You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@kafka.apache.org by "Ismael Juma (JIRA)" <ji...@apache.org> on 2017/01/05 15:16:59 UTC
[jira] [Commented] (KAFKA-4597) Record metadata returned by
producer doesn't consider log append time
[ https://issues.apache.org/jira/browse/KAFKA-4597?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15801615#comment-15801615 ]
Ismael Juma commented on KAFKA-4597:
------------------------------------
Good catch. Log append time was passed to the `RecordMetadata` passed to the callbacks, but not the `RecordMetadata` returned in `send`.
> Record metadata returned by producer doesn't consider log append time
> ---------------------------------------------------------------------
>
> Key: KAFKA-4597
> URL: https://issues.apache.org/jira/browse/KAFKA-4597
> Project: Kafka
> Issue Type: Bug
> Components: clients, producer
> Affects Versions: 0.10.1.1
> Reporter: Alex Fechner
> Assignee: Ismael Juma
> Fix For: 0.10.2.0
>
>
> Kafka topics might be configured to record timestamps of the messages produced. There are two different timestamps which might be stored:
> # Record *create time*: The time the record is created by the client.
> # Log *append time*: The time the record has been added to the log by the broker.
> The [ProducerRecord|https://kafka.apache.org/0101/javadoc/org/apache/kafka/clients/producer/ProducerRecord.html] docs state:
> {quote}
> In either of the cases above, the timestamp that has actually been used will be returned to user in RecordMetadata
> {quote}
> However I found the *create time* used in both cases.
> The following class creates two topics, one configured to store *create time*. The other used *log append time*. It produces 10 messages in each topic and outputs the timestamps from the record meta data as well as those fetched by a consumer client.
> {code:java}
> import kafka.admin.AdminUtils;
> import kafka.admin.RackAwareMode;
> import kafka.utils.ZKStringSerializer$;
> import kafka.utils.ZkUtils;
> import org.I0Itec.zkclient.ZkClient;
> import org.I0Itec.zkclient.ZkConnection;
> import org.apache.kafka.clients.consumer.ConsumerRecord;
> import org.apache.kafka.clients.consumer.KafkaConsumer;
> import org.apache.kafka.clients.producer.KafkaProducer;
> import org.apache.kafka.clients.producer.Producer;
> import org.apache.kafka.clients.producer.ProducerRecord;
> import org.apache.kafka.clients.producer.RecordMetadata;
> import org.apache.kafka.common.TopicPartition;
> import java.util.Arrays;
> import java.util.Properties;
> import java.util.concurrent.ExecutionException;
> public class KafkaTimestampTest {
> public static void main(String[] args) throws ExecutionException, InterruptedException {
> String ip = "127.0.0.1";
> Properties producerProperties = new Properties();
> producerProperties.put("bootstrap.servers", ip + ":9092");
> producerProperties.put("acks", "all");
> producerProperties.put("retries", 0);
> producerProperties.put("batch.size", 16384);
> producerProperties.put("linger.ms", 1);
> producerProperties.put("buffer.memory", 33554432);
> producerProperties.put("key.serializer", "org.apache.kafka.common.serialization.StringSerializer");
> producerProperties.put("value.serializer", "org.apache.kafka.common.serialization.StringSerializer");
> Properties consumerProperties = new Properties();
> consumerProperties.put("bootstrap.servers", ip + ":9092");
> consumerProperties.put("enable.auto.commit", "false");
> consumerProperties.put("session.timeout.ms", "30000");
> consumerProperties.put("key.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");
> consumerProperties.put("value.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");
> Producer<String, String> producer = new KafkaProducer<>(producerProperties);
> KafkaConsumer<String, String> consumer = new KafkaConsumer<>(consumerProperties);
> ZkClient zkClient = new ZkClient(ip + ":2181", 10000, 10000, ZKStringSerializer$.MODULE$);
> ZkConnection zkConnection = new ZkConnection(ip + ":2181");
> ZkUtils zkUtils = new ZkUtils(zkClient, zkConnection, false);
> TopicPartition topicPartitionWithCreateTime = new TopicPartition("test-topic-with-create-time", 0);
> TopicPartition topicPartitionWithLogAppendTime = new TopicPartition("test-topic-with-log-append-time", 0);
> // create topic with create time
> if (!AdminUtils.topicExists(zkUtils, topicPartitionWithCreateTime.topic())) {
> Properties topicProperties = new Properties();
> topicProperties.put("message.timestamp.type", "CreateTime");
> AdminUtils.createTopic(zkUtils, topicPartitionWithCreateTime.topic(), 1, 1, topicProperties, RackAwareMode.Disabled$.MODULE$);
> }
> // create topic with log append time
> if (!AdminUtils.topicExists(zkUtils, topicPartitionWithLogAppendTime.topic())) {
> Properties topicProperties = new Properties();
> topicProperties.put("message.timestamp.type", "LogAppendTime");
> AdminUtils.createTopic(zkUtils, topicPartitionWithLogAppendTime.topic(), 1, 1, topicProperties, RackAwareMode.Disabled$.MODULE$);
> }
> consumer.assign(Arrays.asList(topicPartitionWithLogAppendTime, topicPartitionWithCreateTime));
> String format = "#%s, MetaDataOffset: %s, MetaDataTime: %s, ConsumerRecordOffset: %s, ConsumerRecordTime: %s";
> System.out.println(String.format("Create messages into topic %s ...", topicPartitionWithCreateTime));
> for (int i = 0; i < 10; i++) {
> RecordMetadata recordMetadata = producer.send(new ProducerRecord<>(topicPartitionWithCreateTime.topic(), topicPartitionWithCreateTime.partition(), "", "message")).get();
> consumer.seek(topicPartitionWithCreateTime, recordMetadata.offset());
> ConsumerRecord<String, String> consumerRecord = consumer.poll(1000).records(topicPartitionWithCreateTime).get(0);
> System.out.println(String.format(format, i + 1, recordMetadata.offset(), recordMetadata.timestamp(), consumerRecord.offset(), consumerRecord.timestamp()));
> }
> System.out.println(String.format("Create messages into topic %s...", topicPartitionWithLogAppendTime));
> for (int i = 0; i < 10; i++) {
> RecordMetadata recordMetadata = producer.send(new ProducerRecord<>(topicPartitionWithLogAppendTime.topic(), topicPartitionWithLogAppendTime.partition(), "", "message")).get();
> consumer.seek(topicPartitionWithLogAppendTime, recordMetadata.offset());
> ConsumerRecord<String, String> consumerRecord = consumer.poll(1000).records(topicPartitionWithLogAppendTime).get(0);
> System.out.println(String.format(format, i + 1, recordMetadata.offset(), recordMetadata.timestamp(), consumerRecord.offset(), consumerRecord.timestamp()));
> }
> AdminUtils.deleteTopic(zkUtils, topicPartitionWithCreateTime.topic());
> AdminUtils.deleteTopic(zkUtils, topicPartitionWithLogAppendTime.topic());
> }
> }
> {code}
> The output shows that in case of *log append time* the timestamps differ.
> {code}
> Create messages into topic test-topic-with-create-time-0 ...
> #1, MetaDataOffset: 0, MetaDataTime: 1483623773788, ConsumerRecordOffset: 0, ConsumerRecordTime: 1483623773788
> #2, MetaDataOffset: 1, MetaDataTime: 1483623774178, ConsumerRecordOffset: 1, ConsumerRecordTime: 1483623774178
> #3, MetaDataOffset: 2, MetaDataTime: 1483623774183, ConsumerRecordOffset: 2, ConsumerRecordTime: 1483623774183
> #4, MetaDataOffset: 3, MetaDataTime: 1483623774188, ConsumerRecordOffset: 3, ConsumerRecordTime: 1483623774188
> #5, MetaDataOffset: 4, MetaDataTime: 1483623774193, ConsumerRecordOffset: 4, ConsumerRecordTime: 1483623774193
> #6, MetaDataOffset: 5, MetaDataTime: 1483623774197, ConsumerRecordOffset: 5, ConsumerRecordTime: 1483623774197
> #7, MetaDataOffset: 6, MetaDataTime: 1483623774202, ConsumerRecordOffset: 6, ConsumerRecordTime: 1483623774202
> #8, MetaDataOffset: 7, MetaDataTime: 1483623774207, ConsumerRecordOffset: 7, ConsumerRecordTime: 1483623774207
> #9, MetaDataOffset: 8, MetaDataTime: 1483623774212, ConsumerRecordOffset: 8, ConsumerRecordTime: 1483623774212
> #10, MetaDataOffset: 9, MetaDataTime: 1483623774217, ConsumerRecordOffset: 9, ConsumerRecordTime: 1483623774217
> Create messages into topic test-topic-with-log-append-time-0...
> #1, MetaDataOffset: 0, MetaDataTime: 1483623774224, ConsumerRecordOffset: 0, ConsumerRecordTime: 1483623774992
> #2, MetaDataOffset: 1, MetaDataTime: 1483623774230, ConsumerRecordOffset: 1, ConsumerRecordTime: 1483623774997
> #3, MetaDataOffset: 2, MetaDataTime: 1483623774235, ConsumerRecordOffset: 2, ConsumerRecordTime: 1483623775002
> #4, MetaDataOffset: 3, MetaDataTime: 1483623774239, ConsumerRecordOffset: 3, ConsumerRecordTime: 1483623775007
> #5, MetaDataOffset: 4, MetaDataTime: 1483623774244, ConsumerRecordOffset: 4, ConsumerRecordTime: 1483623775011
> #6, MetaDataOffset: 5, MetaDataTime: 1483623774248, ConsumerRecordOffset: 5, ConsumerRecordTime: 1483623775015
> #7, MetaDataOffset: 6, MetaDataTime: 1483623774253, ConsumerRecordOffset: 6, ConsumerRecordTime: 1483623775020
> #8, MetaDataOffset: 7, MetaDataTime: 1483623774257, ConsumerRecordOffset: 7, ConsumerRecordTime: 1483623775024
> #9, MetaDataOffset: 8, MetaDataTime: 1483623774262, ConsumerRecordOffset: 8, ConsumerRecordTime: 1483623775029
> #10, MetaDataOffset: 9, MetaDataTime: 1483623774267, ConsumerRecordOffset: 9, ConsumerRecordTime: 1483623775034
> {code}
> I assume the timestamps in the record meta data represent the create time but could not ensure that.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)