@Override
public List<WorkUnit> getWorkunits(SourceState state) {
initLogger(state);
List<WorkUnit> workUnits = Lists.newArrayList();
String nameSpaceName = state.getProp(ConfigurationKeys.EXTRACT_NAMESPACE_NAME_KEY);
String entityName = state.getProp(ConfigurationKeys.SOURCE_ENTITY);
String extractTableName = state.getProp(ConfigurationKeys.EXTRACT_TABLE_NAME_KEY);
// If extract table name is not found then use the entity name
if (StringUtils.isBlank(extractTableName)) {
extractTableName =
Utils.escapeSpecialCharacters(
entityName, ConfigurationKeys.ESCAPE_CHARS_IN_TABLE_NAME, "_");
}
TableType tableType =
TableType.valueOf(state.getProp(ConfigurationKeys.EXTRACT_TABLE_TYPE_KEY).toUpperCase());
long previousWatermark = this.getLatestWatermarkFromMetadata(state);
Map<Long, Long> sortedPartitions = Maps.newTreeMap();
sortedPartitions.putAll(new Partitioner(state).getPartitions(previousWatermark));
// Use extract table name to create extract
SourceState partitionState = new SourceState();
partitionState.addAll(state);
Extract extract = partitionState.createExtract(tableType, nameSpaceName, extractTableName);
// Setting current time for the full extract
if (Boolean.valueOf(state.getProp(ConfigurationKeys.EXTRACT_IS_FULL_KEY))) {
extract.setFullTrue(System.currentTimeMillis());
}
for (Entry<Long, Long> entry : sortedPartitions.entrySet()) {
partitionState.setProp(ConfigurationKeys.WORK_UNIT_LOW_WATER_MARK_KEY, entry.getKey());
partitionState.setProp(ConfigurationKeys.WORK_UNIT_HIGH_WATER_MARK_KEY, entry.getValue());
workUnits.add(partitionState.createWorkUnit(extract));
}
LOG.info("Total number of work units for the current run: " + workUnits.size());
List<WorkUnit> previousWorkUnits = this.getPreviousWorkUnitsForRetry(state);
LOG.info("Total number of incomplete tasks from the previous run: " + previousWorkUnits.size());
workUnits.addAll(previousWorkUnits);
return workUnits;
}
/**
* Convert this {@link TaskState} instance to a {@link TaskExecutionInfo} instance.
*
* @return a {@link TaskExecutionInfo} instance
*/
public TaskExecutionInfo toTaskExecutionInfo() {
TaskExecutionInfo taskExecutionInfo = new TaskExecutionInfo();
taskExecutionInfo.setJobId(this.jobId);
taskExecutionInfo.setTaskId(this.taskId);
if (this.startTime > 0) {
taskExecutionInfo.setStartTime(this.startTime);
}
if (this.endTime > 0) {
taskExecutionInfo.setEndTime(this.endTime);
}
taskExecutionInfo.setDuration(this.duration);
taskExecutionInfo.setState(TaskStateEnum.valueOf(getWorkingState().name()));
if (this.contains(ConfigurationKeys.TASK_FAILURE_EXCEPTION_KEY)) {
taskExecutionInfo.setFailureException(
this.getProp(ConfigurationKeys.TASK_FAILURE_EXCEPTION_KEY));
}
taskExecutionInfo.setHighWatermark(this.getHighWaterMark());
// Add extract/table information
Table table = new Table();
Extract extract = this.getExtract();
table.setNamespace(extract.getNamespace());
table.setName(extract.getTable());
if (extract.hasType()) {
table.setType(TableTypeEnum.valueOf(extract.getType().name()));
}
taskExecutionInfo.setTable(table);
// Add task metrics
TaskMetrics taskMetrics = TaskMetrics.get(this);
MetricArray metricArray = new MetricArray();
for (Map.Entry<String, ? extends com.codahale.metrics.Metric> entry :
taskMetrics.getMetricContext().getCounters().entrySet()) {
Metric counter = new Metric();
counter.setGroup(MetricGroup.TASK.name());
counter.setName(entry.getKey());
counter.setType(MetricTypeEnum.valueOf(GobblinMetrics.MetricType.COUNTER.name()));
counter.setValue(Long.toString(((Counter) entry.getValue()).getCount()));
metricArray.add(counter);
}
for (Map.Entry<String, ? extends com.codahale.metrics.Metric> entry :
taskMetrics.getMetricContext().getMeters().entrySet()) {
Metric meter = new Metric();
meter.setGroup(MetricGroup.TASK.name());
meter.setName(entry.getKey());
meter.setType(MetricTypeEnum.valueOf(GobblinMetrics.MetricType.METER.name()));
meter.setValue(Double.toString(((Meter) entry.getValue()).getMeanRate()));
metricArray.add(meter);
}
for (Map.Entry<String, ? extends com.codahale.metrics.Metric> entry :
taskMetrics.getMetricContext().getGauges().entrySet()) {
Metric gauge = new Metric();
gauge.setGroup(MetricGroup.TASK.name());
gauge.setName(entry.getKey());
gauge.setType(MetricTypeEnum.valueOf(GobblinMetrics.MetricType.GAUGE.name()));
gauge.setValue(((Gauge) entry.getValue()).getValue().toString());
metricArray.add(gauge);
}
taskExecutionInfo.setMetrics(metricArray);
// Add task properties
Map<String, String> taskProperties = Maps.newHashMap();
for (String name : this.getPropertyNames()) {
String value = this.getProp(name);
if (!Strings.isNullOrEmpty(value)) taskProperties.put(name, value);
}
taskExecutionInfo.setTaskProperties(new StringMap(taskProperties));
return taskExecutionInfo;
}
