We have an SSIS package that loads data from an Oracle database into a SQL Server data warehouse. There are several environments configured for this data warehouse; Development, testing and production. Dev and test share the machine, Prod is autonomous.
When the SSIS package runs on the PROD machine, it loads the Varchar2 columns from our original Oracle database in MSSQL in the DT_WSTR format and saves it in the NVarchar column. I.E. all steps include Unicode support.
When the same package is launched with the same source database in the DEV / Test window, it somehow sees the external columns as Varchar, displays this on DT_STR in the data stream and refuses to store it in the NVarchar column.
All OSs are Win2K8r2, MSSQL 2008 64 bit. The package starts in 32-bit mode, the same behavior is observed when starting from BIDS or from an SQL agent.
Does anyone want to guess why? I have already seen the suggestion to disable the verification of external metadata ( https://stackoverflow.com/a/165608/ ), but this is not a practical suggestion for our situation.
oracle unicode ssis
steenbergh Oct 21 '13 at 12:56 on 2013-10-21 12:56
source share