actor.scala

object MyActor {
  val inProgress = metrics.counter("fetching-in-progress") // yammer/coda hales metrics
}
// I have around 25.000 actors of this type running
// runs on my-dispatcher
class MyActor extends Actor {
  val id = ...

  implicit val ec = context.system.dispatchers.lookup("dispatcher-for-IO-stuff")

  override def preStart = {
    context.system.scheduler.scheduleOnce(timeToInitialJob +, self, DoJob)
  }

  def receive = {
    case ...  => ... // sporadic and simple thing every few minutes
    case HealthCheck => { // every few minutes
      if (/** it has been way too long since we did something useful **/) {
        throw new Exception("It has been too long since it did some job. It'll be restarted") // actually committing suicide
      }
    }
    case DoJob => {
      inProgress += 1
      lastJobAt = new Date
      logger.debug("Received stuff")
      Future { // using dispatcher-for-IO-stuff
        // IO stuff, first fetching URLs and then searching DB
        logger.debug("Piping it back")
        SuccessfulProcessing(...) // will be piped to self
      } recover {
        case e: Exception => FailedFetch(e)
      } pipeTo self
    }
    case SuccessfulProcessing(...) => {
      ...
      inProgress -= 1
      logger.debug("Done with Job to fetch")
      context.system.scheduler.scheduleOnce(delay(), self, DoJob)
      ...
    }
    case FailedFetch(throwable)      => {
      ...
      inProgress -= 1
      logger.debug("Done with Job Failed to fetch")
      context.system.scheduler.scheduleOnce(delay(), self, DoJob)
      ...
    }
  }
}

common.conf

    my-dispatcher {
      # Dispatcher is the name of the event-based dispatcher
      type = Dispatcher
      executor = "thread-pool-executor"
      thread-pool-executor {
        core-pool-size-min = 1
        core-pool-size-factor = 2
        core-pool-size-max = 8
      }
      throughput = 5
    }
    dispatcher-for-IO-stuff {
      type = Dispatcher
      executor = "thread-pool-executor"
      thread-pool-executor {
        core-pool-size-min = 10
        core-pool-size-factor = 2.0
        core-pool-size-max = 16
      }
      throughput = 1
    }