问题
I looked at the options provided by sqoop export operation but could not find any options to handle bad records. For example once in a while it is possible that a character is present where a number is expected in a huge set of records. Is there a way to handle these scenarios in sqoop without failing the job and providing the bad records in a file.
回答1:
Sqoop currently expects that the data to export is clean and do not provide facilities to handle corrupted data. You can use MR/Pig/Hive job to clean your data up, prior using Sqoop to export them.
来源：https://stackoverflow.com/questions/18446911/handling-bad-records-during-sqoop-import-or-export