Currently we are allowing users to change data type of a field through COM if there is no data in the object. But this change is happening only in Mongo and not in the redshift table. Because of this, there are lot of errors through out S3 and data load API connectors. I believe that this should either be disabled or handled properly. Once the redshift table is created, we should disable data type conversion even if there is no data. Or when the data type conversion is made through COM, we should automatically create a new column with same name and the required data type, and then delete the older field. Or we should have a flag that shows if there is any mismatch between data types between Mongo and Redshift tables.