You'd usually not need two pointing to the same HDFS cluster. I'd use it for Avro Sinks for agent to agent communication to make sure your pipeline does not get clogged due to a single agent failing. You could failover from one HDFS cluster to another if you want.

Gary Malouf wrote:

I have a tough time seeing why one would need a failover sink where
both write to the same HDFS cluster. Can one describe to me why this
is a good idea?

Reply via email to