Timeouts and retries

A Retry Policy works in cooperation with the timeouts to provide fine controls to optimize the execution experience.
In Temporal, you could define retry policies for both Activity execution and Workflow execution.
Refer to Temporal documentation for more details regarding retries.

Activity retries

Let's start with some basic imports that will be required for the whole demonstration:

import zio._
import zio.temporal._
import zio.temporal.worker._
import zio.temporal.workflow._
import zio.temporal.activity._

import java.util.UUID

Consider following simple activity and workflow definitions:

@activityInterface
trait BookingActivity {
  def bookFlight(name: String, surname: String, flightNumber: String): UUID /*Booking ID*/ 
  
  def purchaseFlight(bookingId: UUID, cardId: UUID): UUID /*Booking ID*/ 
}

@workflowInterface
trait BookingWorkflow {
  @workflowMethod
  def bookFlight(name: String, surname: String, flightNumber: String, cardId: UUID): UUID /*Booking ID*/
}

When declaring activities inside the workflow implementation, it's possible to provide custom timeouts and retry policies.
They are provided using ZRetryOptions:

val retryOptions = ZRetryOptions.default
  .withMaximumAttempts(3)
  .withInitialInterval(300.millis)
  .withMaximumAttempts(5)
  .withBackoffCoefficient(1.2)
  .withDoNotRetry(nameOf[IllegalArgumentException])
// retryOptions: ZRetryOptions = ZRetryOptions(
//   maximumAttempts = Some(value = 5),
//   initialInterval = Some(value = PT0.3S),
//   backoffCoefficient = Some(value = 1.2),
//   maximumInterval = None,
//   doNotRetry = ArraySeq("java.lang.IllegalArgumentException"),
//   javaOptionsCustomization = zio.temporal.ZRetryOptions$$$Lambda$8860/0x00007f228da87748@357fd3b3
// )

Important notes:

• withStartToCloseTimeout allows to specify the maximum duration of a single Activity Task Execution
• withRetryOptions allows to specify the retry policy for an activity execution
• withMaximumAttempts limits the number of retries
• withInitialInterval, withMaximumAttempts and withBackoffCoefficient adds a backoff
• withDoNotRetry allows to specify what errors must not be retries. A helper nameOf method is used to get the full type name of the provided Exception

You can then specify retry options into ZActivityOptions when creating the Activity Stub:

class BookingWorkflowImpl extends BookingWorkflow {
  private val bookingActivity: ZActivityStub.Of[BookingActivity] = 
    ZWorkflow.newActivityStub[BookingActivity](
      ZActivityOptions
        .withStartToCloseTimeout(10.seconds)
        .withRetryOptions(retryOptions)
    )
    
  override def bookFlight(name: String, surname: String, flightNumber: String, cardId: UUID): UUID = 
    ???
}

Workflow retries

Adding retry policies for workflows is pretty the same as for activities:

ZIO.serviceWithZIO[ZWorkflowClient] { workflowClient =>
  workflowClient.newWorkflowStub[BookingWorkflow](
    ZWorkflowOptions
      .withWorkflowId("<ANY ID>")
      .withTaskQueue("booking")
      .withWorkflowExecutionTimeout(5.minutes)
      .withWorkflowRunTimeout(10.seconds)
      .withRetryOptions(
        ZRetryOptions.default.withMaximumAttempts(5)
      )
  )
}

Important notes:

• withWorkflowExecutionTimeout allows to specify the maximum time that a Workflow Execution could be in progress, including retries and Continue as New (more details here)
• withWorkflowRunTimeout allows to specify the maximum run time of a single Workflow run
• withRetryOptions allows to specify the retry policy in the same way as for activities