public Value<MapReduceResult<R>> run(
MapReduceSpecification<I, K, V, O, R> mrSpec, MapReduceSettings settings) {
String mrJobId = getJobKey().getName();
FutureValue<ResultAndCounters<List<AppEngineFile>>> mapResult =
futureCall(new MapJob<I, K, V>(mrJobId, mrSpec, settings), Util.jobSettings(settings));
FutureValue<ShuffleResult<K, V, O>> shuffleResult =
futureCall(
new ShuffleJob<K, V, O>(mrJobId, mrSpec, settings),
mapResult,
Util.jobSettings(settings));
futureCall(
new IntermediateCleanupJob(mrJobId, settings),
mapResult,
Util.jobSettings(settings, waitFor(shuffleResult)));
FutureValue<MapReduceResult<R>> reduceResult =
futureCall(
new ReduceJob<K, V, O, R>(mrJobId, mrSpec, settings),
mapResult,
shuffleResult,
Util.jobSettings(settings));
futureCall(
new FinalCleanupJob<K, V, O>(mrJobId, settings),
shuffleResult,
Util.jobSettings(settings, waitFor(reduceResult)));
return reduceResult;
}
@Override
public Value<Void> run(ResultAndCounters<List<AppEngineFile>> mapResult) {
futureCall(
new DeleteFilesJob("" + this),
immediate(mapResult.getOutputResult()),
Util.jobSettings(settings));
return immediate(null);
}
@Override
public Value<Void> run(ShuffleResult<K, V, O> shuffleResult) {
futureCall(
new DeleteFilesJob("" + this),
immediate(shuffleResult.getReducerInputFiles()),
Util.jobSettings(settings));
return immediate(null);
}
@Override
public Value<ResultAndCounters<List<AppEngineFile>>> run() {
@SuppressWarnings("unchecked")
PromisedValue<ResultAndCounters<List<AppEngineFile>>> result =
(PromisedValue) newPromise(ResultAndCounters.class);
String shardedJobId = mrJobId + "-map";
List<? extends InputReader<I>> readers = Util.createReaders(mrSpec.getInput());
Output<KeyValue<K, V>, List<AppEngineFile>> output =
new IntermediateOutput<K, V>(
mrJobId,
readers.size(),
mrSpec.getIntermediateKeyMarshaller(),
mrSpec.getIntermediateValueMarshaller());
startShardedJob(
mrSpec.getJobName() + " (map phase)",
shardedJobId,
new CountersImpl(),
readers,
output,
Util.createWriters(output),
new TaskCreator<I, KeyValue<K, V>, MapperContext<K, V>>() {
@Override
public WorkerShardTask<I, KeyValue<K, V>, MapperContext<K, V>> createTask(
int shard,
int shardCount,
InputReader<I> reader,
OutputWriter<KeyValue<K, V>> writer) {
return new MapShardTask<I, K, V>(
mrJobId,
shard,
shardCount,
reader,
mrSpec.getMapper(),
writer,
settings.getMillisPerSlice());
}
},
result.getHandle(),
settings);
setStatusConsoleUrl(settings.getBaseUrl() + "detail?mapreduce_id=" + shardedJobId);
return result;
}
/**
* Starts a {@link MapReduceJob} with the given parameters in a new Pipeline. Returns the pipeline
* id.
*/
public static <I, K, V, O, R> String start(
MapReduceSpecification<I, K, V, O, R> specification, MapReduceSettings settings) {
PipelineService pipelineService = PipelineServiceFactory.newPipelineService();
return pipelineService.startNewPipeline(
new MapReduceJob<I, K, V, O, R>(), specification, settings, Util.jobSettings(settings));
}
