/**
* Whether the task should directly publish its output data to the final publisher output
* directory.
*
* <p>The task should publish its output data directly if {@link
* ConfigurationKeys#PUBLISH_DATA_AT_JOB_LEVEL} is set to false AND any of the following
* conditions is satisfied:
*
* <ul>
* <li>The {@link JobCommitPolicy#COMMIT_ON_PARTIAL_SUCCESS} policy is used.
* <li>The {@link JobCommitPolicy#COMMIT_SUCCESSFUL_TASKS} policy is used. and all {@link Fork}s
* of this {@link Task} succeeded.
* </ul>
*/
private boolean shouldPublishDataInTask() {
boolean publishDataAtJobLevel =
this.taskState.getPropAsBoolean(
ConfigurationKeys.PUBLISH_DATA_AT_JOB_LEVEL,
ConfigurationKeys.DEFAULT_PUBLISH_DATA_AT_JOB_LEVEL);
if (publishDataAtJobLevel) {
LOG.info(
String.format(
"%s is true. Will publish data at the job level.",
ConfigurationKeys.PUBLISH_DATA_AT_JOB_LEVEL));
return false;
}
JobCommitPolicy jobCommitPolicy = JobCommitPolicy.getCommitPolicy(this.taskState);
if (jobCommitPolicy == JobCommitPolicy.COMMIT_SUCCESSFUL_TASKS) {
return this.taskState.getWorkingState() == WorkUnitState.WorkingState.SUCCESSFUL;
}
if (jobCommitPolicy == JobCommitPolicy.COMMIT_ON_PARTIAL_SUCCESS) {
return true;
}
LOG.info("Will publish data at the job level with job commit policy: " + jobCommitPolicy);
return false;
}
/**
* Get latest water mark from previous work unit states.
*
* @param state Source state
* @return latest water mark (high water mark)
*/
private long getLatestWatermarkFromMetadata(SourceState state) {
LOG.debug("Get latest watermark from the previous run");
long latestWaterMark = ConfigurationKeys.DEFAULT_WATERMARK_VALUE;
List<WorkUnitState> previousWorkUnitStates =
Lists.newArrayList(state.getPreviousWorkUnitStates());
List<Long> previousWorkUnitStateHighWatermarks = Lists.newArrayList();
List<Long> previousWorkUnitLowWatermarks = Lists.newArrayList();
if (previousWorkUnitStates.isEmpty()) {
LOG.info(
"No previous work unit states found; Latest watermark - Default watermark: "
+ latestWaterMark);
return latestWaterMark;
}
boolean hasFailedRun = false;
boolean isCommitOnFullSuccess = false;
boolean isDataProcessedInPreviousRun = false;
JobCommitPolicy commitPolicy =
JobCommitPolicy.forName(
state.getProp(
ConfigurationKeys.JOB_COMMIT_POLICY_KEY,
ConfigurationKeys.DEFAULT_JOB_COMMIT_POLICY));
if (commitPolicy == JobCommitPolicy.COMMIT_ON_FULL_SUCCESS) {
isCommitOnFullSuccess = true;
}
for (WorkUnitState workUnitState : previousWorkUnitStates) {
long processedRecordCount = 0;
LOG.info(
"State of the previous task: "
+ workUnitState.getId()
+ ":"
+ workUnitState.getWorkingState());
if (workUnitState.getWorkingState() == WorkingState.FAILED
|| workUnitState.getWorkingState() == WorkingState.CANCELLED
|| workUnitState.getWorkingState() == WorkingState.RUNNING
|| workUnitState.getWorkingState() == WorkingState.PENDING) {
hasFailedRun = true;
} else {
processedRecordCount =
workUnitState.getPropAsLong(ConfigurationKeys.EXTRACTOR_ROWS_EXPECTED);
if (processedRecordCount != 0) {
isDataProcessedInPreviousRun = true;
}
}
LOG.info(
"Low watermark of the previous task: "
+ workUnitState.getId()
+ ":"
+ workUnitState.getWorkunit().getLowWaterMark());
LOG.info(
"High watermark of the previous task: "
+ workUnitState.getId()
+ ":"
+ workUnitState.getHighWaterMark());
LOG.info("Record count of the previous task: " + processedRecordCount + "\n");
// Consider high water mark of the previous work unit, if it is
// extracted any data
if (processedRecordCount != 0) {
previousWorkUnitStateHighWatermarks.add(workUnitState.getHighWaterMark());
}
previousWorkUnitLowWatermarks.add(this.getLowWatermarkFromWorkUnit(workUnitState));
}
// If commit policy is full and it has failed run, get latest water mark
// as
// minimum of low water marks from previous states.
if (isCommitOnFullSuccess && hasFailedRun) {
long previousLowWatermark = Collections.min(previousWorkUnitLowWatermarks);
WorkUnitState previousState = previousWorkUnitStates.get(0);
ExtractType extractType =
ExtractType.valueOf(
previousState
.getProp(ConfigurationKeys.SOURCE_QUERYBASED_EXTRACT_TYPE)
.toUpperCase());
// add backup seconds only for snapshot extracts but not for appends
if (extractType == ExtractType.SNAPSHOT) {
int backupSecs =
previousState.getPropAsInt(
ConfigurationKeys.SOURCE_QUERYBASED_LOW_WATERMARK_BACKUP_SECS, 0);
String watermarkType =
previousState.getProp(ConfigurationKeys.SOURCE_QUERYBASED_WATERMARK_TYPE);
latestWaterMark = this.addBackedUpSeconds(previousLowWatermark, backupSecs, watermarkType);
} else {
latestWaterMark = previousLowWatermark;
}
LOG.info(
"Previous job was COMMIT_ON_FULL_SUCCESS but it was failed; Latest watermark - "
+ "Min watermark from WorkUnits: "
+ latestWaterMark);
}
// If commit policy is full and there are no failed tasks or commit
// policy is partial,
// get latest water mark as maximum of high water marks from previous
// tasks.
else {
if (isDataProcessedInPreviousRun) {
latestWaterMark = Collections.max(previousWorkUnitStateHighWatermarks);
LOG.info(
"Previous run was successful. Latest watermark - Max watermark from WorkUnitStates: "
+ latestWaterMark);
} else {
latestWaterMark = Collections.min(previousWorkUnitLowWatermarks);
LOG.info(
"Previous run was successful but no data found. Latest watermark - Min watermark from WorkUnitStates: "
+ latestWaterMark);
}
}
return latestWaterMark;
}
