sqoop export from hive table created by sqoop import

Question

I've successfully imported a simple 2 column SQL Server data into HIVE using SQOOP on Cloudera CDH4.1. - id (int) - description (varchar(50))

When I try get the same HIVE table to go back to SQL Server it lumps the two columns together and therefore creates an NumberFormatException.

I've tried the --fields-terminated by \t --lines-terminated-by \n and this doesn't make any difference.

I've also tried to export the sample_08 Hive table as supplied by Cloudera but get a NoSuchElement exception. Wish it would give a clue as to which element it thinks doesn't exist!

I've ensured that the table structures and field names are both the same on both sides including the case and underscores.

I can export a single column table but am absolutely stumped as to how to resolve the issue.

Any help would be gratefully received.

I found that the Microsoft install of their SQOOP connector does not set up the configuration file in managers.d correctly. /usr/lib/sqoop/conf/managers.d/mssqoop-sqlserver should contain the path of the sqljdbc4.jar file but by default it omits it. The file should contain the following:- com.microsoft.sqoop.MSSQLServerManagerFactory/usr/lib/sqoop/lib/sqljdbc4.jar This simply gets me to the next error which is java.lang.NoSuchMethodError — Dave Poole

Jarek Jarcec Cecho Jarek Jarcec Cecho · Accepted Answer · 2012-11-14T22:08:21

Parameters --filds-terminated-by and --lines-terminated-by are used only during "import" operation. You need to use parameters --input-fields-terminated-by and --input-lines-terminated-by to do the same in "export" case.

Would you mind sharing entire exception stack trace and possibly example set of your data that trigger your issue?

sqoop export from hive table created by sqoop import

1 Answers