Apache Hadoop user group information employs the JDK security framework which uses a static "Subject" instance across the application scope. There is no way to change it in runtime. Due to this restriction, and before the Hive SQL query execution in embedded mode, the user must be correctly specified.
The Hive SQL query execution will obtain a user name in this order from Talend Studio:
- Find the user name from the system property of HADOOP_USER_NAME.
- If the user name is not found, then find it from the system property of HADOOP_PROXY_USER.
- If the user name is not found, then find it from OS environment variables.
- If the user name is not found, then use the OS login user name.