Skip to main content
Version: main branch

Source configuration

Quickwit can insert data into an index from one or multiple sources. A source can be added after index creation using the CLI command quickwit source create. It can also be enabled or disabled with the quickwit source enable/disable subcommands.

A source is declared using an object called source config, which defines the source's settings. It consists of multiple parameters:

  • source ID
  • source type
  • source parameters
  • input_format
  • maximum number of pipelines per indexer (optional)
  • desired number of pipelines (optional)
  • transform parameters (optional)

Source ID

The source ID is a string that uniquely identifies the source within an index. It may only contain uppercase or lowercase ASCII letters, digits, hyphens (-), and underscores (_). Finally, it must start with a letter and contain at least 3 characters but no more than 255.

Source type

The source type designates the kind of source being configured. As of version 0.5, available source types are ingest-api, kafka, kinesis, and pulsar. The file type is also supported but only for local ingestion from the CLI.

Source parameters

The source parameters indicate how to connect to a data store and are specific to the source type.

File source (CLI only)

A file source reads data from a local file. The file must consist of JSON objects separated by a newline (NDJSON). As of version 0.5, a file source can only be ingested with the CLI command. Compressed files (bz2, gzip, ...) and remote files (Amazon S3, HTTP, ...) are not supported.

./quickwit tool local-ingest --input-path <INPUT_PATH>

Ingest API source

An ingest API source reads data from the Ingest API. This source is automatically created at the index creation and cannot be deleted nor disabled.

Kafka source

A Kafka source reads data from a Kafka stream. Each message in the stream must hold a JSON object.

A tutorial is available here.

Kafka source parameters

The Kafka source consumes a topic using the client library librdkafka and forwards the key-value pairs carried by the parameter client_params to the underlying librdkafka consumer. Common client_params options are bootstrap servers (bootstrap.servers), or security protocol (security.protocol). Please, refer to Kafka and librdkafka documentation pages for more advanced options.

PropertyDescriptionDefault value
topicName of the topic to consume.required
client_log_levellibrdkafka client log level. Possible values are: debug, info, warn,
client_paramslibrdkafka client configuration parameters.{}
enable_backfill_modeBackfill mode stops the source after reaching the end of the topic.false

Kafka client parameters

  • bootstrap.servers Comma-separated list of host and port pairs that are the addresses of a subset of the Kafka brokers in the Kafka cluster.

  • auto.offset.reset Defines the behavior of the source when consuming a partition for which there is no initial offset saved in the checkpoint. earliest consumes from the beginning of the partition, whereas latest (default) consumes from the end.

  • The Kafka source manages commit offsets manually using the checkpoint API and disables auto-commit.

  • Kafka-based distributed indexing relies on consumer groups. Unless overridden in the client parameters, the default group ID assigned to each consumer managed by the source is quickwit-{index_uid}-{source_id}.

  • Short max poll interval durations may cause a source to crash when back pressure from the indexer occurs. Therefore, Quickwit recommends using the default value of 300000 (5 minutes).

Adding a Kafka source to an index with the CLI

cat << EOF > source-config.yaml
version: 0.7
source_id: my-kafka-source
source_type: kafka
max_num_pipelines_per_indexer: 1
desired_num_pipelines: 2
topic: my-topic
bootstrap.servers: localhost:9092
security.protocol: SSL
./quickwit source create --index my-index --source-config source-config.yaml

Kinesis source

A Kinesis source reads data from an Amazon Kinesis stream. Each message in the stream must hold a JSON object.

A tutorial is available here.

Kinesis source parameters

The Kinesis source consumes a stream identified by a stream_name and a region.

PropertyDescriptionDefault value
stream_nameName of the stream to consume.required
regionThe AWS region of the stream. Mutually exclusive with
endpointCustom endpoint for use with AWS-compatible Kinesis service. Mutually exclusive with region.optional

If no region is specified, Quickwit will attempt to find one in multiple other locations and with the following order of precedence:

  1. Environment variables (AWS_REGION then AWS_DEFAULT_REGION)

  2. Config file, typically located at ~/.aws/config or otherwise specified by the AWS_CONFIG_FILE environment variable if set and not empty.

  3. Amazon EC2 instance metadata service determining the region of the currently running Amazon EC2 instance.

  4. Default value: us-east-1

Adding a Kinesis source to an index with the CLI

cat << EOF > source-config.yaml
version: 0.7
source_id: my-kinesis-source
source_type: kinesis
stream_name: my-stream
quickwit source create --index my-index --source-config source-config.yaml

Pulsar source

A Puslar source reads data from one or several Pulsar topics. Each message in topic(s) must hold a JSON object.

A tutorial is available here.

Pulsar source parameters

The Pulsar source consumes topics using the client library pulsar-rs.

PropertyDescriptionDefault value
topicsList of topics to consume.required
addressPulsar URL (pulsar:// and pulsar+ssl://).required
consumer_nameThe consumer name to register with the pulsar source.quickwit

Adding a Pulsar source to an index with the CLI

cat << EOF > source-config.yaml
version: 0.7
source_id: my-pulsar-source
source_type: pulsar
- my-topic
address: pulsar://localhost:6650
./quickwit source create --index my-index --source-config source-config.yaml

Maximum number of pipelines per indexer

The max_num_pipelines_per_indexer parameter is only available for sources that can be distributed: Kafka, GCP PubSub and Pulsar(coming soon).

The maximum number of indexing pipelines defines the limit of pipelines spawned for the source on a given indexer. This maximum can be reached only if there are enough desired_num_pipelines to run.


With the following parameters, only one pipeline will run on one indexer.

  • max_num_pipelines_per_indexer=2
  • desired_num_pipelines=1

Desired number of pipelines

desired_num_pipelines parameter is only available for sources that can be distributed: Kafka, GCP PubSub and Pulsar (coming soon).

The desired number of indexing pipelines defines the number of pipelines to run on a cluster for the source. It is a "desired" number as it cannot be reach it there is not enough indexers in the cluster.


With the following parameters, only one pipeline will start on the sole indexer.

  • max_num_pipelines_per_indexer=1
  • desired_num_pipelines=2

Transform parameters

For all source types but the ingest-api, ingested documents can be transformed before being indexed using Vector Remap Language (VRL) scripts.

PropertyDescriptionDefault value
scriptSource code of the VRL program executed to transform documents.required
timezoneTimezone used in the VRL program for date and time manipulations. It must be a valid name in the TZ databaseUTC
# Your source config here
# ...
script: |
.message = downcase(string!(.message))
.timestamp = now()
timezone: local

Input format

The input_format parameter specifies the expected data format of the source. Two formats are currently supported:

  • json: JSON, the default
  • plain_text: unstructured text document

Internally, Quickwit can only index JSON data. To allow the ingestion of plain text documents, Quickwit transform them on the fly into JSON objects of the following form: {"plain_text": "<original plain text document>"}. Then, they can be optionally transformed into more complex documents using a VRL script. (see transform feature).

The following is an example of how one could parse and transform a CSV dataset containing a list of users described by 3 attributes: first name, last name, and age.

# Your source config here
# ...
input_format: plain_text
script: |
user = parse_csv!(.plain_text)
.first_name = user[0]
.last_name = user[1]
.age = to_int!(user[2])

Enabling/Disabling a source from an index

A source can be enabled or disabled from an index using the CLI command quickwit source enable or quickwit source disable:

quickwit source disable --index my-index --source my-source

A source is enabled by default. When disabling a source, the related indexing pipelines will be shut down on each relevant indexer and indexing for this source will be paused.

Deleting a source from an index

A source can be removed from an index using the CLI command quickwit source delete:

quickwit source delete --index my-index --source my-source

When deleting a source, the checkpoint associated with the source is also removed.