Maksym Doroshenko created SQOOP-2131:
----------------------------------------
Summary: Export timestamp column from ORC file
Key: SQOOP-2131
URL: https://issues.apache.org/jira/browse/SQOOP-2131
Project: Sqoop
Issue Type: Bug
Affects Versions: 1.4.5
Environment: Hive-Hcatalog 0.13.0, MySQL DB
Reporter: Maksym Doroshenko
The following exception occurs when Hive (v. 0.13) table, which is stored in
ORC format and has a TIMESTAMP column, is exported with Sqoop (v.1.4.5) to
MySQL DB:
"ERROR tool.ExportTool: Encountered IOException running export job:
java.io.IOException: java.lang.TypeNotPresentException: Type timestamp not
present"
Distribution HDP v.2.1 (sandbox)…
I assume the cause is in the namespace.
As far as I know, to operate TIMESTAMP data type in Hive/Hcatalog since version
0.13 the new namespace should be used for HCatalog classes, namely
org.apache.hive.hcatalog instead of org.apache.hcatalog. To check which
namespace sqoop use I have deleted org.apache.hcatalog directory from
hive-hcatalog-core.jar and re-ran sqoop to have another exception
"Exception in thread "main" java.lang.NoClassDefFoundError:
org/apache/hcatalog/mapreduce/HCatInputFormat (wrong name:
org/apache/hive/hcatalog/mapreduce/HCatInputFormat)"
It confirms that Sqoop uses old namespace, so TIMESTAMP features don’t work.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
