HTTP Sink connector

The HTTP Sink connector is a Stateless NiFi dataflow developed by Cloudera that is running in the Kafka Connect framework. Learn about the connector, its properties, and configuration.

The HTTP Sink connector obtains messages from a Kafka topic and transfers their content in a HTTP POST requests to a specified endpoint. The topic the connector receives messages from is determined by the value of the topics parameter in the configuration. The connector can forward the data it reads from Kafka as is (raw data) or can be configured to execute record processing. When record processing is enabled, the connector expects the incoming data in Avro or JSON format. In case of Avro, the connector can either read the record schema from the Avro file it receives (provided that the schema is embedded) or it can fetch the schema from Schema Registry. In case of JSON, it can either infer the schema or fetch it from Schema Registry. What strategy is used to retrieve the schema is determined by the Schema Access Strategy property.

If Schema Registry is used, and it is on a Kerberized cluster, the krb5.file property must point to the krb5.conf file that provides access to the cluster on which Schema Registry is present. This means that the krb5.conf file must be on the same cluster node that the connector runs on. The connections to Schema Registry and the HTTP server can be secured by TLS. The keystore and truststore files necessary for securing these connections must also be on the same cluster node that the connector runs on.

Properties and configuration

Configuration is passed to the connector in a JSON file during creation. The properties of the connector can be categorized into three groups. These are as follows:

Common connector properties
These are the properties of the Kafka Connect framework that are accepted by all connectors. For a comprehensive list of these properties, see the Apache Kafka documentation.
Stateless NiFi Sink properties
These are the properties that are specific to the Stateless NiFi Sink connector. All Stateless NiFi Sink connectors share and accept these properties. For a comprehensive list of these properties, see the Stateless NiFi Sink property reference.
Connector/dataflow-specific properties
These properties are unique to this specific connector. Or to be more precise, unique to the dataflow running within the connector. These properties use the following prefix:
parameter.[***CONNECTOR NAME***] Parameters:
For a comprehensive list of these properties, see the HTTP Sink properties reference.

Notes and limitations

  • Required properties must be assigned a valid value even if they are not used in the particular configuration. If a required property is not used, either leave its default value, or completely remove the property from the configuration JSON.
  • If a property that has a default value is completely removed from the configuration JSON, the system uses the default value.
  • Properties not marked as required must be completely removed from the configuration JSON if not set.
  • Schema Branch and Schema Version can not be specified at the same time.
  • The value of the Schema Access Strategy property is not independent of the value of the Kafka Message Data Format property. As a result, you must exercise caution when configuring Schema Access Strategy.
    • If the value of Kafka Message Data Format is AVRO, the possible values for Schema Access Strategy are Schema Registry, Embedded Schema or HWX Content-Encoded Schema Reference.
    • If the value of Kafka Message Data Format is JSON, the possible values for Schema Access Strategy are Schema Registry or Infer Schema.

Configuration example

In this example, the connector receives data in any format and forwards the raw data as the content of an HTTP POST request.

{
 "connector.class": "org.apache.nifi.kafka.connect.StatelessNiFiSinkConnector",
 "meta.smm.predefined.flow.name": "HTTP Sink",
 "meta.smm.predefined.flow.version": "1.0.0",
 "key.converter": "org.apache.kafka.connect.converters.ByteArrayConverter",
 "value.converter": "org.apache.kafka.connect.converters.ByteArrayConverter",
 "tasks.max": "1",
 "nexus.url": "https://repository.cloudera.com/artifactory/repo",
 "extensions.directory": "/tmp/nifi-stateless-extensions",
 "working.directory": "/tmp/nifi-stateless-working",
 "input.port": "Input from Kafka",
 "failure.ports": "Failure",
 "topics": "[***KAFKA TOPIC NAME***]",
 "parameter.HTTP Sink Parameters:Forward Raw Data": "true",
 "parameter.HTTP Sink Parameters:Remote URL": "http://[***SERVER HOSTNAME***]:[***PORT***]/[***PATH***]"
}
The following list collects the properties from the configuration example that must be customized for this use case.
topics
The name of the Kafka topic the connector fetches messages from.
Forward Raw Data
Specifies whether messages from Kafka should be forwarded as is or converted to JSON. In this example, the property is set to true, meaning that the connector does not process any records. It forwards incoming data as is.
Remote URL
Identifies the HTTP endpoint that receives the messages sent by this connector. In this example, SSL is not used. As a result, the URL starts with http not https. For example, http://my.http-server.com:22000/contentListener.