I want to get RDD from a realtime message source (currently socket) to replace the line of code that is used to read from a local file sc.newAPIHadoopFile()
. In order to maintain the structure of the project, I have to get RDD instead of DStream from that source. I tried to implement a custom receiver, but StreamingContext.receiverStream()
only gives me a DStream. How can I do?
I tried to convert a DStream to RDD with compute
, but that doesn’t make any sense and I failed. So I turned to discover how to directly receive RDD with a receiver.