Kubernetes Autoscaler

Learn about configurations >

• autoscaler_extra_args | string

  Default: {}

  yaml or json mapping of string to use as an extra set arguments provided to cluster-autoscaler service. if an argument can be specified multiple times, for that key specify a list of values https://github.com/charmed-kubernetes/autoscaler/tree/juju/cluster-autoscaler/FAQ.md#what-are-the-parameters-to-ca example cat << EOF > args.yaml --- scale-down-unneeded-time: 5m0s # scale down at the 5 min mark v: 5 # verbose logging node-group-auto-discovery: - label:foo=bar - label:fizz=buzz EOF juju config kubernetes-autoscaler autoscaler_extra_args="$(cat args.yaml)"

• juju_api_endpoints | string

  In order to add and remove juju units, provide the api endpoints of each juju controller which contains the model/application in which to scale. By leaving empty, the application will scale according to the current juju controller's api endpoints. multiple endpoints may be provided by separating with commas example) juju_api_endpoint: '10.2.3.7:17070, 10.2.3.8:17070, 10.2.3.9:17070' juju_api_endpoint: 'my-juju-controller:17070'

• juju_ca_cert | string

  Juju controller CA certificate (base64 encoded)

• juju_default_model_uuid | string

  Juju Model UUID to use if not explicity specified by `juju_scale`.

• juju_password | string

  Juju account password associated with the controller

• juju_scale | string

  yaml or json list associated with the autoscaler node-groups define parameters about the node-group. Each item in the node-group list can be defined as the following yaml or json structure with the necessary components: min [REQUIRED] [int] # minimum number of workers within the node-group max [REQUIRED] [int] # maximum number of workers within the node-group application [REQUIRED] [str] # name of the application within the juju model to scale up or down model [OPTIONAL] [str] # name of the model within which the associated # application is deployed. If unspecified, defaults # to the value within `juju_default_model_uuid`. # if unspecified, this node-group will simply be ignored The following example shows how to create juju-scale config on the shell for two applications which have different scale requirements and differenet models. The first application defines an explicit model, the second application uses the juju_default_model_uuid examples # Simple juju config kubernetes-autoscaler juju_scale="- {min: 3, max: 5, application: kubernetes-worker}" # More complex cat << EOF > juju_scale.yaml --- - # explicit model application: kubernetes-worker-amd64 max: 10 min: 3 model: cdcaed9f-336d-47d3-83ba-d9ea9047b18c - {"min": 0, "max": 3, "application": "kubernetes-worker-amd64"} # default model EOF juju config kubernetes-autoscaler juju_scale="$(cat juju_scale.yaml)"

• juju_username | string

  Juju account username associated with the controller