Github user EronWright commented on a diff in the pull request:

    https://github.com/apache/flink/pull/2315#discussion_r75332580
  
    --- Diff: 
flink-mesos/src/main/scala/org/apache/flink/mesos/scheduler/TaskMonitor.scala 
---
    @@ -0,0 +1,258 @@
    +/*
    + * Licensed to the Apache Software Foundation (ASF) under one
    + * or more contributor license agreements.  See the NOTICE file
    + * distributed with this work for additional information
    + * regarding copyright ownership.  The ASF licenses this file
    + * to you under the Apache License, Version 2.0 (the
    + * "License"); you may not use this file except in compliance
    + * with the License.  You may obtain a copy of the License at
    + *
    + *     http://www.apache.org/licenses/LICENSE-2.0
    + *
    + * Unless required by applicable law or agreed to in writing, software
    + * distributed under the License is distributed on an "AS IS" BASIS,
    + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    + * See the License for the specific language governing permissions and
    + * limitations under the License.
    + */
    +
    +package org.apache.flink.mesos.scheduler
    +
    +import grizzled.slf4j.Logger
    +
    +import akka.actor.{Actor, FSM, Props}
    +import org.apache.flink.configuration.Configuration
    +import org.apache.flink.mesos.scheduler.ReconciliationCoordinator.Reconcile
    +import org.apache.flink.mesos.scheduler.TaskMonitor._
    +import org.apache.flink.mesos.scheduler.messages.{Connected, Disconnected, 
StatusUpdate}
    +import org.apache.mesos.Protos.TaskState._
    +import org.apache.mesos.{SchedulerDriver, Protos}
    +
    +import scala.PartialFunction.empty
    +import scala.concurrent.duration._
    +
    +/**
    +  * Monitors a Mesos task throughout its lifecycle.
    +  *
    +  * Models a task with a state machine reflecting the perceived state of 
the task in Mesos.   The state
    +  * is primarily updated when task status information arrives from Mesos.
    +  *
    +  * The associated state data primarily tracks the task's goal (intended) 
state, as persisted by the scheduler.
    +  * Keep in mind that goal state is persisted before actions are taken.    
The goal state strictly transitions
    +  * thru New->Launched->Released.
    +  *
    +  * Unlike most exchanges with Mesos, task status is delivered 
at-least-once, so status handling should be idempotent.
    +  */
    +class TaskMonitor(
    +    flinkConfig: Configuration,
    +    schedulerDriver: SchedulerDriver,
    +    goalState: TaskGoalState) extends Actor with 
FSM[TaskMonitorState,StateData] {
    +
    +  val LOG = Logger(getClass)
    +
    +  startWith(Suspended, StateData(goalState))
    +
    +  // 
------------------------------------------------------------------------
    +  //  Suspended State
    +  // 
------------------------------------------------------------------------
    +
    +  when(Suspended) {
    +    case Event(update: TaskGoalStateUpdated, _) =>
    +      stay() using StateData(update.state)
    +    case Event(msg: StatusUpdate, _) =>
    +      stay()
    +    case Event(msg: Connected, StateData(goal: New)) =>
    +      goto(New)
    +    case Event(msg: Connected, StateData(goal: Launched)) =>
    +      goto(Reconciling)
    +    case Event(msg: Connected, StateData(goal: Released)) =>
    +      goto(Killing)
    +  }
    +
    +  // 
------------------------------------------------------------------------
    +  //  New State
    +  // 
------------------------------------------------------------------------
    +
    +  when(New) {
    +    case Event(TaskGoalStateUpdated(goal: Launched), _) =>
    +      goto(Staging) using StateData(goal)
    +  }
    +
    +  // 
------------------------------------------------------------------------
    +  //  Reconciliation State
    +  // 
------------------------------------------------------------------------
    +
    +  onTransition {
    +    case _ -> Reconciling =>
    +      nextStateData.goal match {
    +        case goal: Launched =>
    +          val taskStatus = Protos.TaskStatus.newBuilder()
    +            
.setTaskId(goal.taskID).setSlaveId(goal.slaveID).setState(TASK_STAGING).build()
    +          context.parent ! Reconcile(Seq(taskStatus))
    --- End diff --
    
    I agree that the alternative to using the implicit parent reference is to 
always use an explicit reference.     It can be unit tested either way.      I 
actually considered the latter but couldn't think of a nice name for the 
explicit reference.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---

Reply via email to