Documentation ¶
Index ¶
Constants ¶
const ( // RequestQueueQuitPath specifies the path to send quit request to // queue-proxy. This is used for preStop hook of queue-proxy. It: // - marks the service as not ready, so that requests will no longer // be routed to it, // - adds a small delay, so that the container doesn't get killed at // the same time the pod is marked for removal. RequestQueueQuitPath = "quitquitquit" // RequestQueueHealthPath specifies the path for health checks for // queue-proxy. RequestQueueHealthPath = "health" )
const ( // ReportingPeriod interval of time for reporting. ReportingPeriod = 10 * time.Second // OperationsPerSecondN OperationsPerSecondN = "operations_per_second" // AverageConcurrentRequestsN AverageConcurrentRequestsN = "average_concurrent_requests" // LameDuckN LameDuckN = "lame_duck" // OperationsPerSecondM number of operations per second. OperationsPerSecondM Measurement = iota // AverageConcurrentRequestsM average number of requests currently being handled by this pod. AverageConcurrentRequestsM // LameDuckM indicates this Pod has received a shutdown signal. LameDuckM )
Variables ¶
This section is empty.
Functions ¶
func TimeToFirstByteTimeoutHandler ¶ added in v0.3.0
TimeToFirstByteTimeoutHandler returns a Handler that runs h with the given time limit in which the first byte of the response must be written.
The new Handler calls h.ServeHTTP to handle each request, but if a call runs for longer than its time limit, the handler responds with a 503 Service Unavailable error and the given message in its body. (If msg is empty, a suitable default message will be sent.) After such a timeout, writes by h to its ResponseWriter will return ErrHandlerTimeout.
A panic from the underlying handler is propagated as-is to be able to make use of custom panic behavior by HTTP handlers. See https://golang.org/pkg/net/http/#Handler.
The implementation is largely inspired by http.TimeoutHandler.
Types ¶
type Breaker ¶
type Breaker struct {
// contains filtered or unexported fields
}
Breaker is a component that enforces a concurrency limit on the execution of a function. It also maintains a queue of function executions in excess of the concurrency limit. Function call attempts beyond the limit of the queue are failed immediately.
func NewBreaker ¶
NewBreaker creates a Breaker with the desired queue depth, concurrency limit and initial capacity
type Channels ¶
type Channels struct { // Ticks with every request arrived/completed respectively ReqChan chan ReqEvent // Ticks with every stat report request ReportChan <-chan time.Time // Stat reporting channel StatChan chan *autoscaler.Stat }
Channels is a structure for holding the channels for driving Stats. It's just to make the NewStats signature easier to read.
type Reporter ¶ added in v0.3.0
type Reporter struct { Initialized bool // contains filtered or unexported fields }
Reporter structure representing a prometheus expoerter.
func NewStatsReporter ¶ added in v0.3.0
func NewStatsReporter(namespace string, config string, revision string, pod string) (*Reporter, error)
NewStatsReporter creates a reporter that collects and reports queue metrics
func (*Reporter) Report ¶ added in v0.3.0
func (r *Reporter) Report(lameDuck bool, operationsPerSecond float64, averageConcurrentRequests float64) error
Report captures request metrics
func (*Reporter) UnregisterViews ¶ added in v0.3.0
UnregisterViews Unregister views
type ReqEvent ¶
type ReqEvent struct { Time time.Time EventType ReqEventType }
ReqEvent represents either an incoming or closed request.
type ReqEventType ¶ added in v0.2.0
type ReqEventType int
ReqEventType denotes the type (incoming/closed) of a ReqEvent.
const ( // ReqIn represents an incoming request ReqIn ReqEventType = iota // ReqOut represents a finished request ReqOut )
type Semaphore ¶ added in v0.3.0
type Semaphore struct {
// contains filtered or unexported fields
}
Semaphore is an implementation of a semaphore based on Go channels The number of available tokens is the number of elements in the buffered channel
func NewSemaphore ¶ added in v0.3.0
NewSemaphore creates a semaphore with the desired maximal and initial capacity
func (*Semaphore) Acquire ¶ added in v0.3.0
func (s *Semaphore) Acquire()
Acquire receives the token from the semaphore, potentially blocking
func (*Semaphore) AddCapacity ¶ added in v0.3.0
AddCapacity conditionally adds capacity to the semaphore If there are tokens that must be reduced, release them first Otherwise, add tokens to the queue
func (*Semaphore) ReduceCapacity ¶ added in v0.3.0
ReduceCapacity removes tokens from the rotation It tries to acquire as many tokens as possible, if there are not enough tokens in the queue, it postpones the operation for the future by increasing the `reducers` counter