Created 01-04-2017 02:09 PM
My understanding is that Sqoop uses Hiverser(CLI), which does not support authorization. Is there anyway to configure Sqoop to use HiverServer2, so it can use Hive Authorization?
Any help would be much appreciated!
Created 01-04-2017 06:05 PM
Hive authentication using kerberos or simple auth is supported in sqoop2 starting at version 1.99.4:
https://cwiki.apache.org/confluence/display/SQOOP/Security+Guide+On+Sqoop+2
https://issues.apache.org/jira/browse/SQOOP-1525
Nifi can import using hive with authentication using the PutHiveQL or PutHiveStreaming processors:
PutHiveQL specifies the kerberos principals and key tab on the HiveConnectionPool.
PutHiveStreaming sets the security options in the hive-site.xml config file.
Documentation for the Nifi Processors are at:
Created 01-05-2017 02:23 PM
@cduby Thanks! Would I be able to use Ranger for authorization?
Created 01-05-2017 03:36 PM
@Raffi Abberbock Yes, Nifi will use the setup of the hive server. Sqoop2 is not in the HDP distribution yet.
Created 01-16-2017 09:46 PM
This is a great question. As @cduby pointed out its not available until sqoop 2. But, if you have to load data into tables using some sort of authorization, you could directly load the data in hdfs, then use beeline to move data into tables with authorization, then you can apply Ranger policies per user and other features. Its a bit of a long way to achieve the results but will get you closer to audit, authorization requirements.