sv2000 commented on a change in pull request #2900: [GOBBLIN-1040]
HighLevelConsumer re-design by removing references to …
URL: https://github.com/apache/incubator-gobblin/pull/2900#discussion_r386171395
##########
File path:
gobblin-runtime/src/main/java/org/apache/gobblin/runtime/kafka/HighLevelConsumer.java
##########
@@ -125,81 +181,127 @@ protected void createMetrics() {
protected List<Tag<?>> getTagsForMetrics() {
List<Tag<?>> tags = Lists.newArrayList();
tags.add(new Tag<>(RuntimeMetrics.TOPIC, this.topic));
- tags.add(new Tag<>(RuntimeMetrics.GROUP_ID,
this.consumerConfig.groupId()));
+ tags.add(new Tag<>(RuntimeMetrics.GROUP_ID,
ConfigUtils.getString(this.config, GROUP_ID_KEY, DEFAULT_GROUP_ID)));
return tags;
}
/**
- * Called every time a message is read from the stream. Implementation must
be thread-safe if {@link #numThreads} is
+ * Called every time a message is read from the queue. Implementation must
be thread-safe if {@link #numThreads} is
* set larger than 1.
*/
- protected abstract void processMessage(MessageAndMetadata<K, V> message);
+ protected abstract void processMessage(DecodeableKafkaRecord<K,V> message);
@Override
protected void startUp() {
buildMetricsContextAndMetrics();
- this.consumer = createConsumerConnector();
+ processQueues();
+ if(!enableAutoCommit) {
+ offsetCommitExecutor = Executors.newSingleThreadScheduledExecutor();
+ offsetCommitExecutor.scheduleAtFixedRate(new Runnable() {
+ @Override
+ public void run() {
+ offsetCommitter();
+ }
+ }, 0,100, TimeUnit.MILLISECONDS);
+ }
- List<KafkaStream<byte[], byte[]>> streams = createStreams();
- this.executor = Executors.newFixedThreadPool(this.numThreads);
+ mainExecutor.scheduleAtFixedRate(new Runnable() {
+ @Override
+ public void run() {
+ consume();
+ }
+ }, 0, 50, TimeUnit.MILLISECONDS);
+ }
- // now create an object to consume the messages
- //
- int threadNumber = 0;
- for (final KafkaStream stream : streams) {
- this.executor.execute(new MonitorConsumer(stream));
- threadNumber++;
+ /**
+ * Consumes {@link KafkaConsumerRecord}s and adds to a queue
+ * Note: All records from a KafkaPartition are added to the same queue.
+ * A queue can contain records from multiple partitions if partitions >
numThreads(queues)
+ */
+ private void consume() {
+ try {
+ Iterator<KafkaConsumerRecord> itr = gobblinKafkaConsumerClient.consume();
+ while (itr.hasNext()) {
+ KafkaConsumerRecord record = itr.next();
+ int idx = record.getPartition() % numThreads;
+ queues[idx].put(record);
+ }
+ } catch (InterruptedException e) {
+ Thread.currentThread().interrupt();
}
}
- protected ConsumerConfig createConsumerConfig(Config config) {
- Properties props = ConfigUtils.configToProperties(config);
- if (!props.containsKey(GROUP_ID_KEY)) {
- props.setProperty(GROUP_ID_KEY, DEFAULT_GROUP_ID);
+ private void processQueues() {
+ for(BlockingQueue queue : queues) {
+ queueExecutor.execute(new QueueProcessor(queue));
}
- return new ConsumerConfig(props);
}
- protected ConsumerConnector createConsumerConnector() {
- return Consumer.createJavaConsumerConnector(this.consumerConfig);
+ /**
+ * A thread that commits offsets to kafka at regular intervals
+ */
+ private void offsetCommitter() {
+ if(recordsProcessed.intValue() >= offsetsCommitNumRecordsThreshold ||
((System.currentTimeMillis() - lastCommitTime) / 1000 >=
offsetsCommitTimeThresholdSecs)) {
Review comment:
Can you move the if condition to a separate method e.g private boolean
shouldCommitOffsets()?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
