private static JobTaskVertex createReducer(
JobGraph jobGraph,
int numSubTasks,
TypeSerializerFactory<?> inputSerializer,
TypeComparatorFactory<?> inputComparator,
TypeSerializerFactory<?> outputSerializer) {
// ---------------- the tail (co group) --------------------
JobTaskVertex tail =
JobGraphUtils.createTask(
IterationTailPactTask.class,
"Reduce / Iteration Tail",
jobGraph,
numSubTasks,
numSubTasks);
TaskConfig tailConfig = new TaskConfig(tail.getConfiguration());
tailConfig.setIterationId(ITERATION_ID);
tailConfig.setIsWorksetUpdate();
// inputs and driver
tailConfig.setDriver(ReduceDriver.class);
tailConfig.setDriverStrategy(DriverStrategy.SORTED_GROUP);
tailConfig.addInputToGroup(0);
tailConfig.setInputSerializer(inputSerializer, 0);
tailConfig.setDriverComparator(inputComparator, 0);
tailConfig.setInputLocalStrategy(0, LocalStrategy.SORT);
tailConfig.setInputComparator(inputComparator, 0);
tailConfig.setMemoryInput(0, MEMORY_PER_CONSUMER * JobGraphUtils.MEGABYTE);
tailConfig.setFilehandlesInput(0, 128);
tailConfig.setSpillingThresholdInput(0, 0.9f);
// output
tailConfig.addOutputShipStrategy(ShipStrategyType.FORWARD);
tailConfig.setOutputSerializer(outputSerializer);
// the udf
tailConfig.setStubWrapper(
new UserCodeObjectWrapper<RecomputeClusterCenter>(new RecomputeClusterCenter()));
return tail;
}
