> Project: Hive
> Issue Type: Bug
> Affects Versions: 2.0.0, 2.1.0, 2.0.1
> Reporter: Thejas M Nair
> Assignee: Thejas M Nair
> Attachments: HIVE-14322.1.patch
>
>
> With the upgrade to datanucleus 4.x versions in HIVE-6113, hive does not
> work properly with postgres.
> The nullable fields in the database have string "NULL::character varying"
> instead of real NULL values. This causes various issues.
> One example is -
> {code}
> hive> create table t(i int);
> OK
> Time taken: 1.9 seconds
> hive> create view v as select * from t;
> OK
> Time taken: 0.542 seconds
> hive> select * from v;
> FAILED: SemanticException Unable to fetch table v.
> java.net.URISyntaxException: Relative path in absolute URI:
> NULL::character%20varying
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
