reboot

Documentation

Index

• Constants
• type Rebooter
  • func NewWatchdogRebooter(wd watchdog.Watchdog, log logr.Logger) Rebooter
• type SafeTimeCalculator
  • func NewAgentSafeTimeCalculator(k8sClient client.Client, wd watchdog.Watchdog, maxErrorThreshold int, ...) SafeTimeCalculator
  • func NewManagerSafeTimeCalculator(k8sClient client.Client, timeToAssumeNodeRebooted time.Duration) SafeTimeCalculator

Constants

const (
	MaxTimeForNoPeersResponse = 30 * time.Second
	MinNodesNumberInBatch     = 3
	MaxBatchesAfterFirst      = 10
)

const TimeToAssumeRebootHasStarted = time.Second * 30

Types

type Rebooter

type Rebooter interface {
	// Reboot triggers a node reboot
	Reboot() error
}

func NewWatchdogRebooter

func NewWatchdogRebooter(wd watchdog.Watchdog, log logr.Logger) Rebooter

type SafeTimeCalculator

type SafeTimeCalculator interface {
	// GetTimeToAssumeNodeRebooted returns the safe time to assume node was already rebooted
	// note that this time must include the time for a unhealthy node without api-server access to reach the conclusion that it's unhealthy
	// this should be at least worst-case time to reach a conclusion from the other peers * request context timeout + watchdog interval + maxFailuresThreshold * reconcileInterval + padding
	GetTimeToAssumeNodeRebooted() time.Duration
	SetTimeToAssumeNodeRebooted(time.Duration)
	Start(ctx context.Context) error
	//IsAgent return true in case running on an agent pod (responsible for reboot) or false in case running on a manager pod
	IsAgent() bool
}

func NewAgentSafeTimeCalculator

func NewAgentSafeTimeCalculator(k8sClient client.Client, wd watchdog.Watchdog, maxErrorThreshold int, apiCheckInterval, apiServerTimeout, peerDialTimeout, peerRequestTimeout, timeToAssumeNodeRebooted time.Duration) SafeTimeCalculator

func NewManagerSafeTimeCalculator

func NewManagerSafeTimeCalculator(k8sClient client.Client, timeToAssumeNodeRebooted time.Duration) SafeTimeCalculator