keda_enable
Description
The keda_enable parameter installs KEDA (Kubernetes Event-Driven Autoscaling) on the rack. KEDA enables event-driven autoscaling for services, allowing them to scale based on external event sources such as message queues, databases, and custom metrics.
Default Value
The default value for keda_enable is false.
Use Cases
- Event-Driven Scaling: Scale services based on AWS SQS queue depth, Kafka lag, or other event sources.
- Scale to Zero: Reduce costs by scaling idle services to zero replicas when no events are pending.
- Custom Metrics: Autoscale on application-specific metrics from Prometheus, Datadog, or other monitoring sources.
Setting Parameters
To enable KEDA on your rack, use the following command:
$ convox rack params set keda_enable=true -r rackName
Setting parameters... OK
Additional Information
Enabling KEDA installs the KEDA operator and its components in the cluster. Once enabled, services can use the scale.keda or scale.autoscale sections in their convox.yml to define autoscaling triggers.
The Console Service > Scaling page surfaces a status card describing KEDA state on the rack. Whether Range mode (min/max replica editing) is available depends on the service having any autoscaler at all (manifest-declared OR Console-driven). Native HPA backs CPU/memory autoscale on every rack regardless of keda_enable; KEDA only gates the event-driven trigger types (GPU utilization, queue depth, custom Prometheus, SQS, etc.).
The Console Autoscale Triggers Override surface (3.24.6+) also reads this parameter to gate the KEDA-only trigger types (GPU utilization, queue depth) in the Enable dialog. CPU and memory triggers use native Kubernetes HPA and remain available on every rack regardless of keda_enable. See Autoscale Triggers Override for details.
See KEDA Autoscaling for service configuration details.
