KEDA (Kubernetes Event-Driven Autoscaling)

Kamelets of type source can be augmented with KEDA metadata to automatically configure autoscalers. The additional KEDA metadata is needed for the following purposes:

  • Map Kamelet properties to corresponding KEDA parameters

  • Distinguish which KEDA parameters are needed for authentication (and need to be placed in a Secret)

  • Mark KEDA parameters as required to signal an error during reconciliation

this feature is in an experimental phase.

Basic properties to KEDA parameter mapping

Any Kamelet property can be mapped to a KEDA parameter by simply declaring the mapping in the x-descriptors list. For example:

aws-sqs-source.kamelet.yaml
apiVersion: camel.apache.org/v1
kind: Kamelet
metadata:
  name: aws-sqs-source
  labels:
    camel.apache.org/kamelet.type: "source"
spec:
  definition:
    # ...
    properties:
      queueNameOrArn:
        title: Queue Name
        description: The SQS Queue Name or ARN
        type: string
        x-descriptors:
        - urn:keda:metadata:queueURL (1)
        - urn:keda:required (2)
# ...
1 The Kamelet property queueNameOrArn corresponds to a KEDA metadata parameter named queueURL
2 The queueURL parameter is required by KEDA

In the example above, the queueNameOrArn Kamelet property is declared to correspond to a KEDA metadata parameter named queueURL, using the urn:keda:metadata: prefix. The queueURL parameter is documented in the the KEDA AWS SQS Queue scaler together with other options required by KEDA to configure an autoscaler (it can be a full queue URL or a simple queue name).

By using the marker descriptor urn:keda:required, it is also marked as required by KEDA.

The queueURL is a metadata parameter for the autoscaler. In order to configure authentication parameters, the syntax is slightly different:

aws-sqs-source.kamelet.yaml
apiVersion: camel.apache.org/v1
kind: Kamelet
metadata:
  name: aws-sqs-source
  labels:
    camel.apache.org/kamelet.type: "source"
spec:
  definition:
    # ...
    properties:
      # ...
      accessKey:
        title: Access Key
        description: The access key obtained from AWS
        type: string
        format: password
        x-descriptors:
        - urn:alm:descriptor:com.tectonic.ui:password
        - urn:camel:group:credentials
        - urn:keda:authentication:awsAccessKeyID (1)
        - urn:keda:required
# ...
1 The Kamelet property access corresponds to a KEDA authentication parameter named awsAccessKeyID

This time the property mapping uses the urn:keda:authentication: prefix, declaring it as a KEDA authentication parameter. The difference between the two approaches is that authentication parameters will be injected into a secret by the Camel K operator and linked to the KEDA ScaledObject using a TriggerAuthentication (refer to the KEDA documentation for more info).

Advanced KEDA property mapping

There are cases where KEDA requires some static values to be set in a ScaledObject or also values computed from multiple Kamelet properties. To deal with these cases it’s possible to use annotations on the Kamelet prefixed with camel.apache.org/keda.metadata. (for metadata parameters) or camel.apache.org/keda.authentication. (for authentication parameters). Those annotations can contain plain fixed values or also templates (using the Go syntax). For example:

my-source.kamelet.yaml
apiVersion: camel.apache.org/v1
kind: Kamelet
metadata:
  name: my-source
  labels:
    camel.apache.org/kamelet.type: "source"
  annotations:
    camel.apache.org/keda.authentication.sasl: "plaintext" (1)
    camel.apache.org/keda.metadata.queueLength: "5" (2)
    camel.apache.org/keda.metadata.queueAddress: "https://myhost.com/queues/{{.queueName}}" (3)
spec:
  definition:
    # ...
    properties:
      queueName:
        title: Queue Name
        description: The Queue Name
        type: string
# ...
1 An authentication parameter with a fixed value
2 A metadata parameter with a fixed value
3 A metadata parameter with a valued computed from a template

When using the template syntax, all Kamelet properties are available as fields. The default values are used in case they are missing from the user configuration. For information on how to use Kamelets with KEDA see how to run Pipes with KEDA.

Back to top