openstack

Documentation

Index

• Constants
• type CloudProvider
  • func (provider CloudProvider) Enabled() bool
  • func (provider CloudProvider) GetName() string
  • func (provider CloudProvider) Repair(nodes []healthcheck.NodeInfo) error
  • func (provider CloudProvider) UpdateHealthStatus(masters []healthcheck.NodeInfo, workers []healthcheck.NodeInfo) error
• type OpenStackCloudProvider
• type ResourceStackRelationship

Constants

const (
	ProviderName            = "openstack"
	ClusterAutoHealingLabel = "auto_healing_enabled"
)

Types

type CloudProvider

type CloudProvider struct {
	KubeClient           kubernetes.Interface
	Nova                 *gophercloud.ServiceClient
	Heat                 *gophercloud.ServiceClient
	Magnum               *gophercloud.ServiceClient
	Cinder               *gophercloud.ServiceClient
	Config               config.Config
	ResourceStackMapping map[string]ResourceStackRelationship
}

OpenStack is an implementation of cloud provider Interface for OpenStack.

func (CloudProvider) Enabled

func (provider CloudProvider) Enabled() bool

Enabled decides if the repair should be triggered. There are two conditions that we disable the repair: - The cluster admin disables the auto healing via OpenStack API. - The Magnum cluster is not in stable status.

func (CloudProvider) GetName

func (provider CloudProvider) GetName() string

func (CloudProvider) Repair

func (provider CloudProvider) Repair(nodes []healthcheck.NodeInfo) error

Repair For master nodes: detach etcd and docker volumes, find the root

        volume, then shutdown the VM, marks the both the VM and the root
        volume (heat resource) as "unhealthy" then trigger Heat stack update
        in order to rebuild the node. The information this function needs:
        - Nova VM ID
        - Root volume ID
	       - Heat stack ID and resource ID.

For worker nodes: Call Magnum resize API directly.

func (CloudProvider) UpdateHealthStatus

func (provider CloudProvider) UpdateHealthStatus(masters []healthcheck.NodeInfo, workers []healthcheck.NodeInfo) error

UpdateHealthStatus can update the cluster health status to reflect the real-time health status of the k8s cluster.

type OpenStackCloudProvider

type OpenStackCloudProvider = CloudProvider

revive:disable:exported Deprecated: use CloudProvider instead

type ResourceStackRelationship

type ResourceStackRelationship struct {
	ResourceID   string
	ResourceName string
	StackID      string
	StackName    string
}