@Override
public void emitTuples() {
if (currentWindowId <= idempotentStorageManager.getLargestRecoveryWindow()) {
return;
}
if (inputStream == null) {
try {
if (currentFile != null && offset > 0) {
// open file resets offset to 0 so this a way around it.
int tmpOffset = offset;
if (fs.exists(new Path(currentFile))) {
this.inputStream = openFile(new Path(currentFile));
offset = tmpOffset;
skipCount = tmpOffset;
} else {
currentFile = null;
offset = 0;
skipCount = 0;
}
} else if (!unfinishedFiles.isEmpty()) {
retryFailedFile(unfinishedFiles.poll());
} else if (!pendingFiles.isEmpty()) {
String newPathString = pendingFiles.iterator().next();
pendingFiles.remove(newPathString);
if (fs.exists(new Path(newPathString)))
this.inputStream = openFile(new Path(newPathString));
} else if (!failedFiles.isEmpty()) {
retryFailedFile(failedFiles.poll());
} else {
scanDirectory();
}
} catch (IOException ex) {
failureHandling(ex);
}
}
if (inputStream != null) {
int startOffset = offset;
String file = currentFile; // current file is reset to null when closed.
try {
int counterForTuple = 0;
while (counterForTuple++ < emitBatchSize) {
T line = readEntity();
if (line == null) {
LOG.info("done reading file ({} entries).", offset);
closeFile(inputStream);
break;
}
// If skipCount is non zero, then failed file recovery is going on, skipCount is
// used to prevent already emitted records from being emitted again during recovery.
// When failed file is open, skipCount is set to the last read offset for that file.
//
if (skipCount == 0) {
offset++;
emit(line);
} else {
skipCount--;
}
}
} catch (IOException e) {
failureHandling(e);
}
// Only when something was emitted from the file then we record it for entry.
if (offset > startOffset) {
currentWindowRecoveryState.add(new RecoveryEntry(file, startOffset, offset));
}
}
}
protected void replay(long windowId) {
// This operator can partition itself dynamically. When that happens a file can be re-hashed
// to a different partition than the previous one. In order to handle this, the partition loads
// all the recovery data for a window and then processes only those files which would be hashed
// to it in the current run.
try {
Map<Integer, Object> recoveryDataPerOperator = idempotentStorageManager.load(windowId);
for (Object recovery : recoveryDataPerOperator.values()) {
@SuppressWarnings("unchecked")
LinkedList<RecoveryEntry> recoveryData = (LinkedList<RecoveryEntry>) recovery;
for (RecoveryEntry recoveryEntry : recoveryData) {
if (scanner.acceptFile(recoveryEntry.file)) {
// The operator may have continued processing the same file in multiple windows.
// So the recovery states of subsequent windows will have an entry for that file however
// the offset changes.
// In this case we continue reading from previously opened stream.
if (currentFile == null
|| !(currentFile.equals(recoveryEntry.file)
&& offset == recoveryEntry.startOffset)) {
if (inputStream != null) {
closeFile(inputStream);
}
processedFiles.add(recoveryEntry.file);
// removing the file from failed and unfinished queues and pending set
Iterator<FailedFile> failedFileIterator = failedFiles.iterator();
while (failedFileIterator.hasNext()) {
FailedFile ff = failedFileIterator.next();
if (ff.path.equals(recoveryEntry.file) && ff.offset == recoveryEntry.startOffset) {
failedFileIterator.remove();
break;
}
}
Iterator<FailedFile> unfinishedFileIterator = unfinishedFiles.iterator();
while (unfinishedFileIterator.hasNext()) {
FailedFile ff = unfinishedFileIterator.next();
if (ff.path.equals(recoveryEntry.file) && ff.offset == recoveryEntry.startOffset) {
unfinishedFileIterator.remove();
break;
}
}
if (pendingFiles.contains(recoveryEntry.file)) {
pendingFiles.remove(recoveryEntry.file);
}
inputStream =
retryFailedFile(new FailedFile(recoveryEntry.file, recoveryEntry.startOffset));
while (offset < recoveryEntry.endOffset) {
T line = readEntity();
offset++;
emit(line);
}
} else {
while (offset < recoveryEntry.endOffset) {
T line = readEntity();
offset++;
emit(line);
}
}
}
}
}
} catch (IOException e) {
throw new RuntimeException("replay", e);
}
}