Kafkaless installer

An ansible playbook for installing Kafka and Kubeless on a running Openshift instance.

• Kubeless version: v1.0.2
• Strimzi version: 0.11.1

Why

Kubeless is a Kubernetes-native serverless solution which uses Apache Kafka for messaging under the hood. By default Kubeless runs a Kafka instance for you, but a custom cluster can be used instead (for example, if one already exists in the Kubernetes/Openshift cluster, or if a higher level of configuration is needed). This playbook handles the setup necessary to do this, and uses Strimzi for the Kafka cluster. Strimzi provides a way to run an Apache Kafka cluster on OpenShift and Kubernetes in various deployment configurations.

Run

To start the playbook, make sure you have ansible installed, and run the following command:

ansible-playbook site.yml

Variables

Can be edited in group_vars/all.

• kubeless_namespace: The name of the Kubeless namespace
• kafka_namespace: The name of the Strimzi/Kafka namespace
• kafka_cluster_name: The name of the Strimzi/Kafka cluster

Kafka triggers

When setting up the Kubeless instance, a trigger controller is also created. This allows functions to be triggered whenever messages are published to given (trigger) topics in Kafka. To do this,

• Create a kubeless kafka trigger. This needs to associate a function with a topic that, when published to, will cause it to be invoked. For example, a hello_world function could be invoked whenever a producer publishes a message to the hello topic
• Produce a message to that topic (for testing I usually use these Go producers/consumers)
• That's it! The function should be invoked.

A guided example can be found in this blog post.