Hi all, As part of the concurrent random walks that I'm implementing (see: https://issues.apache.org/jira/browse/GIRAPH-824 ), I am using the InternalVertexRunner to locally execute / test my code. Invoking:
InternalVertexRunner.run(conf, graph) always returns null. After much digging / custom printing of error messages, I got the following error message: run: Caught an unrecoverable exception waitFor: ExecutionException occurred while waiting for org.apache.giraph.utils.ProgressableUtils$FutureWaitable@1837acad which didn't help too much either (at least now I knew something was mistakenly configured). Therefore, I ran my example on an actual (distributed) GIRAPH instance, which gave back the following exception: Exception in thread "main" java.lang.IllegalArgumentException: checkClassTypes: vertex value types not assignable, computation - class org.apache.giraph.io.DoubleArrayListWritable, VertexInputFormat - class org.apache.hadoop.io.DoubleWritable which much better describes what the problem is. Note that this is thrown when the graph is being read, and is not related with any computation / algorithmic part of the code. Therefore, I *believe* it would be fairly easy to capture it with the InternalVertexRunner as well. Given that, I would like to ask that InternalVertexRunner prints more meaningful results than the generic one I got in the first place. Having to deploy and run the code in an actual GIRAPH installation to get the actual exceptions, negates the whole idea of such a proxy. I would be happy to help in this direction, but would need some guidance, considering that I've not followed GIRAPH's build-up from the start. I can open the JIRA ticket, but wanted to get some feedback first (perhaps it's minor and easily fixable). Best, George
