Apache Kafka

Use the Apache Kafka connector to get data from a Kafka topic into Decodable. If you are looking for information about how to create a connection to get data into Kafka from Decodable, see Apache Kafka in the Connect to a data destination chapter.

Overview

Type

source, sink

Connector name

kafka

Delivery guarantee

at least once

Compatibility

Most Kafka 2.x and 3.x-compatible broker clusters including:

Apache Kafka®

AWS Managed Service for Apache Kafka (MSK)

Confluent Platform

Confluent Cloud

Create a connection to Apache Kafka

Follow these steps to get data from Apache Kafka. These steps assume that you are using the Decodable web interface. However, if you want to use the Decodable CLI to create the connection, you can refer to the Property Name column for information about what the underlying property names are.

  1. From the Connections page, select the Apache Kafka Streaming Connector and complete the following fields.

    UI Field Property Name in the Decodable CLI Description

    Connection Type

    N/A

    Select source to use this connector to get data into Decodable.

    Bootstrap Servers

    bootstrap.servers

    A comma-separated list of your Kafka brokers. You must enter at least one broker. Enter each broker in the format: <host>:<port>. The broker list cannot contain any spaces.

    Topic

    topic

    The name of the topic to get data from.

    Broker Security Protocol

    security.protocol

    Specify the security protocol to use when connecting to a broker. Must be one of the following:

    -PLAINTEXT

    -TLS

    -SASL_SSL

    -SASL_PLAINTEXT

    See the Security section for more information about these protocols and for the additional fields that you’ll need to fill out if you want to secure your connection with TLS or SASL Authentication.

    Value Format

    value.format

    The format for data in the Kafka topic. Must be one of the following:

    - JSON: Read and write JSON data based on a JSON schema.

    - RAW: Read and write raw (byte based) values as a single column.

    - AVRO: Read and write Avro data based on an Avro schema. You can also use Avro with a schema registry. See the Avro section for more information.

    - Debezium (JSON): A unified format schema for changelogs with additional support for serializing messages using JSON. Select this option if you want to send CDC data through this connector.

    Parse Error Policy

    parse-error-policy

    Select the error handling policy. Must be one of the following:

    - FAIL: When set to FAIL, Decodable stops the connection if any validation errors are encountered in the incoming data stream.

    - IGNORE: When set to IGNORE, Decodable ignores all invalid records. All validated records are sent. With this policy, the connection is not stopped nor will any errors or warnings be shown if there is an invalid record.

    Consumer Group ID

    properties.group.id

    The group ID used by the consumer.

    This argument is only required when group-offsets has been set to scan startup mode.

    Scan Startup Mode

    scan.startup.mode

    Specifies where in the topic to start reading data when the connection is first started or when it’s restarted.

    group-offsets: Start reading data from a specified group consumer.

    earliest-offset: Start reading data from the earliest available point in the stream.

    latest-offset: Start reading data from the latest available point in the stream.

    timestamp: Start reading data from a specific timestamp.

    specific-offset: Start reading data from a specific offset value.

    Consumer Offset Reset Strategy

    properties.auto.offset.reset

    What to do when there is no initial offset in Kafka (e.g. a new consumer group) or if the specified offset does not exist any more on the server (e.g. because that data has been deleted).

  2. Select the stream that you’d like to connect to this connector. This will be the stream that receives events from Apache Kafka. Then, select Next.

  3. Define the connection’s schema. Select New Schema to manually enter the fields and field types present or Import Schema if you want to paste the schema in Avro or JSON format.

    1. The stream’s schema must match the schema of the data that you plan on sending through this connection.

    2. For more information about creating a stream or defining the stream schema, see Create and Manage Streams.

  4. Select Next when you are finished providing defining the connection’s schema.

  5. Give the newly created connection a Name and Description and select Save.

Reference

Avro

A popular choice for data serialization on Kafka is Apache Avro. Decodable supports both plain Avro, as well as Avro with a schema registry.

Plain Avro

In order to use plain Avro, the connection property value.format=avro is required. In this mode, the Avro schema is generated from the connection schema, and is available on the inferred_schema connection property. For example:

decodable connection get 69caa7e1
avro_test_source
 id 69caa7e1
 description A Kafka source connection with plain Avro schema
 connector kafka
 type source
 stream id 865e8555
 schema
 0 field1 INT NOT NULL
 1 field2 DOUBLE NOT NULL
 2 field3 STRING NOT NULL
 properties
 bootstrap.servers kafka-kafka-brokers.kafka.svc.cluster.local:9092
 format avro
 inferred_schema {"type":"record","name":"record","fields":[{"name":"field1","type":"int"},{"name":"field2","type":"double"},{"name":"field3","type":"string"}]}
 topic avro_topic
 target state RUNNING
 actual state RUNNING
 create time 2022-02-04T19:05:19Z
 update time 2022-02-08T17:04:46Z

Avro with Schema Registry

To use Avro with a schema registry, the connection property value.format=confluent-avro is required. In this mode, the Avro schema is still derived from the connection schema, but it is validated and registered against the configured schema registry.

When using this mode, the following properties are used:

Property Required? Description

avro-confluent.schema-registry.url

required

The URL of the schema registry to use.

avro-confluent.basic-auth.credentials-source

optional

The source of the authentication credentials to use, if required. Allowed values are ["USER_INFO", "SASL_INHERIT"]. Typically USER_INFO is used most often when schema registry authentication is enabled.

avro-confluent.basic-auth.user-info

optional

The authentication credentials to use, if credentials are to be used. They must be specified in the form of username:password. Note that the credentials are kept secret, once set, and will not be visible in the connection properties output.

avro-confluent.long-schema-id

optional

Whether to parse the schema ID as an integer or a long. If set to true, this will allow for reading and writing Apicurio-style payloads rather than the Confluent default. Defaults to false

For example, you can create a source Kafka connection using both SASL Authentication, as well as a secured schema registry using:

decodable connection create \
--name avro_registry_source \
--description "A Kafka source connection with SASL auth and a schema registry" \
--type source \
--stream-id 295e2a7f \
--connector kafka \
--prop format="avro-confluent" \
--prop topic="my-topic" \
--prop bootstrap.servers="some.broker.cloud:9092" \
--prop security.protocol=SASL_SSL \
--prop sasl.username="my-api-key" \
--prop sasl.password="my-secret-key" \
--prop sasl.mechanism="PLAIN" \
--prop avro-confluent.schema-registry.url="https://my.schema.registry.cloud" \
--prop avro-confluent.basic-auth.user-info="my-registry-user:my-registry-password" \
--prop avro-confluent.basic-auth.credentials-source="USER_INFO "

Extract message metadata

When consuming records from Apache Kafka in Decodable, you can access specific metadata fields associated with each Kafka record. A metadata field is identified by a string-based key and an associated data type. For example, the Apache Kafka connector has a metadata field with the key timestamp and data type TIMESTAMP_LTZ(3). See Available metadata for a list of metadata fields that you have access to.

A metadata field has its type formed as:

{datatype} METADATA [FROM '{key}']

Thus, a metadata field is indicated by the METADATA keyword, with an optional FROM to provide the key. If the key is not provided explicitly with FROM, it defaults to the name of the field.

For example:

TIMESTAMP_LTZ(3) METADATA FROM 'timestamp'

Available metadata

When you connect to Apache Kafka as a data source, you have access to the following metadata fields.

Key Data Type Description

topic

STRING NOT NULL

Topic name of the Kafka record.

partition

INT NOT NULL

Partition ID of the Kafka record.

headers

MAP<string, bytes> NOT NULL

Headers of the Kafka record as a map of raw bytes.

leader-epoch

INT NULL

Leader epoch of the Kafka record if available.

offset

BIGINT NOT NULL

Offset of the Kafka record in the partition.

timestamp

TIMESTAMP_LTZ(3) NOT NULL

Timestamp of the Kafka record.

timestamp-type

STRING NOT NULL

Timestamp type of the Kafka record. One of: NoTimestampType CreateTime (also set when writing metadata)LogAppendTime

Supported security protocols

Decodable supports four different security protocols for accessing Kafka brokers.

The Kafka connector (and Apache Kafka) actually implements TLS which supersedes the SSL protocol. For historical reasons, Kafka calls this SSL rather than TLS even though the implementation is really TLS. We do the same in order to avoid confusion for experienced Kafka users. That said, for the security-minded audience, it’s really TLS!
Security protocol Description

SASL-authenticated

Username and password authentication is used with SASL. Both SSL/TLS and PLAINTEXT encryption is supported, as well as SCRAM and PLAIN authentication mechanisms.

mTLS-authenticated

Two-way SSL authentication is used, so that Decodable and the Kafka broker(s) authenticate each other using the SSL protocol. Additionally, the connection is encrypted using SSL.

TLS-authenticated

One-way SSL authentication is used. The client (Decodable) holds the server’s (Kafka brokers) public certificate. Data from the Kafka broker(s) is encrypted using the server’s private key and Decodable can decrypt it using the public certificate. Data from Decodable is encrypted using the public certificate and can only be decrypted using the Kafka broker’s private key.

Unauthenticated

No authentication takes place between Decodable and the Kafka broker(s). The connection is not encrypted.

See the following pages for specific instructions on how to create a connection to Apache Kafka using the various security protocols:

SASL Properties

When configuring a Kafka connection to use SASL-authentication, the following connection properties are used.

UI Field Property Name in the Decodable CLI Description

SASL Mechanism

sasl.mechanism

Specify the SASL mechanism as configured by the Kafka broker. The following mechanisms are supported: - PLAIN - SCRAM-SHA-256 - SCRAM-SHA-512

SASL Username

sasl.username

The username or API key for authentication.

SASL Password

sasl.password

The secret associated with your provided API key. If you are using the Decodable CLI, this is the ID of a secret resource in your account. Run decodable secret list to view available secrets or decodable secret --help for help with creating a new secret.

Note: For security purposes, Decodable will never display secret values in plaintext. You can manage which users have permissions to create, delete, or modify secrets in the Access Control management view. See Roles, Groups, and Permissions for more information.

TLS Properties

When configuring a Kafka connection to use TLS or mTLS-authentication, the following connection properties are used.

UI Field Property Name in the Decodable CLI Description

Broker Certificate

tls.broker.certificate

The public certificate for the broker used to encrypt traffic to Decodable.

TLS Client Certificate Type

tls.client.key

The secret associated with the client TLS key used by mTLS connections. The key must be an unencrypted key in PKCS#8 format.

If you are using the Decodable CLI, this is the ID of the secret resource corresponding to the client TLS key in your account. Run decodable secret list to view available secrets or decodable secret --help for help with creating a new secret.

Note: For security purposes, Decodable will never display secret values in plaintext. You can manage which users have permissions to create, delete, or modify secrets in the Access Control management view. See Roles, Groups, and Permissions for more information.

N/A

tls.broker.signed_client_certificate

Specifies the client certificate signed by the broker.

This property is required when the tls.client.certificate.type is set to CSR. You must use the Decodable CLI to specify the client certificate. See Option 2: Using Certificate Sign Request in the How To: Set up Mutual TLS Authentication with Kafka section for more information.