I am trying to rewritten the map matching algorithm graphhopper into spark. I want to broadcast the 'map infomation' to each worker and distribute the traces to each worker.
However, the class MapMatching
(which stores the map infomation) is not is not Serializable
. I have tried using kryo
to serialize it but it does not work.
com.esotericsoftware.kryo.KryoException: Error constructing instance of class: > com.graphhopper.matching.MapMatchin
Can I instantiate the MapMatching
after broadcasting? Or how can I make it serializable? Or are there any solutions?