use of co.cask.cdap.etl.common.DefaultStageMetrics in project cdap by caskdata.
the class RDDCollection method publishAlerts.
@Override
public void publishAlerts(StageSpec stageSpec, StageStatisticsCollector collector) throws Exception {
PluginFunctionContext pluginFunctionContext = new PluginFunctionContext(stageSpec, sec, collector);
AlertPublisher alertPublisher = pluginFunctionContext.createPlugin();
PipelineRuntime pipelineRuntime = new SparkPipelineRuntime(sec);
AlertPublisherContext alertPublisherContext = new DefaultAlertPublisherContext(pipelineRuntime, stageSpec, sec.getMessagingContext(), sec.getAdmin());
alertPublisher.initialize(alertPublisherContext);
StageMetrics stageMetrics = new DefaultStageMetrics(sec.getMetrics(), stageSpec.getName());
TrackedIterator<Alert> trackedAlerts = new TrackedIterator<>(((JavaRDD<Alert>) rdd).collect().iterator(), stageMetrics, Constants.Metrics.RECORDS_IN);
alertPublisher.publish(trackedAlerts);
alertPublisher.destroy();
}
use of co.cask.cdap.etl.common.DefaultStageMetrics in project cdap by caskdata.
the class StreamingAlertPublishFunction method call.
@Override
public Void call(JavaRDD<Alert> data, Time batchTime) throws Exception {
MacroEvaluator evaluator = new DefaultMacroEvaluator(new BasicArguments(sec), batchTime.milliseconds(), sec.getSecureStore(), sec.getNamespace());
PluginContext pluginContext = new SparkPipelinePluginContext(sec.getPluginContext(), sec.getMetrics(), stageSpec.isStageLoggingEnabled(), stageSpec.isProcessTimingEnabled());
String stageName = stageSpec.getName();
AlertPublisher alertPublisher = pluginContext.newPluginInstance(stageName, evaluator);
PipelineRuntime pipelineRuntime = new SparkPipelineRuntime(sec, batchTime.milliseconds());
AlertPublisherContext alertPublisherContext = new DefaultAlertPublisherContext(pipelineRuntime, stageSpec, sec.getMessagingContext(), sec.getAdmin());
alertPublisher.initialize(alertPublisherContext);
StageMetrics stageMetrics = new DefaultStageMetrics(sec.getMetrics(), stageName);
TrackedIterator<Alert> trackedAlerts = new TrackedIterator<>(data.collect().iterator(), stageMetrics, Constants.Metrics.RECORDS_IN);
alertPublisher.publish(trackedAlerts);
alertPublisher.destroy();
return null;
}
use of co.cask.cdap.etl.common.DefaultStageMetrics in project cdap by caskdata.
the class PipelinePluginContext method wrapPlugin.
private Object wrapPlugin(String pluginId, Object plugin) {
Caller caller = getCaller(pluginId);
StageMetrics stageMetrics = new DefaultStageMetrics(metrics, pluginId);
OperationTimer operationTimer = processTimingEnabled ? new MetricsOperationTimer(stageMetrics) : NoOpOperationTimer.INSTANCE;
if (plugin instanceof Action) {
return new WrappedAction((Action) plugin, caller);
} else if (plugin instanceof BatchSource) {
return new WrappedBatchSource<>((BatchSource) plugin, caller, operationTimer);
} else if (plugin instanceof BatchSink) {
return new WrappedBatchSink<>((BatchSink) plugin, caller, operationTimer);
} else if (plugin instanceof ErrorTransform) {
return new WrappedErrorTransform<>((ErrorTransform) plugin, caller, operationTimer);
} else if (plugin instanceof Transform) {
return new WrappedTransform<>((Transform) plugin, caller, operationTimer);
} else if (plugin instanceof BatchAggregator) {
return new WrappedBatchAggregator<>((BatchAggregator) plugin, caller, operationTimer);
} else if (plugin instanceof BatchJoiner) {
return new WrappedBatchJoiner<>((BatchJoiner) plugin, caller, operationTimer);
} else if (plugin instanceof PostAction) {
return new WrappedPostAction((PostAction) plugin, caller);
} else if (plugin instanceof SplitterTransform) {
return new WrappedSplitterTransform<>((SplitterTransform) plugin, caller, operationTimer);
}
return wrapUnknownPlugin(pluginId, plugin, caller);
}
use of co.cask.cdap.etl.common.DefaultStageMetrics in project cdap by caskdata.
the class SmartWorkflow method destroy.
@Override
public void destroy() {
WorkflowContext workflowContext = getContext();
PipelineRuntime pipelineRuntime = new PipelineRuntime(workflowContext, workflowMetrics);
// Execute the post actions only if pipeline is not running in preview mode.
if (!workflowContext.getDataTracer(PostAction.PLUGIN_TYPE).isEnabled()) {
for (Map.Entry<String, PostAction> endingActionEntry : postActions.entrySet()) {
String name = endingActionEntry.getKey();
PostAction action = endingActionEntry.getValue();
StageSpec stageSpec = stageSpecs.get(name);
BatchActionContext context = new WorkflowBackedActionContext(workflowContext, pipelineRuntime, stageSpec);
try {
action.run(context);
} catch (Throwable t) {
LOG.error("Error while running post action {}.", name, t);
}
}
}
// publish all alerts
for (Map.Entry<String, AlertPublisher> alertPublisherEntry : alertPublishers.entrySet()) {
String name = alertPublisherEntry.getKey();
AlertPublisher alertPublisher = alertPublisherEntry.getValue();
PartitionedFileSet alertConnector = workflowContext.getDataset(name);
try (CloseableIterator<Alert> alerts = new AlertReader(alertConnector.getPartitions(PartitionFilter.ALWAYS_MATCH))) {
if (!alerts.hasNext()) {
continue;
}
StageMetrics stageMetrics = new DefaultStageMetrics(workflowMetrics, name);
StageSpec stageSpec = stageSpecs.get(name);
AlertPublisherContext alertContext = new DefaultAlertPublisherContext(pipelineRuntime, stageSpec, workflowContext, workflowContext.getAdmin());
alertPublisher.initialize(alertContext);
TrackedIterator<Alert> trackedIterator = new TrackedIterator<>(alerts, stageMetrics, Constants.Metrics.RECORDS_IN);
alertPublisher.publish(trackedIterator);
} catch (Exception e) {
LOG.warn("Stage {} had errors publishing alerts. Alerts may not have been published.", name, e);
} finally {
try {
alertPublisher.destroy();
} catch (Exception e) {
LOG.warn("Error destroying alert publisher for stage {}", name, e);
}
}
}
ProgramStatus status = getContext().getState().getStatus();
if (status == ProgramStatus.FAILED) {
WRAPPERLOGGER.error("Pipeline '{}' failed.", getContext().getApplicationSpecification().getName());
} else {
WRAPPERLOGGER.info("Pipeline '{}' {}.", getContext().getApplicationSpecification().getName(), status == ProgramStatus.COMPLETED ? "succeeded" : status.name().toLowerCase());
}
MacroEvaluator macroEvaluator = new DefaultMacroEvaluator(pipelineRuntime.getArguments(), workflowContext.getLogicalStartTime(), workflowContext, workflowContext.getNamespace());
// Get resolved plugin properties
Map<String, Map<String, String>> resolvedProperties = new HashMap<>();
for (StageSpec spec : stageSpecs.values()) {
String stageName = spec.getName();
resolvedProperties.put(stageName, workflowContext.getPluginProperties(stageName, macroEvaluator).getProperties());
}
// Add resolved plugin properties to workflow token as a JSON String
workflowContext.getToken().put(RESOLVED_PLUGIN_PROPERTIES_MAP, GSON.toJson(resolvedProperties));
}
use of co.cask.cdap.etl.common.DefaultStageMetrics in project cdap by caskdata.
the class MapReduceTransformExecutorFactory method getTransformation.
@SuppressWarnings("unchecked")
private <IN, OUT> TrackedTransform<IN, OUT> getTransformation(StageSpec stageSpec) throws Exception {
DefaultMacroEvaluator macroEvaluator = new DefaultMacroEvaluator(arguments, taskContext.getLogicalStartTime(), taskContext, taskContext.getNamespace());
String stageName = stageSpec.getName();
String pluginType = stageSpec.getPluginType();
StageMetrics stageMetrics = new DefaultStageMetrics(metrics, stageName);
TaskAttemptContext taskAttemptContext = (TaskAttemptContext) taskContext.getHadoopContext();
StageStatisticsCollector collector = isPipelineContainsCondition ? new MapReduceStageStatisticsCollector(stageName, taskAttemptContext) : new NoopStageStatisticsCollector();
if (BatchAggregator.PLUGIN_TYPE.equals(pluginType)) {
BatchAggregator<?, ?, ?> batchAggregator = pluginInstantiator.newPluginInstance(stageName, macroEvaluator);
BatchRuntimeContext runtimeContext = createRuntimeContext(stageSpec);
batchAggregator.initialize(runtimeContext);
if (isMapPhase) {
return getTrackedEmitKeyStep(new MapperAggregatorTransformation(batchAggregator, mapOutputKeyClassName, mapOutputValClassName), stageMetrics, taskContext.getDataTracer(stageName), collector);
} else {
return getTrackedAggregateStep(new ReducerAggregatorTransformation(batchAggregator, mapOutputKeyClassName, mapOutputValClassName), stageMetrics, taskContext.getDataTracer(stageName), collector);
}
} else if (BatchJoiner.PLUGIN_TYPE.equals(pluginType)) {
BatchJoiner<?, ?, ?> batchJoiner = pluginInstantiator.newPluginInstance(stageName, macroEvaluator);
BatchJoinerRuntimeContext runtimeContext = createRuntimeContext(stageSpec);
batchJoiner.initialize(runtimeContext);
if (isMapPhase) {
return getTrackedEmitKeyStep(new MapperJoinerTransformation(batchJoiner, mapOutputKeyClassName, mapOutputValClassName), stageMetrics, taskContext.getDataTracer(stageName), collector);
} else {
return getTrackedMergeStep(new ReducerJoinerTransformation(batchJoiner, mapOutputKeyClassName, mapOutputValClassName, runtimeContext.getInputSchemas().size()), stageMetrics, taskContext.getDataTracer(stageName), collector);
}
}
Transformation transformation = getInitializedTransformation(stageSpec);
boolean isLimitingSource = taskContext.getDataTracer(stageName).isEnabled() && BatchSource.PLUGIN_TYPE.equals(pluginType) && isMapPhase;
transformation = isLimitingSource ? new LimitingTransform(transformation, numberOfRecordsPreview) : transformation;
// we emit metrics for records into alert publishers when the actual alerts are published,
// not when we write the alerts to the temporary dataset
String recordsInMetric = AlertPublisher.PLUGIN_TYPE.equals(pluginType) ? null : Constants.Metrics.RECORDS_IN;
return new TrackedTransform<>(transformation, stageMetrics, recordsInMetric, Constants.Metrics.RECORDS_OUT, taskContext.getDataTracer(stageName), collector);
}
Aggregations