Event Brokers

Rasa Core allows you to stream events to a message broker. The event broker emits events into the event queue. It becomes part of the TrackerStore which you use when starting an Agent or launching rasa.core.run.

All events are streamed to the broker as serialized dictionaries every time the tracker updates its state. An example event emitted from the default tracker looks like this:

{
    "sender_id": "default",
    "timestamp": 1528402837.617099,
    "event": "bot",
    "text": "what your bot said",
    "data": "some data"
}

The event field takes the event’s type_name (for more on event types, check out the Events docs).

Rasa enables three possible broker types:

• Pika Event Broker
• Kafka Event Broker
• SQL Event Broker

Pika Event Broker

The example implementation uses Pika, the Python client library for RabbitMQ.

Adding a Pika Event Broker Using the Endpoint Configuration

You can use an endpoint configuration file to instruct Rasa Core to stream all events to your event broker. To do so, add the following section to your endpoint configuration, e.g. endpoints.yml:

event_broker:
  url: localhost
  username: username
  password: password
  queue: queue
  type: pika

Then instruct Rasa Core to use the endpoint configuration by adding --endpoints <path to your endpoint configuration as follows:

rasa run -m models --endpoints endpoints.yml

Adding a Pika Event Broker in Python

Here is how you add it using Python code:

from rasa.core.event_brokers.pika_producer import PikaProducer
from rasa_platform.core.tracker_store import InMemoryTrackerStore

pika_broker = PikaProducer('localhost',
                            'username',
                            'password',
                            queue='rasa_core_events')

tracker_store = InMemoryTrackerStore(db=db, event_broker=pika_broker)

Implementing a Pika Event Consumer

You need to have a RabbitMQ server running, as well as another application that consumes the events. Here’s a simple example:

import json
import pika

def _callback(self, ch, method, properties, body):
        print('Received event {}'.format(json.loads(body)))

if __name__ == '__main__':
    credentials = pika.PlainCredentials('username', 'password')
    connection = pika.BlockingConnection(
        pika.ConnectionParameters('rabbit', credentials=credentials))

channel = connection.channel()
    channel.basic_consume(_callback,
                          queue='rasa_core_events',
                          no_ack=True)
    channel.start_consuming()

Kafka Event Broker

It is possible to use Kafka as the main broker for your events. The following implementation shows an example of how to instantiate a Kafka producer:

from rasa.core.event_brokers.kafka_producer import KafkaProducer
from rasa.core.tracker_store import InMemoryTrackerStore

kafka_broker = KafkaProducer(host='localhost:9092',
                             topic='rasa_core_events')

tracker_store = InMemoryTrackerStore(event_broker=kafka_broker)

Authentication and authorization

Rasa Core’s Kafka producer accepts two types of security protocols - SASL_PLAINTEXT and SSL.

For development, use simple authentication with SASL_PLAINTEXT:

kafka_broker = KafkaProducer(host='kafka_broker:9092',
                             sasl_plain_username='kafka_username',
                             sasl_plain_password='kafka_password',
                             security_protocol='SASL_PLAINTEXT',
                             topic='rasa_core_events')

For production, use SSL protocol to secure data:

kafka_broker = KafkaProducer(host='kafka_broker:9092',
                             ssl_cafile='CARoot.pem',
                             ssl_certfile='certificate.pem',
                             ssl_keyfile='key.pem',
                             ssl_check_hostname=True,
                             security_protocol='SSL',
                             topic='rasa_core_events')

Implementing a Kafka Event Consumer

The following code shows an example for consuming events from Kafka:

from kafka import KafkaConsumer
from json import loads

consumer = KafkaConsumer('rasa_core_events',
                          bootstrap_servers=['localhost:29093'],
                          value_deserializer=lambda m: json.loads(m.decode('utf-8')))

for message in consumer:
    print(message.value)

SQL Event Broker

You can use an SQL database as an event broker. Connections are established using SQLAlchemy.

Adding a SQL Event Broker Using the Endpoint Configuration

To save all events to your SQL event broker, add a event_broker section to your endpoint configuration:

SQLite example:

event_broker:
  type: SQL
  dialect: sqlite
  db: events.db

PostgreSQL example:

event_broker:
  type: SQL
  host: 127.0.0.1
  port: 5432
  dialect: postgresql
  username: myuser
  password: mypassword
  db: mydatabase

With this configuration, Rasa will create a table called events where all events will be added.