Documentation ¶
Overview ¶
The consistent hash ring from the original fnlb.
The behaviour of this depends on changes to the runner list leaving it relatively stable.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func NewPlacerTracker ¶
func NewPlacerTracker(requestCtx context.Context, cfg *PlacerConfig, call RunnerCall) *placerTracker
func RegisterPlacerViews ¶
Types ¶
type Placer ¶
type Placer interface { PlaceCall(ctx context.Context, rp RunnerPool, call RunnerCall) error GetPlacerConfig() PlacerConfig }
Placer implements a placement strategy for calls that are load-balanced across runners in a pool
func NewCHPlacer ¶
func NewCHPlacer(cfg *PlacerConfig) Placer
func NewFakeDetachedPlacer ¶
func NewFakeDetachedPlacer(cfg *PlacerConfig, st time.Duration) Placer
func NewNaivePlacer ¶
func NewNaivePlacer(cfg *PlacerConfig) Placer
type PlacerConfig ¶
type PlacerConfig struct { // After all runners in the runner list is tried, apply a delay before retrying. RetryAllDelay time.Duration `json:"retry_all_delay"` // Maximum amount of time a placer can hold a request during runner attempts PlacerTimeout time.Duration `json:"placer_timeout"` // Maximum amount of time a placer can hold an ack sync request during runner attempts DetachedPlacerTimeout time.Duration `json:"detached_placer_timeout"` }
Common config for placers.
func NewPlacerConfig ¶
func NewPlacerConfig() PlacerConfig
type Runner ¶
type Runner interface { TryExec(ctx context.Context, call RunnerCall) (bool, error) Status(ctx context.Context) (*RunnerStatus, error) Close(ctx context.Context) error Address() string }
Runner is the interface to invoke the execution of a function call on a specific runner
type RunnerCall ¶
type RunnerCall interface { SlotHashId() string Extensions() map[string]string RequestBody() io.ReadCloser ResponseWriter() http.ResponseWriter StdErr() io.ReadWriteCloser Model() *models.Call // For metrics/stats, add special accounting for time spent in customer code AddUserExecutionTime(dur time.Duration) GetUserExecutionTime() *time.Duration }
RunnerCall provides access to the necessary details of request in order for it to be processed by a RunnerPool
type RunnerPool ¶
type RunnerPool interface { // returns an error for unrecoverable errors that should not be retried Runners(ctx context.Context, call RunnerCall) ([]Runner, error) Shutdown(ctx context.Context) error }
RunnerPool is the abstraction for getting an ordered list of runners to try for a call
type RunnerStatus ¶
type RunnerStatus struct { ActiveRequestCount int32 // Number of active running requests on Runner RequestsReceived uint64 // Number of requests received by Runner RequestsHandled uint64 // Number of requests handled without NACK by Runner KdumpsOnDisk uint64 // Number of kdumps on disk StatusFailed bool // True if Status execution failed Cached bool // True if Status was provided from cache StatusId string // Call ID for Status Details string // General/Debug Log information ErrorCode int32 // If StatusFailed, then error code is set ErrorStr string // Error details if StatusFailed and ErrorCode is set CreatedAt common.DateTime // Status creation date at Runner StartedAt common.DateTime // Status execution date at Runner CompletedAt common.DateTime // Status completion date at Runner SchedulerDuration time.Duration // Amount of time runner scheduler spent on the request ExecutionDuration time.Duration // Amount of time runner spent on function execution ImagePullWaitDuration time.Duration // Amount of time spent waiting for the image pull CtrPrepDuration time.Duration // Amount of time spent preparing for the container creation CtrCreateDuration time.Duration //Amount of time spent creating the container InitStartTime time.Duration // Container Init UDS Latency time IsNetworkDisabled bool // True if network on runner is offline }
RunnerStatus is general information on Runner health as returned by Runner::Status() call
Click to show internal directories.
Click to hide internal directories.