Changes for Build #1391 [cws] HIVE-2928. Support for Oracle-backed Hive-Metastore (longvarchar to clob in package.jdo) (Mithun Radhakrishnan and Andrew Bayer via cws)
[cws] HIVE-2646. Hive Ivy dependencies on Hadoop should depend on jars directly, not tarballs (Andrew Bayer and Thomas Weise) [hashutosh] HIVE-2970: several jars in hive tar generated are not required (Thejas Nair via Ashutosh Chauhan) Changes for Build #1392 [cws] HIVE-2803 [jira] utc_from_timestamp and utc_to_timestamp returns incorrect results. (Kiyotaka Suzuki via Carl Steinbach) Summary: https://issues.apache.org/jira/browse/HIVE-2803 HIVE-2803 utc_from_timestamp and utc_to_timestamp returns incorrect results. It changed so that applyOffset() return a new instance. How to reproduce: $ echo "2011-12-25 09:00:00.123456" > /tmp/data5.txt hive> create table ts1(t1 timestamp); hive> load data local inpath '/tmp/data5.txt' overwrite into table ts1; hive> select t1, from_utc_timestamp(t1, 'JST'), from_utc_timestamp(t1, 'JST') from ts1 limit 1; The following result is expected: 2011-12-25 09:00:00.123456 2011-12-25 18:00:00.123456 2011-12-25 18:00:00.123456 However, the above query return incorrect result like this: 2011-12-26 03:00:00.492456 2011-12-26 03:00:00.492456 2011-12-26 03:00:00.492456 This is because GenericUDFFromUtcTimestamp.applyOffset() does setTime() improperly. On evaluating query, timestamp argument always returns the same instance. GenericUDFFromUtcTimestamp.applyOffset() does setTime() on the instance. That means it adds all offsets in the query. Test Plan: EMPTY Reviewers: JIRA Differential Revision: https://reviews.facebook.net/D1959 [namit] HIVE-2703 ResultSetMetaData.getColumnType() always returns VARCHAR(string) for partition columns irrespective of partition column type (tamtam180 via namit) [hashutosh] HIVE-2883 [jira] Metastore client doesnt close connection properly Summary: https://issues.apache.org/jira/browse/HIVE-2883 Here is the patch which gets rid of this problem as per my hypothesis. I also tested it and the above noted stack trace doesnt occur anymore. While closing connection, it always fail with following trace. Seemingly, it doesnt have any harmful effects. 12/03/20 10:55:02 ERROR hive.metastore: Unable to shutdown local metastore client org.apache.thrift.transport.TTransportException: Cannot write to null outputStream at org.apache.thrift.transport.TIOStreamTransport.write(TIOStreamTransport.java:142) at org.apache.thrift.protocol.TBinaryProtocol.writeI32(TBinaryProtocol.java:163) at org.apache.thrift.protocol.TBinaryProtocol.writeMessageBegin(TBinaryProtocol.java:91) at org.apache.thrift.TServiceClient.sendBase(TServiceClient.java:62) at com.facebook.fb303.FacebookService$Client.send_shutdown(FacebookService.java:421) at com.facebook.fb303.FacebookService$Client.shutdown(FacebookService.java:415) at org.apache.hadoop.hive.metastore.HiveMetaStoreClient.close(HiveMetaStoreClient.java:310) Test Plan: EMPTY Reviewers: JIRA, cwsteinbach Reviewed By: cwsteinbach Differential Revision: https://reviews.facebook.net/D2613 All tests passed The Apache Jenkins build system has built Hive-trunk-h0.21 (build #1392) Status: Still Failing Check console output at https://builds.apache.org/job/Hive-trunk-h0.21/1392/ to view the results.