Skip to main content

Kafka source connector


You can download all the Pulsar connectors on download page.

The Kafka source connector pulls messages from Kafka topics and persists the messages to Pulsar topics.

This guide explains how to configure and use the Kafka source connector.


The configuration of the Kafka source connector has the following properties.


bootstrapServersStringtrue" " (empty string)A comma-separated list of host and port pairs for establishing the initial connection to the Kafka cluster.
securityProtocolStringfalse" " (empty string)The protocol used to communicate with Kafka brokers.
saslMechanismStringfalse" " (empty string)The SASL mechanism used for Kafka client connections.
saslJaasConfigStringfalse" " (empty string)The JAAS login context parameters for SASL connections in the format used by JAAS configuration files.
sslEnabledProtocolsStringfalse" " (empty string)The list of protocols enabled for SSL connections.
sslEndpointIdentificationAlgorithmStringfalse" " (empty string)The endpoint identification algorithm to validate server hostnames using a server certificate.
sslTruststoreLocationStringfalse" " (empty string)The location of the trust store file.
sslTruststorePasswordStringfalse" " (empty string)The password of the trust store file.
groupIdStringtrue" " (empty string)A unique string that identifies the group of consumer processes to which this consumer belongs.
fetchMinByteslongfalse1The minimum byte expected for each fetch response.
autoCommitEnabledbooleanfalsetrueIf set to true, the consumer's offset is periodically committed in the background.

This committed offset is used when the process fails as the position from which a new consumer begins.
autoCommitIntervalMslongfalse5000The frequency in milliseconds that the consumer offsets are auto-committed to Kafka if autoCommitEnabled is set to true.
heartbeatIntervalMslongfalse3000The interval between heartbeats to the consumer when using Kafka's group management facilities.

Note: heartbeatIntervalMs must be smaller than sessionTimeoutMs.
sessionTimeoutMslongfalse30000The timeout used to detect consumer failures when using Kafka's group management facility.
topicStringtrue" " (empty string)The Kafka topic that sends messages to Pulsar.
consumerConfigPropertiesMapfalse" " (empty string)The consumer configuration properties to be passed to consumers.

Note: other properties specified in the connector configuration file take precedence over this configuration.
keyDeserializationClassStringfalseorg.apache.kafka.common.serialization.StringDeserializerThe deserializer class for Kafka consumers to deserialize keys.
The deserializer is set by a specific implementation of KafkaAbstractSource.
valueDeserializationClassStringfalseorg.apache.kafka.common.serialization.ByteArrayDeserializerThe deserializer class for Kafka consumers to deserialize values.
autoOffsetResetStringfalseearliestThe default offset reset policy.

Schema Managementโ€‹

This Kafka source connector applies the schema to the topic depending on the data type that is present on the Kafka topic. You can detect the data type from the keyDeserializationClass and valueDeserializationClass configuration parameters.

If the valueDeserializationClass is org.apache.kafka.common.serialization.StringDeserializer, you can set Schema.STRING() as schema type on the Pulsar topic.

If valueDeserializationClass is io.confluent.kafka.serializers.KafkaAvroDeserializer, Pulsar downloads the AVRO schema from the Confluent Schema Registryยฎ and sets it properly on the Pulsar topic.

In this case, you need to set schema.registry.url inside of the consumerConfigProperties configuration entry of the source.

If keyDeserializationClass is not org.apache.kafka.common.serialization.StringDeserializer, it means that you do not have a string as a key and the Kafka Source uses the KeyValue schema type with the SEPARATED encoding.

Pulsar supports the AVRO format for keys.

In this case, you can have a Pulsar topic with the following properties:

  • Schema: KeyValue schema with SEPARATED encoding
  • Key: the key content of the Kafka message (base64-encoded)
  • Value: the value content of the Kafka message
  • KeySchema: the schema detected from keyDeserializationClass
  • ValueSchema: the schema detected from valueDeserializationClass

Topic compaction and partition routing use the Pulsar key, which contains the Kafka key, and so they are driven by the same value that you have on Kafka.

When you consume data from Pulsar topics, you can use the KeyValue schema. In this way, you can decode the data properly. If you want to access the raw key, you can use the Message#getKeyBytes() API.


Before using the Kafka source connector, you need to create a configuration file through one of the following methods.

  • JSON

    "bootstrapServers": "pulsar-kafka:9092",
    "groupId": "test-pulsar-io",
    "topic": "my-topic",
    "sessionTimeoutMs": "10000",
    "autoCommitEnabled": false
  • YAML

    bootstrapServers: "pulsar-kafka:9092"
    groupId: "test-pulsar-io"
    topic: "my-topic"
    sessionTimeoutMs: "10000"
    autoCommitEnabled: false


You can make the Kafka source connector as a Pulsar built-in connector and use it on a standalone cluster or an on-premises cluster.

Standalone clusterโ€‹

This example describes how to use the Kafka source connector to feed data from Kafka and write data to Pulsar topics in the standalone mode.


  • Install Docker(Community Edition).


  1. Download and start the Confluent Platform. For details, see the documentation to install the Kafka service locally.

  2. Pull a Pulsar image and start Pulsar in standalone mode.

    docker pull apachepulsar/pulsar:latest

    docker run -d -it -p 6650:6650 -p 8080:8080 -v $PWD/data:/pulsar/data --name pulsar-kafka-standalone apachepulsar/pulsar:latest bin/pulsar standalone
  3. Create a producer file

    from kafka import KafkaProducer
    producer = KafkaProducer(bootstrap_servers='localhost:9092')
    future = producer.send('my-topic', b'hello world')
  4. Create a consumer file

    import pulsar

    client = pulsar.Client('pulsar://localhost:6650')
    consumer = client.subscribe('my-topic', subscription_name='my-aa')

    while True:
    msg = consumer.receive()
    print msg
    print dir(msg)
    print("Received message: '%s'" %

  5. Copy the following files to Pulsar.

    docker cp pulsar-io-kafka.nar pulsar-kafka-standalone:/pulsar
    docker cp kafkaSourceConfig.yaml pulsar-kafka-standalone:/pulsar/conf
  6. Open a new terminal window and start the Kafka source connector in local run mode.

    docker exec -it pulsar-kafka-standalone /bin/bash

    ./bin/pulsar-admin source localrun \
    --archive ./pulsar-io-kafka.nar \
    --tenant public \
    --namespace default \
    --name kafka \
    --destination-topic-name my-topic \
    --source-config-file ./conf/kafkaSourceConfig.yaml \
    --parallelism 1
  7. Open a new terminal window and run the Kafka producer locally.

  8. Open a new terminal window and run the Pulsar consumer locally.


The following information appears on the consumer terminal window.

Received message: 'hello world'

On-premises clusterโ€‹

This example explains how to create a Kafka source connector in an on-premises cluster.

  1. Copy the NAR package of the Kafka connector to the Pulsar connectors directory.

    cp pulsar-io-kafka-{{connector:version}}.nar $PULSAR_HOME/connectors/pulsar-io-kafka-{{connector:version}}.nar
  2. Reload all built-in connectors.

    PULSAR_HOME/bin/pulsar-admin sources reload
  3. Check whether the Kafka source connector is available on the list or not.

    PULSAR_HOME/bin/pulsar-admin sources available-sources
  4. Create a Kafka source connector on a Pulsar cluster using the pulsar-admin sources create command.

    PULSAR_HOME/bin/pulsar-admin sources create \
    --source-config-file <kafka-source-config.yaml>