use of co.cask.cdap.etl.batch.connector.SingleConnectorFactory in project cdap by caskdata.
the class ETLSpark method initialize.
@Override
@TransactionPolicy(TransactionControl.EXPLICIT)
public void initialize() throws Exception {
final SparkClientContext context = getContext();
cleanupFiles = new ArrayList<>();
List<Finisher> finishers = new ArrayList<>();
SparkConf sparkConf = new SparkConf();
sparkConf.set("spark.driver.extraJavaOptions", "-XX:MaxPermSize=256m");
sparkConf.set("spark.executor.extraJavaOptions", "-XX:MaxPermSize=256m");
sparkConf.set("spark.speculation", "false");
context.setSparkConf(sparkConf);
Map<String, String> properties = context.getSpecification().getProperties();
BatchPhaseSpec phaseSpec = GSON.fromJson(properties.get(Constants.PIPELINEID), BatchPhaseSpec.class);
for (Map.Entry<String, String> pipelineProperty : phaseSpec.getPipelineProperties().entrySet()) {
sparkConf.set(pipelineProperty.getKey(), pipelineProperty.getValue());
}
MacroEvaluator evaluator = new DefaultMacroEvaluator(new BasicArguments(context), context.getLogicalStartTime(), context, context.getNamespace());
final SparkBatchSourceFactory sourceFactory = new SparkBatchSourceFactory();
final SparkBatchSinkFactory sinkFactory = new SparkBatchSinkFactory();
final Map<String, Integer> stagePartitions = new HashMap<>();
PluginContext pluginContext = new SparkPipelinePluginContext(context, context.getMetrics(), phaseSpec.isStageLoggingEnabled(), phaseSpec.isProcessTimingEnabled());
PipelinePluginInstantiator pluginInstantiator = new PipelinePluginInstantiator(pluginContext, context.getMetrics(), phaseSpec, new SingleConnectorFactory());
final PipelineRuntime pipelineRuntime = new PipelineRuntime(context);
final Admin admin = context.getAdmin();
PipelinePhase phase = phaseSpec.getPhase();
// go through in topological order so that arguments set by one stage are seen by stages after it
for (final String stageName : phase.getDag().getTopologicalOrder()) {
final StageSpec stageSpec = phase.getStage(stageName);
String pluginType = stageSpec.getPluginType();
boolean isConnectorSource = Constants.Connector.PLUGIN_TYPE.equals(pluginType) && phase.getSources().contains(stageName);
boolean isConnectorSink = Constants.Connector.PLUGIN_TYPE.equals(pluginType) && phase.getSinks().contains(stageName);
SubmitterPlugin submitterPlugin = null;
if (BatchSource.PLUGIN_TYPE.equals(pluginType) || isConnectorSource) {
BatchConfigurable<BatchSourceContext> batchSource = pluginInstantiator.newPluginInstance(stageName, evaluator);
ContextProvider<BatchSourceContext> contextProvider = new ContextProvider<BatchSourceContext>() {
@Override
public BatchSourceContext getContext(DatasetContext datasetContext) {
return new SparkBatchSourceContext(sourceFactory, context, pipelineRuntime, datasetContext, stageSpec);
}
};
submitterPlugin = new SubmitterPlugin(stageName, context, batchSource, contextProvider);
} else if (Transform.PLUGIN_TYPE.equals(pluginType)) {
Transform transform = pluginInstantiator.newPluginInstance(stageName, evaluator);
ContextProvider<StageSubmitterContext> contextProvider = new ContextProvider<StageSubmitterContext>() {
@Override
public StageSubmitterContext getContext(DatasetContext datasetContext) {
return new SparkBatchSourceContext(sourceFactory, context, pipelineRuntime, datasetContext, stageSpec);
}
};
submitterPlugin = new SubmitterPlugin(stageName, context, transform, contextProvider);
} else if (BatchSink.PLUGIN_TYPE.equals(pluginType) || isConnectorSink) {
BatchConfigurable<BatchSinkContext> batchSink = pluginInstantiator.newPluginInstance(stageName, evaluator);
ContextProvider<BatchSinkContext> contextProvider = new ContextProvider<BatchSinkContext>() {
@Override
public BatchSinkContext getContext(DatasetContext datasetContext) {
return new SparkBatchSinkContext(sinkFactory, context, pipelineRuntime, datasetContext, stageSpec);
}
};
submitterPlugin = new SubmitterPlugin(stageName, context, batchSink, contextProvider);
} else if (SparkSink.PLUGIN_TYPE.equals(pluginType)) {
BatchConfigurable<SparkPluginContext> sparkSink = pluginInstantiator.newPluginInstance(stageName, evaluator);
ContextProvider<SparkPluginContext> contextProvider = new ContextProvider<SparkPluginContext>() {
@Override
public SparkPluginContext getContext(DatasetContext datasetContext) {
return new BasicSparkPluginContext(context, pipelineRuntime, stageSpec, datasetContext, admin);
}
};
submitterPlugin = new SubmitterPlugin(stageName, context, sparkSink, contextProvider);
} else if (BatchAggregator.PLUGIN_TYPE.equals(pluginType)) {
BatchAggregator aggregator = pluginInstantiator.newPluginInstance(stageName, evaluator);
ContextProvider<DefaultAggregatorContext> contextProvider = new AggregatorContextProvider(pipelineRuntime, stageSpec, admin);
submitterPlugin = new SubmitterPlugin(stageName, context, aggregator, contextProvider);
} else if (BatchJoiner.PLUGIN_TYPE.equals(pluginType)) {
BatchJoiner joiner = pluginInstantiator.newPluginInstance(stageName, evaluator);
ContextProvider<DefaultJoinerContext> contextProvider = new JoinerContextProvider(pipelineRuntime, stageSpec, admin);
submitterPlugin = new SubmitterPlugin<>(stageName, context, joiner, contextProvider, new SubmitterPlugin.PrepareAction<DefaultJoinerContext>() {
@Override
public void act(DefaultJoinerContext sparkJoinerContext) {
stagePartitions.put(stageName, sparkJoinerContext.getNumPartitions());
}
});
}
if (submitterPlugin != null) {
submitterPlugin.prepareRun();
finishers.add(submitterPlugin);
}
}
File configFile = File.createTempFile("HydratorSpark", ".config");
cleanupFiles.add(configFile);
try (Writer writer = Files.newBufferedWriter(configFile.toPath(), StandardCharsets.UTF_8)) {
SparkBatchSourceSinkFactoryInfo sourceSinkInfo = new SparkBatchSourceSinkFactoryInfo(sourceFactory, sinkFactory, stagePartitions);
writer.write(GSON.toJson(sourceSinkInfo));
}
finisher = new CompositeFinisher(finishers);
context.localize("HydratorSpark.config", configFile.toURI());
WorkflowToken token = context.getWorkflowToken();
if (token != null) {
for (Map.Entry<String, String> entry : pipelineRuntime.getArguments().getAddedArguments().entrySet()) {
token.put(entry.getKey(), entry.getValue());
}
}
}
use of co.cask.cdap.etl.batch.connector.SingleConnectorFactory in project cdap by caskdata.
the class BatchSparkPipelineDriver method run.
@Override
public void run(DatasetContext context) throws Exception {
BatchPhaseSpec phaseSpec = GSON.fromJson(sec.getSpecification().getProperty(Constants.PIPELINEID), BatchPhaseSpec.class);
Path configFile = sec.getLocalizationContext().getLocalFile("HydratorSpark.config").toPath();
try (BufferedReader reader = Files.newBufferedReader(configFile, StandardCharsets.UTF_8)) {
String object = reader.readLine();
SparkBatchSourceSinkFactoryInfo sourceSinkInfo = GSON.fromJson(object, SparkBatchSourceSinkFactoryInfo.class);
sourceFactory = sourceSinkInfo.getSparkBatchSourceFactory();
sinkFactory = sourceSinkInfo.getSparkBatchSinkFactory();
stagePartitions = sourceSinkInfo.getStagePartitions();
}
datasetContext = context;
numOfRecordsPreview = phaseSpec.getNumOfRecordsPreview();
PipelinePluginContext pluginContext = new PipelinePluginContext(sec.getPluginContext(), sec.getMetrics(), phaseSpec.isStageLoggingEnabled(), phaseSpec.isProcessTimingEnabled());
Map<String, StageStatisticsCollector> collectors = new HashMap<>();
if (phaseSpec.pipelineContainsCondition()) {
Iterator<StageSpec> iterator = phaseSpec.getPhase().iterator();
while (iterator.hasNext()) {
StageSpec spec = iterator.next();
collectors.put(spec.getName(), new SparkStageStatisticsCollector(jsc));
}
}
try {
PipelinePluginInstantiator pluginInstantiator = new PipelinePluginInstantiator(pluginContext, sec.getMetrics(), phaseSpec, new SingleConnectorFactory());
runPipeline(phaseSpec.getPhase(), BatchSource.PLUGIN_TYPE, sec, stagePartitions, pluginInstantiator, collectors);
} finally {
updateWorkflowToken(sec.getWorkflowToken(), collectors);
}
}
Aggregations