Runtime implementation of Kerberos spark context

I have a Spark application that performs various tasks for different users simultaneously, through several Spark sessions for several threads.

My client would like to strengthen his cluster. I wonder if there is a way to configure impersonation, for example, each of these tasks will work with different proxy users. From what I see in spark conf and code, this cannot be done at run time for a specific context, but I am not familiar with Kerberos and this part of Spark.

Can anyone confirm / assure this?

+4
source share

Source: https://habr.com/ru/post/1676279/


All Articles