k8s

Documentation

Index

• type K8SManager
  • func NewK8sManager(configPath string, podEventCh chan<- *scalar.PodEvent, ...) (*K8SManager, error)
  • func (k *K8SManager) AckPodEvent(event *scalar.PodEvent)
  • func (k *K8SManager) KillPod(ctx context.Context, podID string) error
  • func (k *K8SManager) LaunchPods(ctx context.Context, pods []*models.LaunchablePod, hostname string) (launched []*models.LaunchablePod, err error)
  • func (k *K8SManager) ReconcileHosts() ([]*scalar.HostInfo, error)
  • func (k *K8SManager) Start() error
  • func (k *K8SManager) Stop()

Types

type K8SManager

type K8SManager struct {
	// contains filtered or unexported fields
}

K8SManager implements the plugin for the Kubernetes cluster manager.

func NewK8sManager

func NewK8sManager(
	configPath string,
	podEventCh chan<- *scalar.PodEvent,
	hostEventCh chan<- *scalar.HostEvent,
) (*K8SManager, error)

NewK8sManager returns a new instance of K8SManager

func (*K8SManager) AckPodEvent

func (k *K8SManager) AckPodEvent(event *scalar.PodEvent)

AckPodEvent is relevant to Mesos. For K8s for now, this is a noop. We could use some smarts here if we decide to write the resource version to DB after the event has been acknowledged by both JM and RM.

func (*K8SManager) KillPod

func (k *K8SManager) KillPod(ctx context.Context, podID string) error

KillPod stops and deletes the given pod

func (*K8SManager) LaunchPods

func (k *K8SManager) LaunchPods(
	ctx context.Context,
	pods []*models.LaunchablePod,
	hostname string,
) (launched []*models.LaunchablePod, err error)

LaunchPods creates a slice of pod objects, binds them to the node specified by hostname.

func (*K8SManager) ReconcileHosts

func (k *K8SManager) ReconcileHosts() ([]*scalar.HostInfo, error)

ReconcileHosts lists all nodes on the API server and converts them into host infos.

func (*K8SManager) Start

func (k *K8SManager) Start() error

Start starts the k8s manager plugin

func (*K8SManager) Stop

func (k *K8SManager) Stop()

Stop stops the K8SManager.