How to Send Message to Specific Partition in Kafka

To send a message to a specific partition in Kafka, use the KafkaProducer API and specify the partition number in the ProducerRecord constructor. You can also control partitioning by providing a key that Kafka hashes to select the partition.

Syntax

The KafkaProducer.send() method allows specifying the partition explicitly by passing the partition number in the ProducerRecord constructor. The method signature looks like this:

• send(ProducerRecord<K, V> record) - sends message with automatic partitioning
• send(ProducerRecord<K, V> record, Callback callback) - sends message with callback
• ProducerRecord(String topic, Integer partition, K key, V value) - constructor where partition is the target partition number

By creating a ProducerRecord with a specific partition number, you control exactly where the message goes.

java

ProducerRecord<String, String> record = new ProducerRecord<>("my-topic", 2, "key1", "message to partition 2");
producer.send(record);

Example

This example shows how to send a message to partition 1 of topic my-topic using the Java Kafka client.

java

import org.apache.kafka.clients.producer.KafkaProducer;
import org.apache.kafka.clients.producer.ProducerRecord;
import org.apache.kafka.clients.producer.ProducerConfig;
import org.apache.kafka.common.serialization.StringSerializer;
import java.util.Properties;

public class SendToPartition {
    public static void main(String[] args) {
        Properties props = new Properties();
        props.put(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, "localhost:9092");
        props.put(ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG, StringSerializer.class.getName());
        props.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG, StringSerializer.class.getName());

        KafkaProducer<String, String> producer = new KafkaProducer<>(props);

        // Create record for partition 1
        ProducerRecord<String, String> record = new ProducerRecord<>("my-topic", 1, "myKey", "Hello Partition 1");

        producer.send(record, (metadata, exception) -> {
            if (exception == null) {
                System.out.println("Sent to partition: " + metadata.partition() + ", offset: " + metadata.offset());
            } else {
                exception.printStackTrace();
            }
        });

        producer.flush();
        producer.close();
    }
}

Output

Sent to partition: 1, offset: 0

Common Pitfalls

• Not specifying partition: If you omit the partition number, Kafka uses the key or round-robin to assign partitions automatically.
• Invalid partition number: Specifying a partition number that does not exist for the topic causes an error.
• Ignoring key: If you want consistent partitioning by key, provide a key instead of hardcoding partition.

Example of wrong and right ways:

java

// Wrong: No partition specified, message goes to any partition
ProducerRecord<String, String> record1 = new ProducerRecord<>("my-topic", "key1", "message");

// Right: Specify partition explicitly
ProducerRecord<String, String> record2 = new ProducerRecord<>("my-topic", 0, "key1", "message to partition 0");

Quick Reference

Concept	Usage
Specify partition	Use ProducerRecord(topic, partition, key, value)
Send message	producer.send(record)
Partition by key	Use ProducerRecord(topic, key, value) without partition
Check partition count	Use AdminClient or Kafka tools to verify topic partitions

Key Takeaways

Use ProducerRecord constructor with partition number to send to a specific partition.

Providing a key lets Kafka decide partition consistently via hashing.

Always verify the partition number exists for the topic to avoid errors.

Omitting partition sends messages using Kafka's default partitioner.

Use callbacks to confirm message delivery and partition assignment.