public static String runTask(String[] args) throws Exception {
final String schema = args[0];
String version = args[1];
final HadoopIngestionSpec theSchema =
HadoopDruidIndexerConfig.jsonMapper.readValue(schema, HadoopIngestionSpec.class);
final HadoopDruidIndexerConfig config =
HadoopDruidIndexerConfig.fromSchema(
theSchema.withTuningConfig(theSchema.getTuningConfig().withVersion(version)));
HadoopDruidIndexerJob job = new HadoopDruidIndexerJob(config);
log.info("Starting a hadoop index generator job...");
if (job.run()) {
return HadoopDruidIndexerConfig.jsonMapper.writeValueAsString(job.getPublishedSegments());
}
return null;
}
@Override
public boolean isReady(TaskActionClient taskActionClient) throws Exception {
Optional<SortedSet<Interval>> intervals =
spec.getDataSchema().getGranularitySpec().bucketIntervals();
if (intervals.isPresent()) {
Interval interval = JodaUtils.umbrellaInterval(JodaUtils.condenseIntervals(intervals.get()));
return taskActionClient.submit(new LockTryAcquireAction(interval)).isPresent();
} else {
return true;
}
}
public static String runTask(String[] args) throws Exception {
final String schema = args[0];
final String workingPath = args[1];
final String segmentOutputPath = args[2];
final HadoopIngestionSpec theSchema =
HadoopDruidIndexerConfig.jsonMapper.readValue(schema, HadoopIngestionSpec.class);
final HadoopDruidIndexerConfig config =
HadoopDruidIndexerConfig.fromSchema(
theSchema
.withIOConfig(theSchema.getIOConfig().withSegmentOutputPath(segmentOutputPath))
.withTuningConfig(theSchema.getTuningConfig().withWorkingPath(workingPath)));
Jobby job = new HadoopDruidDetermineConfigurationJob(config);
log.info("Starting a hadoop determine configuration job...");
if (job.run()) {
return HadoopDruidIndexerConfig.jsonMapper.writeValueAsString(config.getSchema());
}
return null;
}
private static String getTheDataSource(HadoopIngestionSpec spec, HadoopIngestionSpec config) {
if (spec != null) {
return spec.getDataSchema().getDataSource();
}
return config.getDataSchema().getDataSource();
}
@SuppressWarnings("unchecked")
@Override
public TaskStatus run(TaskToolbox toolbox) throws Exception {
final List<String> finalHadoopDependencyCoordinates =
hadoopDependencyCoordinates != null
? hadoopDependencyCoordinates
: toolbox.getConfig().getDefaultHadoopCoordinates();
final DefaultTeslaAether aetherClient = Initialization.getAetherClient(extensionsConfig);
final List<URL> extensionURLs = Lists.newArrayList();
for (String coordinate : extensionsConfig.getCoordinates()) {
final ClassLoader coordinateLoader =
Initialization.getClassLoaderForCoordinates(aetherClient, coordinate);
extensionURLs.addAll(Arrays.asList(((URLClassLoader) coordinateLoader).getURLs()));
}
final List<URL> nonHadoopURLs = Lists.newArrayList();
nonHadoopURLs.addAll(
Arrays.asList(((URLClassLoader) HadoopIndexTask.class.getClassLoader()).getURLs()));
final List<URL> driverURLs = Lists.newArrayList();
driverURLs.addAll(nonHadoopURLs);
// put hadoop dependencies last to avoid jets3t & apache.httpcore version conflicts
for (String hadoopDependencyCoordinate : finalHadoopDependencyCoordinates) {
final ClassLoader hadoopLoader =
Initialization.getClassLoaderForCoordinates(aetherClient, hadoopDependencyCoordinate);
driverURLs.addAll(Arrays.asList(((URLClassLoader) hadoopLoader).getURLs()));
}
final URLClassLoader loader =
new URLClassLoader(driverURLs.toArray(new URL[driverURLs.size()]), null);
Thread.currentThread().setContextClassLoader(loader);
final List<URL> jobUrls = Lists.newArrayList();
jobUrls.addAll(nonHadoopURLs);
jobUrls.addAll(extensionURLs);
System.setProperty(
"druid.hadoop.internal.classpath", Joiner.on(File.pathSeparator).join(jobUrls));
boolean determineIntervals =
!spec.getDataSchema().getGranularitySpec().bucketIntervals().isPresent();
final Class<?> determineConfigurationMainClass =
loader.loadClass(HadoopDetermineConfigInnerProcessing.class.getName());
final Method determineConfigurationMainMethod =
determineConfigurationMainClass.getMethod("runTask", String[].class);
String[] determineConfigArgs =
new String[] {
toolbox.getObjectMapper().writeValueAsString(spec),
toolbox.getConfig().getHadoopWorkingPath(),
toolbox.getSegmentPusher().getPathForHadoop(getDataSource())
};
String config =
(String) determineConfigurationMainMethod.invoke(null, new Object[] {determineConfigArgs});
HadoopIngestionSpec indexerSchema =
toolbox.getObjectMapper().readValue(config, HadoopIngestionSpec.class);
// We should have a lock from before we started running only if interval was specified
final String version;
if (determineIntervals) {
Interval interval =
JodaUtils.umbrellaInterval(
JodaUtils.condenseIntervals(
indexerSchema.getDataSchema().getGranularitySpec().bucketIntervals().get()));
TaskLock lock = toolbox.getTaskActionClient().submit(new LockAcquireAction(interval));
version = lock.getVersion();
} else {
Iterable<TaskLock> locks = getTaskLocks(toolbox);
final TaskLock myLock = Iterables.getOnlyElement(locks);
version = myLock.getVersion();
}
log.info("Setting version to: %s", version);
final Class<?> indexGeneratorMainClass =
loader.loadClass(HadoopIndexGeneratorInnerProcessing.class.getName());
final Method indexGeneratorMainMethod =
indexGeneratorMainClass.getMethod("runTask", String[].class);
String[] indexGeneratorArgs =
new String[] {toolbox.getObjectMapper().writeValueAsString(indexerSchema), version};
String segments =
(String) indexGeneratorMainMethod.invoke(null, new Object[] {indexGeneratorArgs});
if (segments != null) {
List<DataSegment> publishedSegments =
toolbox.getObjectMapper().readValue(segments, new TypeReference<List<DataSegment>>() {});
toolbox.pushSegments(publishedSegments);
return TaskStatus.success(getId());
} else {
return TaskStatus.failure(getId());
}
}