In an ideal world, we would want a single Kapacitor setup to capture and calculate all metrics with. While it is easier to couple a Kapacitor instance with a deployment, the scale of deployments in Kubernetes makes the management of it unsustainable. For each Kapacitor instance, a port has to be designated for inbound InfluxDB subscriptions. Also, there is a parity of InfluxDB subscription per Kapacitor instance which will eventually result in memory management issues.
The kapacitor-configmap-listener (KCL) supports the ideal world scenario for running a single Kapacitor instance for the entire Kubernetes cluster workload. Assuming you are using helm packaging or something similar, a ConfigMap can be created along side the deployment with Kapacitor specified Task details. The KCL will capture the ConfigMap and translate the to Kapacitor Task. Kapacitor will then run data through these Task to create an action with. A packaged action with KCL is a TICK script used for autoscaling a ReplicaSet.