p2p/discovery/backoff/backoff library

Classes

AttemptBackoff
A base class for backoff strategies that track attempt numbers
BackoffCache
Cache for peer information with backoff
BackoffConnector
A utility to connect to peers, but only if we have not recently tried connecting to them already
BackoffDiscovery
BackoffDiscovery is an implementation of discovery that caches peer data and attenuates repeated queries
BackoffStrategy
Describes how backoff will be implemented. BackoffStrategies are stateful.
Clock
Interface for getting the current time
ConnCacheData
Data stored in the connection cache
ExponentialBackoff
A backoff strategy based on an exponential function of the attempt number
ExponentialDecorrelatedJitter
A backoff strategy that uses decorrelated jitter with exponential backoff
FixedBackoff
A backoff strategy with a constant delay
LRUCache<K, V>
A simple LRU (Least Recently Used) cache implementation
PolynomialBackoff
A backoff strategy based on a polynomial function of the attempt number
RandomizedBackoff
A base class for randomized backoff strategies
RealClock
Real implementation of Clock using system time

Functions

boundedDuration(Duration d, Duration min, Duration max) → Duration
Returns a duration bounded between min and max
checkUpdates(Map<PeerId, AddrInfo> orig, Map<PeerId, AddrInfo> update) → bool
Checks if the peer addresses have changed
findPeerDispatcher(BackoffCache c, Stream<AddrInfo> peerStream) → void
Dispatches peers from a discovery query to all registered channels
findPeerReceiver(StreamController<AddrInfo> peerController, Stream<AddrInfo> evtStream, List<AddrInfo> rcvPeers) → void
Receives peers from a dispatcher and forwards them to a result channel
fullJitter(Duration duration, Duration min, Duration max, Random rng) → Duration
FullJitter returns a random number, uniformly chosen from the range min, boundedDur. boundedDur is the duration bounded between min and max.
newExponentialBackoff(Duration min, Duration max, Jitter jitter, Duration timeUnits, double base, Duration offset, Random rng) → BackoffFactory
Creates a BackoffFactory with backoff of the form base^x + offset where x is the attempt number jitter is the function for adding randomness around the backoff timeUnits are the units of time the base^x is evaluated in
newExponentialDecorrelatedJitter(Duration min, Duration max, double base, Random rng) → BackoffFactory
Creates a BackoffFactory with backoff of the roughly of the form base^x where x is the attempt number. Delays start at the minimum duration and after each attempt delay = rand(min, delay * base), bounded by the max See https://aws.amazon.com/blogs/architecture/exponential-backoff-and-jitter/ for more information
newFixedBackoff(Duration delay) → BackoffFactory
Creates a BackoffFactory with a constant backoff duration
newPolynomialBackoff(Duration min, Duration max, Jitter jitter, Duration timeUnits, List<double> polyCoefs, Random rng) → BackoffFactory
Creates a BackoffFactory with backoff of the form c0x^0, c1x^1, ...cn*x^n where x is the attempt number jitter is the function for adding randomness around the backoff timeUnits are the units of time the polynomial is evaluated in polyCoefs is the array of polynomial coefficients from c0, c1, ... cn
noJitter(Duration duration, Duration min, Duration max, Random rng) → Duration
NoJitter returns the duration bounded between min and max

Typedefs

BackoffFactory = BackoffStrategy Function()
A factory function that creates a BackoffStrategy
Jitter = Duration Function(Duration duration, Duration min, Duration max, Random rng)
Jitter implementations taken roughly from https://aws.amazon.com/blogs/architecture/exponential-backoff-and-jitter/ Jitter must return a duration between min and max. Min must be lower than, or equal to, max.