Packages

object RetryFlow

Source
RetryFlow.scala
Linear Supertypes
Ordering
  1. Alphabetic
  2. By Inheritance
Inherited
  1. RetryFlow
  2. AnyRef
  3. Any
  1. Hide All
  2. Show All
Visibility
  1. Public
  2. Protected

Value Members

  1. def withBackoff[In, Out, Mat](minBackoff: FiniteDuration, maxBackoff: FiniteDuration, randomFactor: Double, maxRetries: Int, flow: Flow[In, Out, Mat])(decideRetry: (In, Out) => Option[In]): Flow[In, Out, Mat]

    API may change!

    API may change!

    Allows retrying individual elements in the stream with an exponential backoff.

    The retry condition is controlled by the decideRetry function. It takes the originally emitted element and the response emitted by flow, and may return a request to be retried.

    The implementation of the RetryFlow requires that flow follows strict first-in-first-out and one-in-one-out semantics, i.e., the Flow may neither filter elements, nor emit more than one element per incoming element. The RetryFlow will fail if two elements are emitted for one incoming element. Any sort of batching, grouping, or filtering will make it hang forever.

    Just one element will be emitted into the flow at any time. Let's say the flow is handling an element, either first-time executing some calculation, or retrying. The next element won't be emitted into the flow until the current element has been finished processing. By finished, it means either succeed the very first attempt, succeed after a few attempts, or get dropped after using up maxRetries retries.

    minBackoff

    minimum duration to backoff between issuing retries

    maxBackoff

    maximum duration to backoff between issuing retries

    randomFactor

    adds jitter to the retry delay. Use 0 for no jitter

    maxRetries

    total number of allowed retries, when reached the last result will be emitted even if unsuccessful

    flow

    a flow to retry elements from

    decideRetry

    retry condition decision function

    Annotations
    @ApiMayChange()
  2. def withBackoffAndContext[In, CtxIn, Out, CtxOut, Mat](minBackoff: FiniteDuration, maxBackoff: FiniteDuration, randomFactor: Double, maxRetries: Int, flow: FlowWithContext[In, CtxIn, Out, CtxOut, Mat])(decideRetry: ((In, CtxIn), (Out, CtxOut)) => Option[(In, CtxIn)]): FlowWithContext[In, CtxIn, Out, CtxOut, Mat]

    API may change!

    API may change!

    Allows retrying individual elements in the stream with an exponential backoff.

    The retry condition is controlled by the decideRetry function. It takes the originally emitted element with its context, and the response emitted by flow, and may return a request to be retried.

    The implementation of the RetryFlow requires that flow follows one-in-one-out semantics, the FlowWithContext may not filter elements, nor emit more than one element per incoming element. The RetryFlow will fail if two elements are emitted from the flow, it will be stuck "forever" if nothing is emitted. Just one element will be emitted into the flow at any time. The flow needs to emit an element before the next will be emitted to it.

    The wrapped flow and decideRetry take the additional context parameters which can be a context, or used to control retrying with other information.

    minBackoff

    minimum duration to backoff between issuing retries

    maxBackoff

    maximum duration to backoff between issuing retries

    randomFactor

    adds jitter to the retry delay. Use 0 for no jitter

    maxRetries

    total number of allowed retries, when reached the last result will be emitted even if unsuccessful

    flow

    a flow with context to retry elements from

    decideRetry

    retry condition decision function

    Annotations
    @ApiMayChange()