Benthos is a high performance and resilient message streaming service, able to connect various sources and sinks and perform arbitrary actions, transformations and filters on payloads. It is easy to deploy and monitor, and ready to drop into your pipeline either as a static binary or a docker image. It can also be used as a framework for building your own resilient stream processors in Go.

A Benthos stream consists of four layers: inputs, optional buffer, processor workers and outputs. Inputs and outputs can be combined in a range of broker patterns. It is possible to run multiple isolated streams within a single Benthos instance using --streams mode, and perform CRUD operations on the running streams via REST endpoints.

Benthos is crash resilient by default. When connecting to at-least-once sources and sinks without a buffer it guarantees at-least-once delivery without needing to persist messages during transit.

When running a Benthos stream with a buffer there are various options for choosing a level of resiliency that meets your needs.

• AWS (S3, SQS, Kinesis)
• Elasticsearch (output only)
• File
• GCP (pub/sub)
• HDFS
• HTTP(S)
• Kafka
• MQTT
• Nanomsg
• NATS
• NATS Streaming
• NSQ
• RabbitMQ (AMQP 0.91)
• Redis (streams, list, pubsub)
• Stdin/Stdout
• Websocket
• ZMQ4

Documentation for Benthos components, concepts and recommendations can be found in the docs directory.

For building your own stream processors using Benthos as a framework check out the stream package, which also includes some examples.

For some applied examples of Benthos such as streaming and deduplicating the Twitter firehose to Kafka check out the cookbook section.

benthos -c ./config.yaml

Or, with docker:

# Send HTTP /POST data to Kafka:
docker run --rm \
	-e "INPUT_TYPE=http_server" \
	-e "OUTPUT_TYPE=kafka" \
	-e "OUTPUT_KAFKA_ADDRESSES=kafka-server:9092" \
	-e "OUTPUT_KAFKA_TOPIC=benthos_topic" \
	-p 4195:4195 \
	jeffail/benthos

# Using your own config file:
docker run --rm -v /path/to/your/config.yaml:/benthos.yaml jeffail/benthos

Benthos exposes lots of metrics either to Statsd, Prometheus or for debugging purposes an HTTP endpoint that returns a JSON formatted object.

The configuration file for a Benthos stream is made up of four main sections; input, buffer, pipeline, output. If we were to pipe stdin directly to Kafka it would look like this:

input:
  type: stdin
buffer:
  type: none
pipeline:
  threads: 1
  processors: []
output:
  type: kafka
  kafka:
    addresses:
    - localhost:9092
    topic: benthos_stream

There are also sections for setting logging, metrics and HTTP server options.

Benthos provides lots of tools for making configuration discovery and debugging easy. You can read about them here.

You can also find runnable example configs demonstrating each input, output, buffer and processor option here.

It is possible to select fields inside a configuration file to be set via environment variables. The docker image, for example, is built with a config file where all common fields can be set this way.

Grab a binary for your OS from here.

Or pull the docker image:

docker pull jeffail/benthos

Build with Go (1.11 or later):

git clone [email protected]:Jeffail/benthos
cd benthos
make

There's a multi-stage Dockerfile for creating a Benthos docker image which results in a minimal image from scratch. You can build it with:

make docker

Then use the image:

docker run --rm \
	-v /path/to/your/benthos.yaml:/config.yaml \
	-v /tmp/data:/data \
	-p 4195:4195 \
	benthos -c /config.yaml

There are a few examples here that show you some ways of setting up Benthos containers using docker-compose.

Benthos supports ZMQ4 for both data input and output. To add this you need to install libzmq4 and use the compile time flag when building Benthos:

make TAGS=ZMQ4

Or to build a docker image with ZMQ support:

make docker-zmq

Contributions are welcome, please read the guidelines.