Requirement: We have a client who had Netezza tables and columns in a non-English format. We are trying to create a Oozie workflow to periodically sqoop the DB and then continue in the workflow for some ETL.
Tested: We have successfully called Sqoop from the command line with non-English table names and column names. Our locale is set to KO_KR.UTF-8 in /etc/default/locale.
Problem: After confirming that we can sqoop the db tables as necessary from CLI, we have tried to reimplement as a Oozie workflow. We had lots of problems with using the Sqoop action so we decided to call a command line script from Oozie with the sqoop command. The job fails, and in inspecting the job flow, we seen that Oozie is setting the user.language=en.
We suspect that the problem will be fixed if the user.language=ko is set but we are not sure how to do this.
1. have you confirmed that all hosts in the cluster has set locale to KO_KR.UTF-8 in /etc/default/locale? As Oozie job will be run on random hosts in the cluster, you need to make sure that they are consistent.
2. where did you see Oozie setting "user.language=en"?