stats

Documentation

Index

• Constants
• Variables
• func NewCounters(ns, svc, p string, isTCP bool) *counters
• type LBKind
• type Stats
  • func NewStats(ctx context.Context, kind LBKind, device, statsHost, prometheusPort string, ...) (*Stats, error)
  • func (s *Stats) EnableBPFStats() error
  • func (s *Stats) UpdateConfig(c *types.ClusterConfig) error
• type WorkerStateMetrics
  • func NewWorkerStateMetrics(kind, secZone string) *WorkerStateMetrics
  • func (w *WorkerStateMetrics) ArpingFailure(err error)
  • func (w *WorkerStateMetrics) ConfigUpdate()
  • func (w *WorkerStateMetrics) IptablesWriteFailure(status int)
  • func (w *WorkerStateMetrics) LoopbackAdditionErr(errs int, addrKind string)
  • func (w *WorkerStateMetrics) LoopbackAdditions(additions int, addrKind string)
  • func (w *WorkerStateMetrics) LoopbackConfigHealthy(up int, addrKind string)
  • func (w *WorkerStateMetrics) LoopbackRemovalErr(errs int, addrKind string)
  • func (w *WorkerStateMetrics) LoopbackRemovals(removals int, addrKind string)
  • func (w *WorkerStateMetrics) LoopbackTotalDesired(totals int, addrKind string)
  • func (w *WorkerStateMetrics) NodeUpdate(outcome string)
  • func (w *WorkerStateMetrics) QueueDepth(depth int)
  • func (w *WorkerStateMetrics) Reconfigure(outcome string, d time.Duration)

Constants

const KindBGP = "bgp"

const KindDirector = "director"

const KindRealServer = "realserver"

const Prefix = "rdei_lb_"

Variables

var (
	LatencyBuckets []float64 = []float64{100, 1000, 10000, 50000, 100000, 200000, 300000, 400000, 500000, 600000, 700000, 800000, 900000, 1000000, 1500000, 2000000, 3000000}
)

consts for prometheus initialization

Functions

func NewCounters

func NewCounters(ns, svc, p string, isTCP bool) *counters

NewCounters returns an initialized counters object. A HyperLogLog++ will be intantiated as a part of this in order to keep track of the unique number of flows observed in the window period. Where this is a challenge is with respect to measuring bidirectional flows. In the case of TCP, you can simply divide the HLL cardinality count by two, because all flows are by definition bidirectional. For UDP, there doesn't seem to be a good way to track the total number of bidirectional flows. But outbound flows from the VIP address are actually not possible, so it only makes sense to track inbound flows.

Types

type LBKind

type LBKind string

type Stats

type Stats struct {
	sync.Mutex
	// contains filtered or unexported fields
}

func NewStats

func NewStats(ctx context.Context, kind LBKind, device, statsHost, prometheusPort string, freq time.Duration, logger logrus.FieldLogger) (*Stats, error)

func (*Stats) EnableBPFStats

func (s *Stats) EnableBPFStats() error

func (*Stats) UpdateConfig

func (s *Stats) UpdateConfig(c *types.ClusterConfig) error

type WorkerStateMetrics

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

func NewWorkerStateMetrics

func NewWorkerStateMetrics(kind, secZone string) *WorkerStateMetrics

func (*WorkerStateMetrics) ArpingFailure

func (w *WorkerStateMetrics) ArpingFailure(err error)

ArpingFailure switch on what type of metric we should increment

func (*WorkerStateMetrics) ConfigUpdate

func (w *WorkerStateMetrics) ConfigUpdate()

func (*WorkerStateMetrics) IptablesWriteFailure

func (w *WorkerStateMetrics) IptablesWriteFailure(status int)

func (*WorkerStateMetrics) LoopbackAdditionErr

func (w *WorkerStateMetrics) LoopbackAdditionErr(errs int, addrKind string)

func (*WorkerStateMetrics) LoopbackAdditions

func (w *WorkerStateMetrics) LoopbackAdditions(additions int, addrKind string)

func (*WorkerStateMetrics) LoopbackConfigHealthy

func (w *WorkerStateMetrics) LoopbackConfigHealthy(up int, addrKind string)

func (*WorkerStateMetrics) LoopbackRemovalErr

func (w *WorkerStateMetrics) LoopbackRemovalErr(errs int, addrKind string)

func (*WorkerStateMetrics) LoopbackRemovals

func (w *WorkerStateMetrics) LoopbackRemovals(removals int, addrKind string)

func (*WorkerStateMetrics) LoopbackTotalDesired

func (w *WorkerStateMetrics) LoopbackTotalDesired(totals int, addrKind string)

func (*WorkerStateMetrics) NodeUpdate

func (w *WorkerStateMetrics) NodeUpdate(outcome string)

func (*WorkerStateMetrics) QueueDepth

func (w *WorkerStateMetrics) QueueDepth(depth int)

QueueDepth is the depth of the configuration channel gauge config_chan_depth

func (*WorkerStateMetrics) Reconfigure

func (w *WorkerStateMetrics) Reconfigure(outcome string, d time.Duration)

Reconfigure is the end-to-end reconfiguration event. counter reconfigure_count bucket reconfigure_latency