use of io.druid.java.util.common.parsers.ParseException in project druid by druid-io.
the class InlineSchemaAvroBytesDecoder method parse.
@Override
public GenericRecord parse(ByteBuffer bytes) {
DatumReader<GenericRecord> reader = new GenericDatumReader<GenericRecord>(schemaObj);
ByteBufferInputStream inputStream = new ByteBufferInputStream(Collections.singletonList(bytes));
try {
return reader.read(null, DecoderFactory.get().binaryDecoder(inputStream, null));
} catch (Exception e) {
throw new ParseException(e, "Fail to decode avro message!");
}
}
use of io.druid.java.util.common.parsers.ParseException in project druid by druid-io.
the class InlineSchemasAvroBytesDecoder method parse.
// It is assumed that record has following format.
// byte 1 : version, static 0x1
// byte 2-5 : int schemaId
// remaining bytes would have avro data
@Override
public GenericRecord parse(ByteBuffer bytes) {
if (bytes.remaining() < 5) {
throw new ParseException("record must have at least 5 bytes carrying version and schemaId");
}
byte version = bytes.get();
if (version != V1) {
throw new ParseException("found record of arbitrary version [%s]", version);
}
int schemaId = bytes.getInt();
Schema schemaObj = schemaObjs.get(schemaId);
if (schemaObj == null) {
throw new ParseException("Failed to find schema for id [%s]", schemaId);
}
try {
DatumReader<GenericRecord> reader = new GenericDatumReader<GenericRecord>(schemaObj);
ByteBufferInputStream inputStream = new ByteBufferInputStream(Collections.singletonList(bytes));
return reader.read(null, DecoderFactory.get().binaryDecoder(inputStream, null));
} catch (Exception e) {
throw new ParseException(e, "Fail to decode avro message with schemaId [%s].", schemaId);
}
}
use of io.druid.java.util.common.parsers.ParseException in project druid by druid-io.
the class Plumbers method addNextRow.
public static void addNextRow(final Supplier<Committer> committerSupplier, final Firehose firehose, final Plumber plumber, final boolean reportParseExceptions, final FireDepartmentMetrics metrics) {
final InputRow inputRow;
try {
inputRow = firehose.nextRow();
} catch (ParseException e) {
if (reportParseExceptions) {
throw e;
} else {
log.debug(e, "Discarded row due to exception, considering unparseable.");
metrics.incrementUnparseable();
return;
}
}
if (inputRow == null) {
if (reportParseExceptions) {
throw new ParseException("null input row");
} else {
log.debug("Discarded null input row, considering unparseable.");
metrics.incrementUnparseable();
return;
}
}
final int numRows;
try {
numRows = plumber.add(inputRow, committerSupplier);
} catch (IndexSizeExceededException e) {
// plumber.add should be swapping out indexes before they fill up.
throw new ISE(e, "WTF?! Index size exceeded, this shouldn't happen. Bad Plumber!");
}
if (numRows == -1) {
metrics.incrementThrownAway();
log.debug("Discarded row[%s], considering thrownAway.", inputRow);
return;
}
metrics.incrementProcessed();
}
use of io.druid.java.util.common.parsers.ParseException in project druid by druid-io.
the class RealtimeManagerTest method setUp.
@Before
public void setUp() throws Exception {
final List<TestInputRowHolder> rows = Arrays.asList(makeRow(new DateTime("9000-01-01").getMillis()), makeRow(new ParseException("parse error")), null, makeRow(new DateTime().getMillis()));
ObjectMapper jsonMapper = new DefaultObjectMapper();
schema = new DataSchema("test", null, new AggregatorFactory[] { new CountAggregatorFactory("rows") }, new UniformGranularitySpec(Granularities.HOUR, Granularities.NONE, null), jsonMapper);
schema2 = new DataSchema("testV2", null, new AggregatorFactory[] { new CountAggregatorFactory("rows") }, new UniformGranularitySpec(Granularities.HOUR, Granularities.NONE, null), jsonMapper);
RealtimeIOConfig ioConfig = new RealtimeIOConfig(new FirehoseFactory() {
@Override
public Firehose connect(InputRowParser parser) throws IOException {
return new TestFirehose(rows.iterator());
}
}, new PlumberSchool() {
@Override
public Plumber findPlumber(DataSchema schema, RealtimeTuningConfig config, FireDepartmentMetrics metrics) {
return plumber;
}
}, null);
RealtimeIOConfig ioConfig2 = new RealtimeIOConfig(null, new PlumberSchool() {
@Override
public Plumber findPlumber(DataSchema schema, RealtimeTuningConfig config, FireDepartmentMetrics metrics) {
return plumber2;
}
}, new FirehoseFactoryV2() {
@Override
public FirehoseV2 connect(InputRowParser parser, Object arg1) throws IOException, ParseException {
return new TestFirehoseV2(rows.iterator());
}
});
RealtimeTuningConfig tuningConfig = new RealtimeTuningConfig(1, new Period("P1Y"), null, null, null, null, null, null, null, null, 0, 0, null, null);
plumber = new TestPlumber(new Sink(new Interval("0/P5000Y"), schema, tuningConfig.getShardSpec(), new DateTime().toString(), tuningConfig.getMaxRowsInMemory(), tuningConfig.isReportParseExceptions()));
realtimeManager = new RealtimeManager(Arrays.<FireDepartment>asList(new FireDepartment(schema, ioConfig, tuningConfig)), null);
plumber2 = new TestPlumber(new Sink(new Interval("0/P5000Y"), schema2, tuningConfig.getShardSpec(), new DateTime().toString(), tuningConfig.getMaxRowsInMemory(), tuningConfig.isReportParseExceptions()));
realtimeManager2 = new RealtimeManager(Arrays.<FireDepartment>asList(new FireDepartment(schema2, ioConfig2, tuningConfig)), null);
tuningConfig_0 = new RealtimeTuningConfig(1, new Period("P1Y"), null, null, null, null, null, new LinearShardSpec(0), null, null, 0, 0, null, null);
tuningConfig_1 = new RealtimeTuningConfig(1, new Period("P1Y"), null, null, null, null, null, new LinearShardSpec(1), null, null, 0, 0, null, null);
schema3 = new DataSchema("testing", null, new AggregatorFactory[] { new CountAggregatorFactory("ignore") }, new UniformGranularitySpec(Granularities.HOUR, Granularities.NONE, null), jsonMapper);
FireDepartment department_0 = new FireDepartment(schema3, ioConfig, tuningConfig_0);
FireDepartment department_1 = new FireDepartment(schema3, ioConfig2, tuningConfig_1);
QueryRunnerFactoryConglomerate conglomerate = new QueryRunnerFactoryConglomerate() {
@Override
public <T, QueryType extends Query<T>> QueryRunnerFactory<T, QueryType> findFactory(QueryType query) {
return factory;
}
};
chiefStartedLatch = new CountDownLatch(2);
RealtimeManager.FireChief fireChief_0 = new RealtimeManager.FireChief(department_0, conglomerate) {
@Override
public void run() {
super.initPlumber();
chiefStartedLatch.countDown();
}
};
RealtimeManager.FireChief fireChief_1 = new RealtimeManager.FireChief(department_1, conglomerate) {
@Override
public void run() {
super.initPlumber();
chiefStartedLatch.countDown();
}
};
realtimeManager3 = new RealtimeManager(Arrays.asList(department_0, department_1), conglomerate, ImmutableMap.<String, Map<Integer, RealtimeManager.FireChief>>of("testing", ImmutableMap.of(0, fireChief_0, 1, fireChief_1)));
startFireChiefWithPartitionNum(fireChief_0, 0);
startFireChiefWithPartitionNum(fireChief_1, 1);
}
use of io.druid.java.util.common.parsers.ParseException in project druid by druid-io.
the class KafkaIndexTask method run.
@Override
public TaskStatus run(final TaskToolbox toolbox) throws Exception {
log.info("Starting up!");
startTime = DateTime.now();
mapper = toolbox.getObjectMapper();
status = Status.STARTING;
if (chatHandlerProvider.isPresent()) {
log.info("Found chat handler of class[%s]", chatHandlerProvider.get().getClass().getName());
chatHandlerProvider.get().register(getId(), this, false);
} else {
log.warn("No chat handler detected");
}
runThread = Thread.currentThread();
// Set up FireDepartmentMetrics
final FireDepartment fireDepartmentForMetrics = new FireDepartment(dataSchema, new RealtimeIOConfig(null, null, null), null);
fireDepartmentMetrics = fireDepartmentForMetrics.getMetrics();
toolbox.getMonitorScheduler().addMonitor(new RealtimeMetricsMonitor(ImmutableList.of(fireDepartmentForMetrics), ImmutableMap.of(DruidMetrics.TASK_ID, new String[] { getId() })));
try (final Appenderator appenderator0 = newAppenderator(fireDepartmentMetrics, toolbox);
final FiniteAppenderatorDriver driver = newDriver(appenderator0, toolbox, fireDepartmentMetrics);
final KafkaConsumer<byte[], byte[]> consumer = newConsumer()) {
appenderator = appenderator0;
final String topic = ioConfig.getStartPartitions().getTopic();
// Start up, set up initial offsets.
final Object restoredMetadata = driver.startJob();
if (restoredMetadata == null) {
nextOffsets.putAll(ioConfig.getStartPartitions().getPartitionOffsetMap());
} else {
final Map<String, Object> restoredMetadataMap = (Map) restoredMetadata;
final KafkaPartitions restoredNextPartitions = toolbox.getObjectMapper().convertValue(restoredMetadataMap.get(METADATA_NEXT_PARTITIONS), KafkaPartitions.class);
nextOffsets.putAll(restoredNextPartitions.getPartitionOffsetMap());
// Sanity checks.
if (!restoredNextPartitions.getTopic().equals(ioConfig.getStartPartitions().getTopic())) {
throw new ISE("WTF?! Restored topic[%s] but expected topic[%s]", restoredNextPartitions.getTopic(), ioConfig.getStartPartitions().getTopic());
}
if (!nextOffsets.keySet().equals(ioConfig.getStartPartitions().getPartitionOffsetMap().keySet())) {
throw new ISE("WTF?! Restored partitions[%s] but expected partitions[%s]", nextOffsets.keySet(), ioConfig.getStartPartitions().getPartitionOffsetMap().keySet());
}
}
// Set up sequenceNames.
final Map<Integer, String> sequenceNames = Maps.newHashMap();
for (Integer partitionNum : nextOffsets.keySet()) {
sequenceNames.put(partitionNum, String.format("%s_%s", ioConfig.getBaseSequenceName(), partitionNum));
}
// Set up committer.
final Supplier<Committer> committerSupplier = new Supplier<Committer>() {
@Override
public Committer get() {
final Map<Integer, Long> snapshot = ImmutableMap.copyOf(nextOffsets);
return new Committer() {
@Override
public Object getMetadata() {
return ImmutableMap.of(METADATA_NEXT_PARTITIONS, new KafkaPartitions(ioConfig.getStartPartitions().getTopic(), snapshot));
}
@Override
public void run() {
// Do nothing.
}
};
}
};
Set<Integer> assignment = assignPartitionsAndSeekToNext(consumer, topic);
// Main loop.
// Could eventually support leader/follower mode (for keeping replicas more in sync)
boolean stillReading = !assignment.isEmpty();
status = Status.READING;
try {
while (stillReading) {
if (possiblyPause(assignment)) {
// The partition assignments may have changed while paused by a call to setEndOffsets() so reassign
// partitions upon resuming. This is safe even if the end offsets have not been modified.
assignment = assignPartitionsAndSeekToNext(consumer, topic);
if (assignment.isEmpty()) {
log.info("All partitions have been fully read");
publishOnStop = true;
stopRequested = true;
}
}
if (stopRequested) {
break;
}
// The retrying business is because the KafkaConsumer throws OffsetOutOfRangeException if the seeked-to
// offset is not present in the topic-partition. This can happen if we're asking a task to read from data
// that has not been written yet (which is totally legitimate). So let's wait for it to show up.
ConsumerRecords<byte[], byte[]> records = ConsumerRecords.empty();
try {
records = consumer.poll(POLL_TIMEOUT);
} catch (OffsetOutOfRangeException e) {
log.warn("OffsetOutOfRangeException with message [%s]", e.getMessage());
possiblyResetOffsetsOrWait(e.offsetOutOfRangePartitions(), consumer, toolbox);
stillReading = ioConfig.isPauseAfterRead() || !assignment.isEmpty();
}
for (ConsumerRecord<byte[], byte[]> record : records) {
if (log.isTraceEnabled()) {
log.trace("Got topic[%s] partition[%d] offset[%,d].", record.topic(), record.partition(), record.offset());
}
if (record.offset() < endOffsets.get(record.partition())) {
if (record.offset() != nextOffsets.get(record.partition())) {
throw new ISE("WTF?! Got offset[%,d] after offset[%,d] in partition[%d].", record.offset(), nextOffsets.get(record.partition()), record.partition());
}
try {
final byte[] valueBytes = record.value();
if (valueBytes == null) {
throw new ParseException("null value");
}
final InputRow row = Preconditions.checkNotNull(parser.parse(ByteBuffer.wrap(valueBytes)), "row");
if (!ioConfig.getMinimumMessageTime().isPresent() || !ioConfig.getMinimumMessageTime().get().isAfter(row.getTimestamp())) {
final SegmentIdentifier identifier = driver.add(row, sequenceNames.get(record.partition()), committerSupplier);
if (identifier == null) {
// If we allow continuing, then consider blacklisting the interval for a while to avoid constant checks.
throw new ISE("Could not allocate segment for row with timestamp[%s]", row.getTimestamp());
}
fireDepartmentMetrics.incrementProcessed();
} else {
fireDepartmentMetrics.incrementThrownAway();
}
} catch (ParseException e) {
if (tuningConfig.isReportParseExceptions()) {
throw e;
} else {
log.debug(e, "Dropping unparseable row from partition[%d] offset[%,d].", record.partition(), record.offset());
fireDepartmentMetrics.incrementUnparseable();
}
}
nextOffsets.put(record.partition(), record.offset() + 1);
}
if (nextOffsets.get(record.partition()).equals(endOffsets.get(record.partition())) && assignment.remove(record.partition())) {
log.info("Finished reading topic[%s], partition[%,d].", record.topic(), record.partition());
assignPartitions(consumer, topic, assignment);
stillReading = ioConfig.isPauseAfterRead() || !assignment.isEmpty();
}
}
}
} finally {
// persist pending data
driver.persist(committerSupplier.get());
}
synchronized (statusLock) {
if (stopRequested && !publishOnStop) {
throw new InterruptedException("Stopping without publishing");
}
status = Status.PUBLISHING;
}
final TransactionalSegmentPublisher publisher = new TransactionalSegmentPublisher() {
@Override
public boolean publishSegments(Set<DataSegment> segments, Object commitMetadata) throws IOException {
final KafkaPartitions finalPartitions = toolbox.getObjectMapper().convertValue(((Map) commitMetadata).get(METADATA_NEXT_PARTITIONS), KafkaPartitions.class);
// Sanity check, we should only be publishing things that match our desired end state.
if (!endOffsets.equals(finalPartitions.getPartitionOffsetMap())) {
throw new ISE("WTF?! Driver attempted to publish invalid metadata[%s].", commitMetadata);
}
final SegmentTransactionalInsertAction action;
if (ioConfig.isUseTransaction()) {
action = new SegmentTransactionalInsertAction(segments, new KafkaDataSourceMetadata(ioConfig.getStartPartitions()), new KafkaDataSourceMetadata(finalPartitions));
} else {
action = new SegmentTransactionalInsertAction(segments, null, null);
}
log.info("Publishing with isTransaction[%s].", ioConfig.isUseTransaction());
return toolbox.getTaskActionClient().submit(action).isSuccess();
}
};
final SegmentsAndMetadata published = driver.finish(publisher, committerSupplier.get());
if (published == null) {
throw new ISE("Transaction failure publishing segments, aborting");
} else {
log.info("Published segments[%s] with metadata[%s].", Joiner.on(", ").join(Iterables.transform(published.getSegments(), new Function<DataSegment, String>() {
@Override
public String apply(DataSegment input) {
return input.getIdentifier();
}
})), published.getCommitMetadata());
}
} catch (InterruptedException | RejectedExecutionException e) {
// handle the InterruptedException that gets wrapped in a RejectedExecutionException
if (e instanceof RejectedExecutionException && (e.getCause() == null || !(e.getCause() instanceof InterruptedException))) {
throw e;
}
// if we were interrupted because we were asked to stop, handle the exception and return success, else rethrow
if (!stopRequested) {
Thread.currentThread().interrupt();
throw e;
}
log.info("The task was asked to stop before completing");
} finally {
if (chatHandlerProvider.isPresent()) {
chatHandlerProvider.get().unregister(getId());
}
}
return success();
}
Aggregations