QuestDB Kafka connector

QuestDB ships a QuestDB Kafka connector for fast ingestion from Kafka into QuestDB.

This page has the following main sections:

Integration guide#

This guide shows the steps to use the QuestDB Kafka connector to read JSON data from Kafka topics and write them as rows into a QuestDB table.

Prerequisites#

You will need the following:

  • Kafka
  • A running QuestDB instance

Configure Kafka#

Before starting Kafka, the following steps must be completed:

  1. Download the connector file.

The Apache Kafka distribution contains the Kafka Connect framework, but the QuestDB-specific components need to be downloaded from the QuestDB Kafka connector GH page, under the zip archive named kafka-questdb-connector-<version>-bin.zip.

  1. Copy the file to the Kafka libs directory.

Once downloaded, unzip the contents of the archive and copy the required .jar files to the Kafka libs directory:

unzip kafka-questdb-connector-*-bin.zip
cd kafka-questdb-connector
cp ./*.jar /path/to/kafka_2.13-2.6.0/libs
tip

You can automate downloading the latest connector package by running this command:

curl -s https://api.github.com/repos/questdb/kafka-questdb-connector/releases/latest |
jq -r '.assets[]|select(.content_type == "application/zip")|.browser_download_url'|
wget -qi -
  1. Set the configuration file.

A configuration file /path/to/kafka/config/questdb-connector.properties must be created for Kafka Connect in the standalone mode. The host and port of the running QuestDB server must be defined. A topic can be specified under the topics={mytopic} key.

The example below creates a configuration file. It assumes a running QuestDB server on the ILP default port, 9009, creates a reader from a Kafka topic, example-topic, and writes into a QuestDB table, example_table:

Create a configuration file
name=questdb-sink
connector.class=io.questdb.kafka.QuestDBSinkConnector
host=localhost:9009
topics=example-topic
table=example_table
include.key=false
value.converter=org.apache.kafka.connect.json.JsonConverter
value.converter.schemas.enable=false
key.converter=org.apache.kafka.connect.storage.StringConverter

Start Kafka#

The commands listed in this section must be run from the Kafka home directory and in the order shown below.

  1. Start the Kafka Zookeeper used to coordinate the server:
bin/zookeeper-server-start.sh config/zookeeper.properties
  1. Start a Kafka server:
bin/kafka-server-start.sh config/server.properties
  1. Start the QuestDB Kafka connector:
bin/connect-standalone.sh config/connect-standalone.properties config/questdb-connector.properties

Publish messages#

Messages can be published via the console producer script:

bin/kafka-console-producer.sh --topic example-topic --bootstrap-server localhost:9092

A greater-than symbol, >, indicates that a message can be published to the example topic from the interactive session. Paste the following minified JSON as a single line to publish the message and create the table example-topic in the QuestDB instance:

{"firstname": "Arthur", "lastname": "Dent", "age": 42}

Verify the integration#

To verify that the data has been ingested into the example-topic table, the following request to QuestDB's /exp REST API endpoint can be made to export the table contents via the curl command:

curl -G \
--data-urlencode "query=select * from 'example_table'" \
http://localhost:9000/exp

The expected response based on the example JSON message published above will be similar to the following:

"firstname","age","lastname","timestamp"
"Arthur",42,"Dent","2022-11-01T13:11:55.558108Z"

If you can see the expected result then congratulations, you have successfully created and executed your first Kafka to QuestDB pipeline! ๐ŸŽ‰

Additional sample projects#

You can find additional sample projects on the QuestDB Kafka connector Github project page. It includes a sample integration with Debezium for Change Data Capture from PostgreSQL.

Configuration manual#

This section lists configuration options as well as further information about the Kafka Connect connector.

Configuration Options#

The connector supports the following configuration options:

NameTypeExampleDefaultMeaning
topicsstringordersN/ATopics to read from
key.converterstringorg.apache.kafka.connect.storage.StringConverterN/AConverter for keys stored in Kafka
value.converterstringorg.apache.kafka.connect.json.JsonConverterN/AConverter for values stored in Kafka
hoststringlocalhost:9009N/AHost and port where QuestDB server is running
tablestringmy_tableSame as Topic nameTarget table in QuestDB
key.prefixstringfrom_keykeyPrefix for key fields
value.prefixstringfrom_valueN/APrefix for value fields
skip.unsupported.typesbooleanfalsefalseSkip unsupported types
timestamp.field.namestringpickup_timeN/ADesignated timestamp field name
include.keybooleanfalsetrueInclude message key in target table
symbolsstringinstrument,stockN/AComma separated list of columns that should be symbol type
usernamestringuser1adminUser name for QuestDB. Used only when token is non-empty
tokenstringQgHCOyq35D5HocCMrUGJinEsjEscJlCN/AToken for QuestDB authentication
tlsbooleantruefalseUse TLS for QuestDB connection

How does the connector work?#

The connector reads data from Kafka topics and writes it to QuestDB tables via ILP. The connector converts each field in the Kafka message to a column in the QuestDB table. Structures and maps are flatted into columns.

Example: Consider the following Kafka message:

{
"firstname": "John",
"lastname": "Doe",
"age": 30,
"address": {
"street": "Main Street",
"city": "New York"
}
}

The connector will create a table with the following columns:

firstname stringlastname stringage longaddress_street stringaddress_city string
JohnDoe30Main StreetNew York

Supported serialization formats#

The connector does not deserialize data independently. It relies on Kafka Connect converters. The connector has been tested predominantly with JSON, but it should work with any converter, including Avro. Converters can be configured using key.converter and value.converter options, both are included in the Configuration options table above.

Designated timestamps#

The connector supports designated timestamps. If the message contains a timestamp field, the connector can use it as a timestamp for the row. The field name must be configured using the timestamp.field.name option. The field must either be an integer or a timestamp. When the field is set to an integer the connector will interpret it as a Unix timestamp in milliseconds.

Symbol#

QuestDB supports a special type called symbol. Use the symbols configuration option to specify which columns should be created as the symbol type.

Target table considerations#

When a target table does not exist in QuestDB then it will be created automatically. This is the recommended approach for development and testing.

In production, it's recommended to use the SQL CREATE TABLE keyword, because it gives you more control over the table schema, allowing per-table partitioning, creating indexes, etc.

FAQ#

Does this connector work with Schema Registry?

The Connector works independently of the serialization strategy used. It relies on Kafka Connect converters to deserialize data. Converters can be configured using key.converter and value.converter options, see the configuration section above.

I'm getting this error: "org.apache.kafka.connect.errors.DataException: JsonConverter with schemas.enable requires 'schema' and 'payload' fields and may not contain additional fields. If you are trying to deserialize plain JSON data, set schemas.enable=false in your converter configuration."

This error means that the connector is trying to deserialize data using a converter that expects a schema. The connector does not require schemas, so you need to configure the converter to not expect a schema. For example, if you are using a JSON converter, you need to set value.converter.schemas.enable=false or key.converter.schemas.enable=false in the connector configuration.

Does this connector work with Debezium?

Yes, it's been tested with Debezium as a source and a sample project is available. Bear in mind that QuestDB is meant to be used as an append-only database; hence, updates should be translated as new inserts. The connector supports Debezium's ExtractNewRecordState transformation to extract the new state of the record. The transformation by default drops DELETE events, so there is no need to handle them explicitly.

QuestDB is a time series database, how does it fit into Change Data Capture via Debezium?

QuestDB works with Debezium just great! This is the recommended pattern: Transactional applications use a relational database to store the current state of the data. QuestDB is used to store the history of changes. Example: Imagine you have a PostgreSQL table with the most recent stock prices. Whenever a stock price changes, an application updates the PostgreSQL table. Debezium captures each UPDATE/INSERT and pushes it as an event to Kafka. Kafka Connect QuestDB connector reads the events and inserts them into QuestDB. In this way, PostgreSQL will have the most recent stock prices and QuestDB will have the history of changes. You can use QuestDB to build a dashboard with the most recent stock prices and a chart with the history of changes.

How I can select which fields to include in the target table?

Use the ReplaceField transformation to remove unwanted fields. For example, if you want to remove the address field, you can use the following configuration:

{
"name": "questdb-sink",
"config": {
"connector.class": "io.questdb.kafka.QuestDBSinkConnector",
"host": "localhost:9009",
"topics": "Orders",
"table": "orders_table",
"key.converter": "org.apache.kafka.connect.storage.StringConverter",
"value.converter": "org.apache.kafka.connect.json.JsonConverter",
"transforms": "removeAddress",
"transforms.removeAddress.type": "org.apache.kafka.connect.transforms.ReplaceField$Value",
"transforms.removeAddress.blacklist": "address"
}
}

See ReplaceField documentation for more details.

I need to run Kafka Connect on Java 8, but the connector says it requires Java 11. What should I do?

The Kafka Connect-specific part of the connectors works with Java 8. The requirement for Java 11 is coming from QuestDB client itself. The zip archive contains 2 JARs: questdb-kafka-connector-VERSION.jar and questdb-VERSION.jar. You can replace the latter with questdb-VERSION-jdk8.jar from the Maven central. Please note that this setup is not officially supported, and you may encounter issues. If you do, please report them to us.