// at this priority level the request's flow identifier (a string
  // pair) is hashed and the hash value is used to shuffle the list
  // of queues and deal a hand of the size specified here.  The
  // request is put into one of the shortest queues in that hand.
  // `handSize` must be no larger than `queues`, and should be
  // significantly smaller (so that a few heavy flows do not

  // at this priority level the request's flow identifier (a string
  // pair) is hashed and the hash value is used to shuffle the list
  // of queues and deal a hand of the size specified here.  The
  // request is put into one of the shortest queues in that hand.
  // `handSize` must be no larger than `queues`, and should be
  // significantly smaller (so that a few heavy flows do not

  KUBECONFIG_DIR="${ARTIFACTS:-$(mktemp -d)}/kubeconfig"
  IP_FAMILY="${2:-ipv4}"

  check_default_cluster_yaml

  # Trap replaces any previous trap's, so we need to explicitly cleanup clusters here
  trap cleanup_kind_clusters EXIT

  function deploy_kind() {
    IDX="${1}"
    CLUSTER_NAME="${CLUSTER_NAMES[$IDX]}"
    CLUSTER_POD_SUBNET="${CLUSTER_POD_SUBNETS[$IDX]}"

The details for the deployment & installation of this plugin were pretty much lifted directly from the
[Calico CNI plugin](https://github.com/projectcalico/cni-plugin).

Specifically:

- The CNI installation script is containerized and deployed as a daemonset in k8s.  The relevant calico k8s manifests were used as the model for the istio-cni plugin's manifest:

				return fmt.Errorf("--for must be 'delete' or 'distribution', got: %s", forFlag)
			}
			if proxyFlag != "" && threshold != 1 {
				printVerbosef(cmd, "both the proxy and threshold options were provided; the threshold option is being ignored.")
				threshold = 1
			}
			var w *watcher
			ctx, cancel := context.WithTimeout(context.Background(), timeout)
			defer cancel()
			if generation == "" {

// is passed in on stdin. Your plugin may wish to expose its functionality via
// runtime args, see CONVENTIONS.md in the CNI spec.
type Config struct {
	types.NetConf

	// Add plugin-specific flags here
	LogLevel        string     `json:"log_level"`
	LogUDSAddress   string     `json:"log_uds_address"`
	CNIEventAddress string     `json:"cni_event_address"`
	AmbientEnabled  bool       `json:"ambient_enabled"`

In the end, [after evaluation](https://docs.google.com/document/d/1c2123cKuYsBDpIon9FFdctWTUIMFweSjgwG7r8l3818/edit), the decision was to move forward with a Rust implementation.
This offered performance benefits that were too large to leave on the table, as well as opportunities to tune to our specific needs.

## Configuration protocol

  // the selector.  Failure to pick a unique label may cause this
  // and other jobs to not function correctly.  However, You may see
  // `manualSelector=true` in jobs that were created with the old `extensions/v1beta1`
  // API.
  // More info: https://kubernetes.io/docs/concepts/workloads/controllers/jobs-run-to-completion/#specifying-your-own-pod-selector
  // +optional

```

As you can see, the landscape of controllers is pretty extensive at this point.

[Service Discovery](#ServiceDiscovery) and [Config Store](#ConfigStore) were already discussed above, so do not need more explanation here.

#### Mesh Config

    #
    # The PSP controller would add corresponding annotations to the pod spec for each container. However, this happens before
    # the inject adds additional containers, so we must specify them explicitly here. With the above example, we could specify:
    # injectedAnnotations:
    #   container.apparmor.security.beta.kubernetes.io/istio-init: runtime/default