Hi Sudar,
This is due to how Sqoop splits data. It inserts limit offset into your query in order to split the db query across multiple mappers. This is not good, as limit offset is a bad access pattern, and it also uses non-standard SQL.
In CDAP 3.3, you can use the Teradata source instead of the Database source, which works in a very similar way, except you need to specify a field to split on, a bounding query to determine the min and max values for that field, and your import query needs to include 'where $CONDITIONS' in it. The Sqoop input format will replace '$CONDITIONS' with a value range on the split field.
Note that in CDAP 3.4, we've removed the 'Teradata' plugin and moved its behavior to the Database plugin because of these issues.
Regards,
Albert