So if I am using GraphX on Spark and I created a graph, which gets called a lot later, do I want to cache graph? Or do I want to cache the vertices and edges (actual data) that I use to create the graph?
e.g. val graph = Graph(vertices, edges) graph.blahblahblah graph.blahblahblah graph.blahblahblah FYI, I wanted to measured the time it takes to run my algorithm, so once I create the graph, I force it Spark to read the data in by calling graph.vertices.count and graph.edges.count since it does the lazy evalutation. Then I run the actual algorithm with time measure on. But basically it doesn't even get to the algorithm portion because it breaks at graph.edges.count when it reads the data... -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/ERROR-TaskSchedulerImpl-Lost-an-executor-tp4566p4701.html Sent from the Apache Spark User List mailing list archive at Nabble.com.
