[原始碼解析] Flink的Slot究竟是什麼?(2)

羅西的思考發表於2020-09-04

[原始碼解析] Flink 的slot究竟是什麼?(2)

0x00 摘要

Flink的Slot概念大家應該都聽說過,但是可能很多朋友還不甚瞭解其中細節,比如具體Slot究竟代表什麼?在程式碼中如何實現?Slot在生成執行圖、排程、分配資源、部署、執行階段分別起到什麼作用?本文和上文將帶領大家一起分析原始碼,為你揭開Slot背後的機理。

0x01 前文回顧

書接上回 [原始碼解析] Flink 的slot究竟是什麼?(1)。前文中我們已經從系統架構和資料結構角度來分析了Slot,本文我們將從業務流程角度來分析Slot。我們重新放出系統架構圖

和資料結構邏輯關係圖

下面我們從幾個流程入手一一分析。

0x02 註冊/更新Slot

有兩個途徑會註冊Slot/更新Slot狀態。

  • 當TaskExecutor註冊成功之後會和RM互動進行註冊時,一併註冊Slot;
  • 定時心跳時,會在心跳payload中附加Slot狀態資訊;

2.1 TaskExecutor註冊成功

當TaskExecutor註冊成功之後會和RM互動進行註冊。會通過如下的程式碼呼叫路徑來向ResourceManager(SlotManagerImpl)註冊Slot。SlotManagerImpl 在獲取訊息之後,會更新Slot狀態,如果此時已經有如果有pendingSlotRequest,就直接分配,否則就更新freeSlots變數。

  • TaskExecutor#establishResourceManagerConnection;

  • TaskSlotTableImpl#createSlotReport;建立 report

    • 這時候的 report如下:

      slotReport = {SlotReport@9633} 
      
        0 = {SlotStatus@8969} "SlotStatus{slotID=40d390ec-7d52-4f34-af86-d06bb515cc48_0, resourceProfile=ResourceProfile{managedMemory=64.000mb (67108864 bytes), networkMemory=32.000mb (33554432 bytes)}, allocationID=null, jobID=null}"
         slotID = {SlotID@8629} "40d390ec-7d52-4f34-af86-d06bb515cc48_0"
         resourceProfile = {ResourceProfile@4194} "ResourceProfile{managedMemory=64.000mb (67108864 bytes), networkMemory=32.000mb (33554432 bytes)}"
         allocationID = null
         jobID = null
          
        1 = {SlotStatus@9638} "SlotStatus{slotID=40d390ec-7d52-4f34-af86-d06bb515cc48_1, resourceProfile=ResourceProfile{managedMemory=64.000mb (67108864 bytes), networkMemory=32.000mb (33554432 bytes)}, allocationID=null, jobID=null}"
         slotID = {SlotID@9643} "40d390ec-7d52-4f34-af86-d06bb515cc48_1"
         resourceProfile = {ResourceProfile@4194} "ResourceProfile{managedMemory=64.000mb (67108864 bytes), networkMemory=32.000mb (33554432 bytes)}"
         allocationID = null
         jobID = null
      
  • ResourceManager#sendSlotReport;通過RPC(resourceManagerGateway.sendSlotReport)呼叫到RM

  • SlotManagerImpl#registerTaskManager;把TaskManager註冊到SlotManager

  • SlotManagerImpl#registerSlot;

  • SlotManagerImpl#createAndRegisterTaskManagerSlot;生成註冊了TaskManagerSlot

    • 這時候程式碼 & 變數如下,我們可以看到,就是把TM的Slot資訊註冊到SlotManager中

      private TaskManagerSlot createAndRegisterTaskManagerSlot(SlotID slotId, ResourceProfile resourceProfile, TaskExecutorConnection taskManagerConnection) {
         final TaskManagerSlot slot = new TaskManagerSlot(
                                          slotId, resourceProfile, taskManagerConnection);
         slots.put(slotId, slot);
         return slot;
      }
      
      slot = {TaskManagerSlot@13322} 
       slotId = {SlotID@8629} "40d390ec-7d52-4f34-af86-d06bb515cc48_0"
       resourceProfile = {ResourceProfile@4194} 
        cpuCores = {CPUResource@11616} "Resource(CPU: 89884656743115785...0)"
        taskHeapMemory = {MemorySize@11617} "4611686018427387903 bytes"
        taskOffHeapMemory = {MemorySize@11618} "4611686018427387903 bytes"
        managedMemory = {MemorySize@11619} "64 mb"
        networkMemory = {MemorySize@11620} "32 mb"
        extendedResources = {HashMap@11621}  size = 0
       taskManagerConnection = {WorkerRegistration@11121} 
       allocationId = null
       jobId = null
       assignedSlotRequest = null
       state = {TaskManagerSlot$State@13328} "FREE"
      
  • SlotManagerImpl#updateSlot

  • SlotManagerImpl#updateSlotState;如果有pendingSlotRequest,就直接分配

  • SlotManagerImpl#handleFreeSlot;否則就更新freeSlots變數

流程結束後,SlotManager如下,可以看到此時slots個數是兩個,freeSlots也是兩個,說明都是空閒的:

this = {SlotManagerImpl@11120} 
 scheduledExecutor = {ActorSystemScheduledExecutorAdapter@11125} 
 slotRequestTimeout = {Time@11127} "300000 ms"
 taskManagerTimeout = {Time@11128} "30000 ms"
 slots = {HashMap@11122}  size = 2
  {SlotID@9643} "40d390ec-7d52-4f34-af86-d06bb515cc48_1" -> {TaskManagerSlot@19206} 
  {SlotID@8629} "40d390ec-7d52-4f34-af86-d06bb515cc48_0" -> {TaskManagerSlot@13322} 
 freeSlots = {LinkedHashMap@11129}  size = 2
  {SlotID@8629} "40d390ec-7d52-4f34-af86-d06bb515cc48_0" -> {TaskManagerSlot@13322} 
  {SlotID@9643} "40d390ec-7d52-4f34-af86-d06bb515cc48_1" -> {TaskManagerSlot@19206} 
 taskManagerRegistrations = {HashMap@11130}  size = 1
 fulfilledSlotRequests = {HashMap@11131}  size = 0
 pendingSlotRequests = {HashMap@11132}  size = 0
 pendingSlots = {HashMap@11133}  size = 0
 slotMatchingStrategy = {AnyMatchingSlotMatchingStrategy@11134} "INSTANCE"
 slotRequestTimeoutCheck = {ActorSystemScheduledExecutorAdapter$ScheduledFutureTask@11139} 

2.2 心跳機制更新Slot狀態

Flink的心跳機制也會被利用來進行Slots資訊的彙報,Slot Report被包括在心跳payload中。

首先在 TE 中建立Slot Report

  • TaskExecutor#heartbeatFromResourceManager
  • HeartbeatManagerImpl#requestHeartbeat
  • TaskExecutor$ResourceManagerHeartbeatListener # retrievePayload
  • TaskSlotTableImpl # createSlotReport

程式執行到 RM,於是 SlotManagerImpl 呼叫到 reportSlotStatus,進行Slot狀態更新。

  • ResourceManager#heartbeatFromTaskManager

  • HeartbeatManagerImpl#receiveHeartbeat

  • ResourceManager$TaskManagerHeartbeatListener#reportPayload

  • SlotManagerImpl#reportSlotStatus,此時的SlotReport如下:

    • slotReport = {SlotReport@8718} 
       slotsStatus = {ArrayList@8717}  size = 2
        0 = {SlotStatus@9025} "SlotStatus{slotID=d99e16d7-a30c-4e21-b270-f82884b1813f_0, resourceProfile=ResourceProfile{managedMemory=64.000mb (67108864 bytes), networkMemory=32.000mb (33554432 bytes)}, allocationID=null, jobID=null}"
         slotID = {SlotID@9032} "d99e16d7-a30c-4e21-b270-f82884b1813f_0"
         resourceProfile = {ResourceProfile@4194} "ResourceProfile{managedMemory=64.000mb (67108864 bytes), networkMemory=32.000mb (33554432 bytes)}"
         allocationID = null
         jobID = null
        1 = {SlotStatus@9026} "SlotStatus{slotID=d99e16d7-a30c-4e21-b270-f82884b1813f_1, resourceProfile=ResourceProfile{managedMemory=64.000mb (67108864 bytes), networkMemory=32.000mb (33554432 bytes)}, allocationID=null, jobID=null}"
         slotID = {SlotID@9029} "d99e16d7-a30c-4e21-b270-f82884b1813f_1"
         resourceProfile = {ResourceProfile@4194} "ResourceProfile{managedMemory=64.000mb (67108864 bytes), networkMemory=32.000mb (33554432 bytes)}"
         allocationID = null
         jobID = null
      
  • SlotManagerImpl#updateSlot

  • SlotManagerImpl#updateSlotState;如果有pendingSlotRequest,就直接分配

  • SlotManagerImpl#handleFreeSlot;否則就更新freeSlots變數

    • freeSlots.put(freeSlot.getSlotId(), freeSlot);
      

0x03 生成ExecutionGraph階段

當Job提交之後,經過一系列處理,Scheduler會建立ExecutionGraph。ExecutionGraph 是 JobGraph 的並行版本。而通過一系列的分析,才可以最終把任務分發到相關的任務槽中。槽會根據CPU的數量提前指定出來,這樣可以最大限度的利用CPU的計算資源。如果Slot耗盡,也就意味著新分發的作業任務是無法執行的。

ExecutionGraphJobManager根據JobGraph生成的分散式執行圖,是排程層最核心的資料結構。

一個JobVertex / ExecutionJobVertex代表的是一個operator,而具體的ExecutionVertex則代表了一個Task。

在生成StreamGraph時候,StreamGraph.addOperator方法就已經確定了operator是什麼型別,比如OneInputStreamTask,或者SourceStreamTask等。

假設OneInputStreamTask.class即為生成的StreamNode的vertexClass。這個值會一直傳遞,當StreamGraph被轉化成JobGraph的時候,這個值會被傳遞到JobVertex的invokableClass。然後當JobGraph被轉成ExecutionGraph的時候,這個值被傳入到ExecutionJobVertex.TaskInformation.invokableClassName中,最後一直傳到Task中。

本系列程式碼執行序列如下:

  • JobMaster#createScheduler

  • DefaultSchedulerFactory#createInstance

  • DefaultScheduler#init

  • SchedulerBase#init

  • SchedulerBase#createAndRestoreExecutionGraph

  • SchedulerBase#createExecutionGraph

  • ExecutionGraphBuilder#buildGraph

  • ExecutionGraph#attachJobGraph

  • ExecutionJobVertex#init,這裡根據並行度來確定要建立多少個Task,即多少個ExecutionVertex。

    • int numTaskVertices = vertexParallelism > 0 ? vertexParallelism : defaultParallelism;
      this.taskVertices = new ExecutionVertex[numTaskVertices];
      
  • ExecutionVertex#init,這裡會生成Execution。

    • this.currentExecution = new Execution( 
      			getExecutionGraph().getFutureExecutor(),
      			this,	0, initialGlobalModVersion,	createTimestamp, timeout);
      

0x04 排程階段

任務的流程就是通過作業分發到TaskManager,然後再分發到指定的Slot進行執行。

這部分排程階段的程式碼只是利用CompletableFuture把程式執行架構搭建起來,可以把認為是自頂之下進行操作

Job開始排程之後,程式碼執行序列如下:

  • JobMaster#startJobExecution

  • JobMaster#resetAndStartScheduler

  • Future操作

  • JobMaster#startScheduling

  • SchedulerBase#startScheduling

  • DefaultScheduler#startSchedulingInternal

  • LazyFromSourcesSchedulingStrategy#startScheduling,這裡開始針對Vertices進行資源分配和部署

    • allocateSlotsAndDeployExecutionVertices(schedulingTopology.getVertices());
      
  • LazyFromSourcesSchedulingStrategy#allocateSlotsAndDeployExecutionVertices,這裡會遍歷ExecutionVertex,篩選出Create狀態的 & 輸入Ready的節點。

    • private void allocateSlotsAndDeployExecutionVertices(
            final Iterable<? extends SchedulingExecutionVertex<?, ?>> vertices) {
         // 取出狀態是CREATED,且輸入Ready的 ExecutionVertex
      	 final Set<ExecutionVertexID> verticesToDeploy = IterableUtils.toStream(vertices)
      			.filter(IS_IN_CREATED_EXECUTION_STATE.and(isInputConstraintSatisfied()))
      			.map(SchedulingExecutionVertex::getId)
      			.collect(Collectors.toSet());
         // 根據 ExecutionVertex 建立 DeploymentOption
         final List<ExecutionVertexDeploymentOption> vertexDeploymentOptions = ...;
         // 分配資源並且部署
         schedulerOperations.allocateSlotsAndDeploy(vertexDeploymentOptions);
      }
      
  • DefaultScheduler#allocateSlotsAndDeploy

這裡來到了本文第一個關鍵函式 allocateSlotsAndDeploy。其主要功能是:

  1. allocateSlots分配Slot,其實這時候並沒有分配,而是建立一系列Future,然後根據Future返回SlotExecutionVertexAssignment列表。
  2. 根據SlotExecutionVertexAssignment建立DeploymentHandle
  3. 根據deploymentHandles進行部署,其實是根據Future把部署搭建起來,具體如何部署需要在slot分配成功之後再執行。
@Override
public void allocateSlotsAndDeploy(final List<ExecutionVertexDeploymentOption> executionVertexDeploymentOptions) {
   validateDeploymentOptions(executionVertexDeploymentOptions);

   final Map<ExecutionVertexID, ExecutionVertexDeploymentOption> deploymentOptionsByVertex =
      groupDeploymentOptionsByVertexId(executionVertexDeploymentOptions);

   final List<ExecutionVertexID> verticesToDeploy = executionVertexDeploymentOptions.stream()
      .map(ExecutionVertexDeploymentOption::getExecutionVertexId)
      .collect(Collectors.toList());

   final Map<ExecutionVertexID, ExecutionVertexVersion> requiredVersionByVertex =
      executionVertexVersioner.recordVertexModifications(verticesToDeploy);

   transitionToScheduled(verticesToDeploy);

   // 分配Slot,其實這時候並沒有分配,而是建立一系列Future,然後根據Future返回SlotExecutionVertexAssignment列表
   final List<SlotExecutionVertexAssignment> slotExecutionVertexAssignments =
      allocateSlots(executionVertexDeploymentOptions);

   // 根據SlotExecutionVertexAssignment建立DeploymentHandle
   final List<DeploymentHandle> deploymentHandles = createDeploymentHandles(
      requiredVersionByVertex,
      deploymentOptionsByVertex,
      slotExecutionVertexAssignments);
  
   // 根據deploymentHandles進行部署,其實是根據Future把部署搭建起來,具體如何部署需要在slot分配成功之後再執行
   if (isDeployIndividually()) {
      deployIndividually(deploymentHandles);
   } else {
      waitForAllSlotsAndDeploy(deploymentHandles);
   }
}

接下來 兩個小章節我們分別針對 allocateSlots 和 deployIndividually / waitForAllSlotsAndDeploy 進行分析。

0x05 分配資源階段

注意,此處的入口為 allocateSlotsAndDeploy 的allocateSlots 呼叫

在分配slot時,首先會在JobMaster中SlotPool中進行分配,具體是先SlotPool中獲取所有slot,然後嘗試選擇一個最合適的slot進行分配,這裡的選擇有兩種策略,即按照位置優先和按照之前已分配的slot優先;若從SlotPool無法分配,則通過RPC請求向ResourceManager請求slot,若此時並未連線上ResourceManager,則會將請求快取起來,待連線上ResourceManager後再申請。

5.1 CompletableFuture

CompletableFuture 首先是一個 Future,它擁有 Future 所有的功能,包括取得非同步執行結果,取消正在執行的任務等,其次是 一個CompleteStage,其最大作用是將回撥改為鏈式呼叫,從而將 Future 組合起來。

此處生成了執行框架,即通過三個 CompletableFuture 構成了執行框架

我們按照出現順序命名為 Future 1,Future 2,Future 3。

但是這個反過來說明反而更方便。我們可以看到,'

出現次序是 Future 1,Future 2,Future 3

呼叫順序是 Future 3 ---> Future 2 ---> Future 1

5.1.1 Future 3

我們可以稱之為 PhysicalSlot Future

型別是:CompletableFuture

生成在:requestNewAllocatedSlot 函式中對 PendingRequest 的生成。PendingRequest 的建構函式中有 new CompletableFuture<>(),這個 Future 3 是 PendingRequest 的成員變數。

用處是:

  • PendingRequest 會 加入到 waitingForResourceManager

回撥函式作用是:

  • 在 allocateMultiTaskSlot 的 whenComplete 會把payload賦值給slot,allocatedSlot.tryAssignPayload
  • 進一步回撥在 createRootSlot 函式 的 forward . thenApply 語句,會 設定為 Future 3 回撥 Future 2 的回撥函式

何時回撥

  • TM,TE offer Slot的時候,會根據 PendingRequest 間接回撥到這裡

6.1.2 Future 2

我們可以稱之為 allocationFuture

型別是:

  • CompletableFuture ,CompletableFuture 有型別轉換

生成在:

  • createRootSlot函式中。final CompletableFuture slotContextFutureAfterRootSlotResolution = new CompletableFuture<>();

用處是:

  • 把 Future 2 設定為 multiTaskSlot 的成員變數 private final CompletableFuture<? extends SlotContext> slotContextFuture;
  • Future 2 其實也就是 SingleTaskSlot 的 parent.getSlotContextFuture(),因為 multiTaskSlot 和 SingleTaskSlot 是父子關係
  • 在 SingleTaskSlot 建構函式 中,Future 2 會賦值給 SingleTaskSlot 的成員變數 singleLogicalSlotFuture。
  • 即 Future 2 實際上是 SingleTaskSlot 的成員變數 singleLogicalSlotFuture
  • SchedulerImpl # allocateSharedSlot 函式,return leaf.getLogicalSlotFuture(); 會被返回 singleLogicalSlotFuture 給外層呼叫,就是外層看到的 allocationFuture。

回撥函式作用是:

  • 在 SingleTaskSlot 建構函式 中,會生成一個 SingleLogicalSlot(未來回撥時候會真正生成 )
  • 在 internalAllocateSlot 函式中,會回撥 Future 1,allocationResultFuture的回撥函式

何時回撥

  • 被 Future 3 的回撥函式呼叫

6.1.3 Future 1

我們可以稱之為 allocationResultFuture

型別是:

  • CompletableFuture

生成在:

  • SchedulerImpl#allocateSlotInternal,這裡生成了第一個 CompletableFuture

用處是:

  • 後續 Deploy 時候會用到 這個 Future 1,會通過 handle 給 Future 1 再加上兩個後續呼叫,是在 Future 1 結束之後的後續呼叫。

回撥函式作用是:

  • allocateSlotsFor 函式中有錯誤處理
  • 後續 Deploy 時候會用到 這個 Future 1,會通過 handle 給 Future 1 再加上兩個後續呼叫,是在 Future 1 結束之後的後續呼叫。

何時回撥

  • 語句在internalAllocateSlot中,但是在 Future 2 回撥函式中呼叫

5.2 流程圖

這裡比較複雜,先給出流程圖

 *  Run in Job Manager
 *
 *    DefaultScheduler#allocateSlotsAndDeploy 
 *        |
 *        +----> DefaultScheduler#allocateSlots
 *        |     //把ExecutionVertex轉化為ExecutionVertexSchedulingRequirements
 *        |     
 *        +----> DefaultExecutionSlotAllocator#allocateSlotsFor( 呼叫 1 開始 )
 *        |     // 得到 我們的第一個 CompletableFuture,我們稱之為 Future 1
 *        |  
 *        |    
 *        +--------------> NormalSlotProviderStrategy#allocateSlot 
 *        |    
 *        |        
 *        +--------------> SchedulerImpl#allocateSlotInternal
 *        |     // 生成了第一個 CompletableFuture,以後稱之為 allocationResultFuture
 *        |  
 *  ┌────────────┐   
 *  │  Future 1  │ 生成 allocationResultFuture
 *  └────────────┘ 
 *        │     
 *        │            
 *        +----> SchedulerImpl#internalAllocateSlot( 呼叫 2 開始 )  
 *        |      // Future 1 做為引數被傳進來,這裡會繼續呼叫,生成 Future 2, Future 3
 *        |     
 *        |       
 *        +-----------> SchedulerImpl#allocateSharedSlot( 呼叫 3 開始 )
 *        |    	// 這裡涉及到 MultiTaskSlot 和 SingleTaskSlot
 *        |        
 *        +-----------> SchedulerImpl # allocateMultiTaskSlot ( 呼叫 4 開始 )
 *        |  
 *        |       
 *        +--------------------> SchedulerImpl # requestNewAllocatedSlot
 *        |    
 *        |    
 *        +--------------------> SlotPoolImpl#requestNewAllocatedSlot
 *        |    	// 這裡生成一個 PendingRequest
 *        | 	// PendingRequest的建構函式中有 new CompletableFuture<>(),
 *        |     // 所以這裡是生成了第三個 Future,注意這裡的 Future 是針對 PhysicalSlot  
 *        |  
 *        |  
 *  ┌────────────┐   
 *  │  Future 3  │ 生成 Future<PhysicalSlot>,這個 Future 3 實際是對使用者不可見的。
 *  └────────────┘      
 *        |  
 *        | 
 *        +-----------> SchedulerImpl # allocateMultiTaskSlot( 呼叫 4 結束 )
 *        |      // 回到 ( 呼叫 4 ) 這裡,得倒 Future 3
 *        |      // 這裡得倒了第三個 Future<PhysicalSlot> 
 *        |      // 第三是因為從使用者角度看,它是第三個出現的     
 *        |     
 *        +-----------------------> slotSharingManager # createRootSlot  
 *        |      // 把 Future 3 做為引數傳進去     
 *        |      // 這裡馬上生成 Future 2
 *        |      // Future 2 被設定為 multiTaskSlot 的成員變數 slotContextFuture;  
 *        |      // 然後forward . thenApply 語句 會 設定為 Future 3 回撥 Future 2 的回撥函式
 *        |     
 *        |       
 *        +-----------> SchedulerImpl#allocateSharedSlot
 *        |    	// 回到 ( 呼叫 3 ) 這裡   
 *        |     
 *        | 
 *        +-----------------------> SlotSharingManager#allocateSingleTaskSlo  
 *        |  // 在 rootMultiTaskSlot 之上生成一個 SingleTaskSlot leaf加入到allTaskSlots。    
 *        |  // leaf.getLogicalSlotFuture(); 這個就是Future 2,設定好的
 *        |   
 *        |        
 *        +-----------> SchedulerImpl#allocateSharedSlot
 *        |    	// 還在 ( 呼叫 3 ) 這裡  
 *        |     // return leaf.getLogicalSlotFuture(); 返回 Future 2   
 *        | 
 *        |       
 *  ┌────────────┐   
 *  │  Future 2  │
 *  └────────────┘  
 *        |     
 *        |       
 *        |       
 *        +----> SchedulerImpl#internalAllocateSlot    
 *        |      // 回到 ( 呼叫 2 ) 這裡 
 *        |      // 設定,在 Future 2 的回撥函式中會呼叫 Future 1    
 *        |  
 *        |      
 *        +----> DefaultExecutionSlotAllocator#allocateSlotsFor 
 *        |     // 回到 ( 呼叫 1 ) 這裡 
 *        |
 *        |  
 *        |       
 *  ┌────────────┐   
 *  │  Future 1  │ 
 *  └────────────┘    
 *        |  
 *        |      
 *        +---->  createDeploymentHandles    
 *        |  // 生成 DeploymentHandle
 *        |     
 *        |       
 *        +-----------> deployIndividually(deploymentHandles);    
 *        |           // 這裡會給 Future 1 再加上兩個 回撥函式,作為 部署回撥
 *        | 

下圖是為了手機閱讀。

5.3 具體執行路徑

預設情況下,Flink 允許subtasks共享slot,條件是它們都來自同一個Job的不同task的subtask。結果可能一個slot持有該job的整個pipeline。允許slot共享有以下兩點好處:

  • Flink 叢集所需的task slots數與job中最高的並行度一致。也就是說我們不需要再去計算一個程式總共會起多少個task了。
  • 更容易獲得更充分的資源利用。如果沒有slot共享,那麼非密集型操作source/flatmap就會佔用同密集型操作 keyAggregation/sink 一樣多的資源。如果有slot共享,將基線的2個並行度增加到6個,能充分利用slot資源,同時保證每個TaskManager能平均分配到重的subtasks。

此處執行路徑大致如下:

  • DefaultScheduler#allocateSlotsAndDeploy

  • DefaultScheduler#allocateSlots;該過程會把ExecutionVertex轉化為ExecutionVertexSchedulingRequirements,會封裝包含一些location資訊、sharing資訊、資源資訊等

  • DefaultExecutionSlotAllocator#allocateSlotsFor;我們小節實際是從這裡開始分析,這裡會進行一系列操作,一層層呼叫下去。首先這個函式會得到我們的第一個 CompletableFuture,我們稱之為 allocationResultFuture,這個名字的由來後續就會知道。這個 slotFuture 會賦值給 SlotExecutionVertexAssignment,然後傳遞給外面。後續 Deploy 時候會用到 這個 slotFuture,會通過 handle 給 slotFuture 再加上兩個後續呼叫,是在slotFuture結束之後的後續呼叫。

    • public List<SlotExecutionVertexAssignment> allocateSlotsFor(...) {
      		for (ExecutionVertexSchedulingRequirements schedulingRequirements : executionVertexSchedulingRequirements) {
            
            // 得到第一個 CompletableFuture,具體是在 calculatePreferredLocations 中通過 
      			CompletableFuture<LogicalSlot> slotFuture = 
              calculatePreferredLocations(...).thenCompose(...) ->
      								slotProviderStrategy.allocateSlot( // 函式裡面生成了第一個CompletableFuture
      									slotRequestId,
      									new ScheduledUnit(...),
      									SlotProfile.priorAllocation(...)));
      
      			SlotExecutionVertexAssignment slotExecutionVertexAssignment =
      					new SlotExecutionVertexAssignment(executionVertexId, slotFuture);
      
      			slotFuture.whenComplete(
      					(ignored, throwable) -> { // 第一個CompletableFuture的回撥函式,裡其實只是異常處理,後續有人會呼叫到這裡
      						pendingSlotAssignments.remove(executionVertexId);
      						if (throwable != null) {
      							slotProviderStrategy.cancelSlotRequest(slotRequestId, slotSharingGroupId, throwable);
      						}
      					});
      
      			slotExecutionVertexAssignments.add(slotExecutionVertexAssignment);
      		}
      
      		return slotExecutionVertexAssignments;
      }
      
      
  • NormalSlotProviderStrategy#allocateSlot(slotProviderStrategy.allocateSlot)

  • SchedulerImpl#allocateSlotInternal,這裡生成了第一個 CompletableFuture,我們可以稱之為 allocationResultFuture

    • private CompletableFuture<LogicalSlot> allocateSlotInternal(...) {
          // 這裡生成了第一個 CompletableFuture,我們以後稱之為 allocationResultFuture
      		final CompletableFuture<LogicalSlot> allocationResultFuture = new CompletableFuture<>();
          // allocationResultFuture 會傳送進去繼續處理
      		internalAllocateSlot(allocationResultFuture, slotRequestId, scheduledUnit, 
                               slotProfile, allocationTimeout);
          // 返回 allocationResultFuture
      		return allocationResultFuture;
      }
      
      
  • SchedulerImpl#allocateSlot

  • SchedulerImpl#internalAllocateSlot,該方法會根據vertex是否共享slot來分配singleSlot/SharedSlot。這裡得到第二個 CompletableFuture,我們以後成為 allocationFuture

    • private void internalAllocateSlot(
      			CompletableFuture<LogicalSlot> allocationResultFuture, ...) {
        	// 這裡得到第二個 CompletableFuture,我們以後稱為 allocationFuture,注意目前只是得到,不是生成。
      		CompletableFuture<LogicalSlot> allocationFuture = scheduledUnit.getSlotSharingGroupId() == null ?
      			allocateSingleSlot(slotRequestId, slotProfile, allocationTimeout) :
      			allocateSharedSlot(slotRequestId, scheduledUnit, slotProfile, allocationTimeout);
      		// 第二個Future,allocationFuture的回撥函式。注意,CompletableFuture可以連續呼叫多個whenComplete。
      		allocationFuture.whenComplete((LogicalSlot slot, Throwable failure) -> {
      			if (failure != null) { // 異常處理
      				cancelSlotRequest(...);
      				allocationResultFuture.completeExceptionally(failure);
      			} else {
      				allocationResultFuture.complete(slot); // 它將回撥第一個 allocationResultFuture的回撥函式
      			}
      		});
      }
      
      
  • SchedulerImpl#allocateSharedSlot,這裡也比較複雜,涉及到 MultiTaskSlot 和 SingleTaskSlot

    • private CompletableFuture<LogicalSlot> allocateSharedSlot(...) {
       		// allocate slot with slot sharing
       		final SlotSharingManager multiTaskSlotManager = slotSharingManagers.computeIfAbsent(
       			scheduledUnit.getSlotSharingGroupId(),
       			id -> new SlotSharingManager(id,slotPool,this)); // 生成 SlotSharingManager
       
       		final SlotSharingManager.MultiTaskSlotLocality multiTaskSlotLocality;
       
       			if (scheduledUnit.getCoLocationConstraint() != null) {
       				multiTaskSlotLocality = allocateCoLocatedMultiTaskSlot(...);
       			} else {
       				multiTaskSlotLocality = allocateMultiTaskSlot(...); // 這裡生成 MultiTaskSlot
       			}
       
         	// 這裡生成 SingleTaskSlot
       		final SlotSharingManager.SingleTaskSlot leaf = multiTaskSlotLocality.getMultiTaskSlot().allocateSingleTaskSlot(...);
         
       		return leaf.getLogicalSlotFuture(); // 返回 SingleTaskSlot 的 future,就是第二個Future,具體生成我們在下面會詳述
       	}
      
      
  • SchedulerImpl # allocateMultiTaskSlot,這裡是一個難點函式。因為這裡生成了第三個 Future ,這裡把第三個 Future 提前說明第三是因為從使用者角度看,它是第三個出現的

    • private SlotSharingManager.MultiTaskSlotLocality allocateMultiTaskSlot(...) {
       
       		SlotSharingManager.MultiTaskSlot multiTaskSlot = slotSharingManager.getUnresolvedRootSlot(groupId);
       
       		if (multiTaskSlot == null) { 
            // requestNewAllocatedSlot 會呼叫 SlotPoolImpl 的同名函式
            // 得到第 三 個 Future,注意,這個 Future 針對的是 PhysicalSlot
       			final CompletableFuture<PhysicalSlot> slotAllocationFuture = requestNewAllocatedSlot(...); 
       
           // 使用 第 三 個 Future 來構建 multiTaskSlot
       			multiTaskSlot = slotSharingManager.createRootSlot(...,slotAllocationFuture,...);
       
           // 第 三 個 Future的回撥函式,這裡會把payload賦值給slot
       			slotAllocationFuture.whenComplete(
       				(PhysicalSlot allocatedSlot, Throwable throwable) -> {
       					final SlotSharingManager.TaskSlot taskSlot = slotSharingManager.getTaskSlot(multiTaskSlotRequestId);
       
       					if (taskSlot != null) {
                   // 會把payload賦值給slot
       							if (!allocatedSlot.tryAssignPayload(((SlotSharingManager.MultiTaskSlot) taskSlot))) {...}
       					} 
       				});
       		}
       
       		return SlotSharingManager.MultiTaskSlotLocality.of(multiTaskSlot, Locality.UNKNOWN);
       	}
      
      
      
  • SchedulerImpl # requestNewAllocatedSlot 會呼叫 SlotPoolImpl 的同名函式

  • SlotPoolImpl#requestNewAllocatedSlot,這裡生成一個 PendingRequest

    • public CompletableFuture<PhysicalSlot> requestNewAllocatedSlot(...) {
       
       		// 生成 PendingRequest
       		final PendingRequest pendingRequest = PendingRequest.createStreamingRequest(slotRequestId, resourceProfile);
       
       		// 新增 PendingRequest 到 waitingForResourceManager,然後返回Future
       		return requestNewAllocatedSlotInternal(pendingRequest)
       			.thenApply((Function.identity()));
       	}
      
      
    • PendingRequest的建構函式中有 new CompletableFuture<>(),所以這裡是生成了第三個 Future,注意這裡的 Future 是針對 PhysicalSlot

    • requestNewAllocatedSlotInternal

      • private CompletableFuture<AllocatedSlot> requestNewAllocatedSlotInternal(PendingRequest pendingRequest) {
        
           if (resourceManagerGateway == null) {
              // 就是把 pendingRequest 加到 waitingForResourceManager 之中
              stashRequestWaitingForResourceManager(pendingRequest);
           } else {
              requestSlotFromResourceManager(resourceManagerGateway, pendingRequest);
           }
           return pendingRequest.getAllocatedSlotFuture(); // 第三個Future
        }
        
        
  • SlotSharingManager#createRootSlot,這裡才是生成 第二個 Future 的地方

    • MultiTaskSlot createRootSlot(
            SlotRequestId slotRequestId,
            CompletableFuture<? extends SlotContext> slotContextFuture, // 引數是第三個Future
            SlotRequestId allocatedSlotRequestId) {
      
         // 生成第二個Future<SlotContext>
         final CompletableFuture<SlotContext> slotContextFutureAfterRootSlotResolution = new CompletableFuture<>();
        
         final MultiTaskSlot rootMultiTaskSlot = createAndRegisterRootSlot(...
            slotContextFutureAfterRootSlotResolution); // 第二個Future 在 createAndRegisterRootSlot 函式中 被賦值為 MultiTaskSlot的 slotContextFuture 成員變數
      
         FutureUtils.forward(
            slotContextFuture.thenApply( // 第三個Future進一步回撥時候,會回撥第二個Future
               (SlotContext slotContext) -> {
                  // add the root node to the set of resolved root nodes once the SlotContext future has
                  // been completed and we know the slot's TaskManagerLocation
                  tryMarkSlotAsResolved(slotRequestId, slotContext);
                  return slotContext;
               }),
            slotContextFutureAfterRootSlotResolution); // 在這裡回撥第二個Future
      
         return rootMultiTaskSlot;
      }
      
      
  • SlotSharingManager#allocateSingleTaskSlot,這裡的目的是在 rootMultiTaskSlot 之上生成一個 SingleTaskSlot leaf加入到allTaskSlots。

    • SingleTaskSlot allocateSingleTaskSlot(
      				SlotRequestId slotRequestId, ResourceProfile resourceProfile,
      				AbstractID groupId, Locality locality) {
      
      			final SingleTaskSlot leaf = new SingleTaskSlot(
      				slotRequestId, resourceProfile, groupId, this, locality);
      
      			children.put(groupId, leaf);
      
      			// register the newly allocated slot also at the SlotSharingManager
      			allTaskSlots.put(slotRequestId, leaf);
      
      			reserveResource(resourceProfile);
      
      			return leaf;
      }
      
      
  • 最後回到 SchedulerImpl # allocateSharedSlot 函式,return leaf.getLogicalSlotFuture(); 這裡也是一個難點,即 getLogicalSlotFuture 返回的是一個 CompletableFuture(就是第二個 Future),但是這個 SingleLogicalSlot 是未來回撥時候才會生成。

    • public final class SingleTaskSlot extends TaskSlot {
      		private final MultiTaskSlot parent;
         // future containing a LogicalSlot which is completed once the underlying SlotContext future is completed
      		private final CompletableFuture<SingleLogicalSlot> singleLogicalSlotFuture;
        
          private SingleTaskSlot() {
                singleLogicalSlotFuture = parent.getSlotContextFuture()
                  .thenApply(
                    (SlotContext slotContext) -> {
                      return new SingleLogicalSlot( // 未來回撥時候才會生成
                        slotRequestId,
                        slotContext,
                        slotSharingGroupId,
                        locality,
                        slotOwner);
                    });
              }
      
          CompletableFuture<LogicalSlot> getLogicalSlotFuture() {
             return singleLogicalSlotFuture.thenApply(Function.identity());
          }  
      }
      
      

0x06 Deploy階段

注意,此處的入口為 allocateSlotsAndDeploy函式中 的 deployIndividually / waitForAllSlotsAndDeploy 語句

此處執行路徑大致如下:

  • DefaultScheduler#allocateSlotsAndDeploy

  • DefaultScheduler#allocateSlots;得到 SlotExecutionVertexAssignment 列表,上節已經詳細介紹(該過程會ExecutionVertex轉化為ExecutionVertexSchedulingRequirements,會封裝包含一些location資訊、sharing資訊、資源資訊等)

  • List deploymentHandles = createDeploymentHandles() 根據SlotExecutionVertexAssignment建立DeploymentHandle

  • DefaultScheduler#deployIndividually 根據deploymentHandles進行部署,其實是根據Future把部署搭建起來,具體如何部署需要在slot分配成功之後再執行。我們小節實際是從這裡開始分析,具體程式碼可以看出,取出了 Future 1 進行一些列操作

    • private void deployIndividually(final List<DeploymentHandle> deploymentHandles) {
         for (final DeploymentHandle deploymentHandle : deploymentHandles) {
            FutureUtils.assertNoException(
               deploymentHandle
                  .getSlotExecutionVertexAssignment()
                  .getLogicalSlotFuture()
                  .handle(assignResourceOrHandleError(deploymentHandle))
                  .handle(deployOrHandleError(deploymentHandle)));
         }
      }
      
      
  • DefaultScheduler#assignResourceOrHandleError;就是返回函式,以備後續回撥使用

    • private BiFunction<LogicalSlot, Throwable, Void> assignResourceOrHandleError(final DeploymentHandle deploymentHandle) {
        
         final ExecutionVertexVersion requiredVertexVersion = deploymentHandle.getRequiredVertexVersion();
         final ExecutionVertexID executionVertexId = deploymentHandle.getExecutionVertexId();
      
         return (logicalSlot, throwable) -> {
            if (throwable == null) {
               final ExecutionVertex executionVertex = getExecutionVertex(executionVertexId);
               final boolean sendScheduleOrUpdateConsumerMessage = deploymentHandle.getDeploymentOption().sendScheduleOrUpdateConsumerMessage();
               executionVertex
                  .getCurrentExecutionAttempt()
                  .registerProducedPartitions(logicalSlot.getTaskManagerLocation(), sendScheduleOrUpdateConsumerMessage);
               executionVertex.tryAssignResource(logicalSlot);
            } else {
               handleTaskDeploymentFailure(executionVertexId, maybeWrapWithNoResourceAvailableException(throwable));
            }
            return null;
         };
      }
      
      
  • deployOrHandleError 就是返回函式,以備後續回撥使用

    • private BiFunction<Object, Throwable, Void> deployOrHandleError(final DeploymentHandle deploymentHandle) {
        
         final ExecutionVertexVersion requiredVertexVersion = deploymentHandle.getRequiredVertexVersion();
         final ExecutionVertexID executionVertexId = requiredVertexVersion.getExecutionVertexId();
      
         return (ignored, throwable) -> {
            if (throwable == null) {
               deployTaskSafe(executionVertexId);
            } else {
               handleTaskDeploymentFailure(executionVertexId, throwable);
            }
            return null;
         };
      }
      
      

0x07 RM分配資源

之前的工作基本都是在 JM 之中。通過 Scheduler 和 SlotPool 來完成申請資源和部署階段。目前 SlotPool 之中已經積累了一個 PendingRequest,等 SlotPool 連線上 RM,就可以開始向 RM 申請資源了。

當ResourceManager收到申請slot請求時,若發現該JobManager未註冊,則直接丟擲異常;否則將請求轉發給SlotManager處理,SlotManager中維護了叢集所有空閒的slot(TaskManager會向ResourceManager上報自己的資訊,在ResourceManager中由SlotManager儲存Slot和TaskManager對應關係),並從其中找出符合條件的slot,然後向TaskManager傳送RPC請求申請對應的slot。

程式碼執行路徑如下:

  • JobMaster # establishResourceManagerConnection 程式執行在 JM 之中

  • SlotPoolImpl # connectToResourceManager

  • SlotPoolImpl # requestSlotFromResourceManager,這裡 Pool 會向 RM 進行 RPC 請求。

    • private void requestSlotFromResourceManager(
      			final ResourceManagerGateway resourceManagerGateway,
      			final PendingRequest pendingRequest) {
          // 生成一個 AllocationID,這個會傳到 TM 那裡,註冊到 TaskSlot上。
          final AllocationID allocationId = new AllocationID();
          // 生成一個SlotRequest,並且向 RM 進行 RPC 請求。
      	CompletableFuture<Acknowledge> rmResponse = 
              				resourceManagerGateway.requestSlot(
                                      jobMasterId,
                                      new SlotRequest(jobId, allocationId, 
                                           		pendingRequest.getResourceProfile(), 
                                      jobManagerAddress),
                                      rpcTimeout);
      }
      
  • RPC

  • ResourceManager # requestSlot 程式切換到 RM 之中

  • SlotManagerImpl # registerSlotRequest。registerSlotRequest方法會先執行checkDuplicateRequest判斷是否有重複,沒有重複的話,則將該slotRequest維護到pendingSlotRequests,然後呼叫internalRequestSlot進行分配,如果出現異常則從pendingSlotRequests中異常,然後丟擲SlotManagerException。

    • pendingSlotRequests.put
      
  • SlotManagerImpl # internalRequestSlot

  • SlotManagerImpl # findMatchingSlot

  • SlotManagerImpl # internalAllocateSlot,此時是沒有資源的,需要向 TM 要求資源

    • private void internalRequestSlot(PendingSlotRequest pendingSlotRequest) throws ResourceManagerException {
         final ResourceProfile resourceProfile = pendingSlotRequest.getResourceProfile();
         OptionalConsumer.of(findMatchingSlot(resourceProfile))
            .ifPresent(taskManagerSlot -> allocateSlot(taskManagerSlot, pendingSlotRequest))
            .ifNotPresent(() -> fulfillPendingSlotRequestWithPendingTaskManagerSlot(pendingSlotRequest));
      }
      
  • SlotManagerImpl # allocateSlot,向task manager要求資源。TaskExecutorGateway介面用來通過RPC分配任務槽,或者說分配任務的資源。

    • TaskExecutorGateway gateway = taskExecutorConnection.getTaskExecutorGateway();
      CompletableFuture<Acknowledge> requestFuture = gateway.requestSlot(
      			slotId,
      			pendingSlotRequest.getJobId(),
      			allocationId,
      			pendingSlotRequest.getResourceProfile(),
      			pendingSlotRequest.getTargetAddress(),
      			resourceManagerId,
      			taskManagerRequestTimeout);
      
  • RPC

  • TaskExecutor # requestSlot,程式切換到 TE

  • TaskSlotTableImpl # allocateSlot,分配資源,更新task slot map,把slot加入到 set of job slots 中。

    • public boolean allocateSlot(int index, JobID jobId, AllocationID allocationId,
      			ResourceProfile resourceProfile,Time slotTimeout) {
          taskSlot = new TaskSlot<>(index, resourceProfile, memoryPageSize, jobId, allocationId);
          taskSlots.put(index, taskSlot);
          allocatedSlots.put(allocationId, taskSlot);
          slots.add(allocationId);
      }
      
      

0x08 Offer資源階段

此階段是由 TE,TM 開始,就是TE 向 RM 提供 Slot,然後 RM 通知 JM 可以執行 Job。也可以認為這部分是從底向上的執行。

等待所有的slot申請完成後,然後會將ExecutionVertex對應的Execution分配給對應的Slot,即從Slot中分配對應的資源給Execution,完成分配後可開始部署作業。

這裡兩個關鍵點是:

  • 當 JM 收到 SlotOffer時候,就會根據 RPC傳遞過來的 taskManagerId 引數,構建一個 taskExecutorGateway,然後這個 taskExecutorGateway 被賦予為 AllocatedSlot . taskManagerGateway。這樣就把 JM 範疇的 Slot 和 Slot 所在的 taskManager 聯絡起來
  • Execution 部署時候,是 從 SingleLogicalSlot ---> AllocatedSlot ---> TaskManagerGateway 這個順序獲取了 TaskManager 的 RPC 閘道器,然後通過 taskManagerGateway.submitTask 才能提交任務的。這樣就把 Execution 部署階段和執行階段聯絡起來了
---------- Task Executor ----------
       │ 
       │ 
┌─────────────┐   
│  TaskSlot   │  requestSlot
└─────────────┘     
       │ 
       │                  
┌──────────────┐   
│  SlotOffer   │  offerSlotsToJobManager
└──────────────┘       
       │ 
       │      
------------- Job Manager -------------
       │ 
       │       
┌──────────────┐   
│  SlotOffer   │  JobMaster#offerSlots(taskManagerId,slots)
└──────────────┘     
       │ //taskManager = registeredTaskManagers.get(taskManagerId);     
       │ //taskManagerLocation = taskManager.f0;     
       │ //taskExecutorGateway = taskManager.f1;     
       │    
       │       
┌──────────────┐   
│  SlotOffer   │  SlotPoolImpl#offerSlots
└──────────────┘       
       │ 
       │      
┌───────────────┐   
│ AllocatedSlot │  SlotPoolImpl#offerSlot
└───────────────┘      
       │ 
       │      
┌───────────────┐   
│ 回撥 Future 3  │ SlotSharingManager#createRootSlot
└───────────────┘      
       │ 
       │      
┌───────────────┐   
│ 回撥 Future 2  │  SingleTaskSlot#SingleTaskSlot 
└───────────────┘      
       │ 
       │      
┌───────────────────┐   
│ SingleLogicalSlot │ new SingleLogicalSlot
└───────────────────┘    
       │ 
       │     
┌───────────────────┐   
│ SingleLogicalSlot │  
│ 回撥 Future 1      │ allocationResultFuture.complete()
└───────────────────┘   
       │    
       │        
┌───────────────────────────────┐  
│     SingleLogicalSlot         │  
│回撥 assignResourceOrHandleError│ 
└───────────────────────────────┘
       │    
       │        
┌────────────────┐   
│ ExecutionVertex│ tryAssignResource
└────────────────┘    
       │    
       │        
┌────────────────┐   
│    Execution   │ tryAssignResource
└────────────────┘       
       │    
       │        
┌──────────────────┐   
│ SingleLogicalSlot│ tryAssignPayload
└──────────────────┘  
       │    
       │        
┌───────────────────────┐   
│   SingleLogicalSlot   │      
│ 回撥deployOrHandleError│ 
└───────────────────────┘   
       │    
       │       
┌────────────────┐   
│ ExecutionVertex│ deploy
└────────────────┘    
       │    
       │        
┌────────────────┐   
│    Execution   │ deploy // 關鍵點
└────────────────┘        
       │  
       │    
       │        
 ---------- Task Executor ----------
       │    
       │     
┌────────────────┐   
│  TaskExecutor  │ submitTask
└────────────────┘     
       │    
       │        
┌────────────────┐   
│  TaskExecutor  │ startTaskThread
└────────────────┘         

執行路徑如下:

  • TaskExecutor # establishJobManagerConnection

  • TaskExecutor # offerSlotsToJobManager,這裡就是遍歷已經分配的TaskSlot,然後每個TaskSlot會生成一個SlotOffer(裡面是allocationId,slotIndex,resourceProfile),這個會通過RPC發給 JM。

    • private void offerSlotsToJobManager(final JobID jobId) {
      				final Iterator<TaskSlot<Task>> reservedSlotsIterator = taskSlotTable.getAllocatedSlots(jobId);
      				final JobMasterId jobMasterId = jobManagerConnection.getJobMasterId();
      
      				final Collection<SlotOffer> reservedSlots = new HashSet<>(2);
      
      				while (reservedSlotsIterator.hasNext()) {
      					SlotOffer offer = reservedSlotsIterator.next().generateSlotOffer();
      					reservedSlots.add(offer);
      				}
          			// 把 SlotOffer 通過RPC發給 JM
      				CompletableFuture<Collection<SlotOffer>> acceptedSlotsFuture = 
                          jobMasterGateway.offerSlots(
                                  getResourceID(),
                                  reservedSlots,
                                  taskManagerConfiguration.getTimeout());    
      }
      
      
  • RPC

  • JobMaster # offerSlots 。程式執行到 JM。當 JM 收到 SlotOffer時候,就會根據 RPC傳遞過來的 taskManagerId 引數,構建一個 taskExecutorGateway,然後這個 taskExecutorGateway 被賦予為 AllocatedSlot . taskManagerGateway。這樣就把 JM 範疇的 Slot 和 Slot 所在的 taskManager 聯絡起來

    • public CompletableFuture<Collection<SlotOffer>> offerSlots(
      			final ResourceID taskManagerId,
      			final Collection<SlotOffer> slots,
      			final Time timeout) {
      
      		Tuple2<TaskManagerLocation, TaskExecutorGateway> taskManager = registeredTaskManagers.get(taskManagerId);
      
      		final TaskManagerLocation taskManagerLocation = taskManager.f0;
      		final TaskExecutorGateway taskExecutorGateway = taskManager.f1;
      
      		final RpcTaskManagerGateway rpcTaskManagerGateway = new RpcTaskManagerGateway(taskExecutorGateway, getFencingToken());
      
      		return CompletableFuture.completedFuture(
      			slotPool.offerSlots(
      				taskManagerLocation,
      				rpcTaskManagerGateway,
      				slots));
      	}
      
      
  • SlotPoolImpl # offerSlots

  • SlotPoolImpl # offerSlot,這裡根據 SlotOffer 的資訊生成一個 AllocatedSlot,對於 AllocatedSlot 來說,有效資訊就是 slotIndex, resourceProfile。提醒,AllocatedSlot implements PhysicalSlot。

    • boolean offerSlot(
      			final TaskManagerLocation taskManagerLocation,
      			final TaskManagerGateway taskManagerGateway,
      			final SlotOffer slotOffer) {
              
          // 根據 SlotOffer 的資訊生成一個 AllocatedSlot,對於 AllocatedSlot 來說,有效資訊就是 slotIndex, resourceProfile
      	final AllocatedSlot allocatedSlot = new AllocatedSlot(
             allocationID,
             taskManagerLocation,
             slotOffer.getSlotIndex(),
             slotOffer.getResourceProfile(),
             taskManagerGateway);
          
          allocatedSlots.add(pendingRequest.getSlotRequestId(), allocatedSlot);
      		if (pendingRequest != null) {
      			allocatedSlots.add(pendingRequest.getSlotRequestId(), allocatedSlot);
      
            // 這裡取出了 pendingRequest 的 Future, 就是我們之前的 Future 3,進行回撥
      			if (!pendingRequest.getAllocatedSlotFuture().complete(allocatedSlot)) 
                  {
      				// we could not complete the pending slot future --> try to fulfill another pending request
      				allocatedSlots.remove(pendingRequest.getSlotRequestId());
      				tryFulfillSlotRequestOrMakeAvailable(allocatedSlot);
      			} 
      		}
      }
      
      
  • 開始回撥 Future 3,程式碼在 SlotSharingManager # createRootSlot 這裡

    • FutureUtils.forward(
         slotContextFuture.thenApply(
            (SlotContext slotContext) -> {
               // add the root node to the set of resolved root nodes once the SlotContext future has
               // been completed and we know the slot's TaskManagerLocation
               tryMarkSlotAsResolved(slotRequestId, slotContext); // 執行到這裡
               return slotContext;
            }),
         slotContextFutureAfterRootSlotResolution); // 然後到這裡
      
      
  • 開始回撥 Future 2,程式碼在 SingleTaskSlot 建構函式 ,因為有 PhysicalSlot extends SlotContext, 所以這裡就把 物理Slot 對映成了一個 邏輯Slot

    • singleLogicalSlotFuture = parent.getSlotContextFuture()
         .thenApply(
            (SlotContext slotContext) -> {
               return new SingleLogicalSlot( // 回撥生成了 SingleLogicalSlot
                  slotRequestId,
                  slotContext,
                  slotSharingGroupId,
                  locality,
                  slotOwner);
            });
      
      
  • 開始回撥 Future 1,程式碼在這裡,呼叫到 後續 Deploy 時候設定的回撥函式

    • allocationFuture.whenComplete((LogicalSlot slot, Throwable failure) -> {
         if (failure != null) {
            cancelSlotRequest(
               slotRequestId,
               scheduledUnit.getSlotSharingGroupId(),
               failure);
            allocationResultFuture.completeExceptionally(failure);
         } else {
            allocationResultFuture.complete(slot); // 程式碼在這裡
         }
      });
      
      
  • 繼續回撥到 Deploy 階段設定的回撥函式 assignResourceOrHandleError,就是分配資源

    • private BiFunction<LogicalSlot, Throwable, Void> assignResourceOrHandleError(final DeploymentHandle deploymentHandle) {
       
       		return (logicalSlot, throwable) -> {
       			if (executionVertexVersioner.isModified(requiredVertexVersion)) {
       
       			if (throwable == null) {
       				final ExecutionVertex executionVertex = getExecutionVertex(executionVertexId);
       				final boolean sendScheduleOrUpdateConsumerMessage = deploymentHandle.getDeploymentOption().sendScheduleOrUpdateConsumerMessage();
       				executionVertex
       					.getCurrentExecutionAttempt()
       					.registerProducedPartitions(logicalSlot.getTaskManagerLocation(), sendScheduleOrUpdateConsumerMessage);
       				executionVertex.tryAssignResource(logicalSlot); // 執行到這裡
       			} 
       			return null;
       		};
       	}
      
      
    • 回撥函式會深入呼叫 executionVertex.tryAssignResource,

    • ExecutionVertex # tryAssignResource

    • Execution # tryAssignResource

    • SingleLogicalSlot# tryAssignPayload(this),這裡會把 Execution 自己 賦值給Slot.payload,最後 Execution 在 runtime 的變數舉例如下:

      • payload = {Execution@10669} "Attempt #0 (CHAIN DataSource (at getDefaultTextLineDataSet(WordCountData.java:47) (org.apache.flink.api.java.io.CollectionInputFormat)) -> FlatMap (FlatMap at main(WordCount.java:64)) -> Combine (SUM(1), at main(WordCount.java:67) (1/1)) @ org.apache.flink.runtime.jobmaster.slotpool.SingleLogicalSlot@61c7928f - [SCHEDULED]"
         executor = {ScheduledThreadPoolExecutor@5928} "java.util.concurrent.ScheduledThreadPoolExecutor@6a2c6c71[Running, pool size = 3, active threads = 0, queued tasks = 1, completed tasks = 2]"
         vertex = {ExecutionVertex@10534} "CHAIN DataSource (at getDefaultTextLineDataSet(WordCountData.java:47) (org.apache.flink.api.java.io.CollectionInputFormat)) -> FlatMap (FlatMap at main(WordCount.java:64)) -> Combine (SUM(1), at main(WordCount.java:67) (1/1)"
         attemptId = {ExecutionAttemptID@10792} "2f8b6c7297527225ee4c8036c457ba27"
         globalModVersion = 1
         stateTimestamps = {long[9]@10793} 
         attemptNumber = 0
         rpcTimeout = {Time@5924} "18000000 ms"
         partitionInfos = {ArrayList@10794}  size = 0
         terminalStateFuture = {CompletableFuture@10795} "java.util.concurrent.CompletableFuture@2eb8f94c[Not completed]"
         releaseFuture = {CompletableFuture@10796} "java.util.concurrent.CompletableFuture@7c794914[Not completed]"
         taskManagerLocationFuture = {CompletableFuture@10797} "java.util.concurrent.CompletableFuture@2e11ac18[Not completed]"
         state = {ExecutionState@10789} "SCHEDULED"
         assignedResource = {SingleLogicalSlot@10507} 
         failureCause = null
         taskRestore = null
         assignedAllocationID = null
         accumulatorLock = {Object@10798} 
         userAccumulators = null
         ioMetrics = null
         producedPartitions = {LinkedHashMap@10799}  size = 1
        
        
  • 繼續回撥到 Deploy 階段設定的回撥函式 deployOrHandleError,就是部署

    • private BiFunction<Object, Throwable, Void> deployOrHandleError(final DeploymentHandle deploymentHandle) {
      
         return (ignored, throwable) -> {
            if (executionVertexVersioner.isModified(requiredVertexVersion)) {
      
            if (throwable == null) {
               deployTaskSafe(executionVertexId); // 在這裡部署
            } else {
               handleTaskDeploymentFailure(executionVertexId, throwable);
            }
            return null;
         };
      }
      
      
    • 回撥函式深入呼叫其他函式

    • DefaultScheduler # deployTaskSafe

    • ExecutionVertex # deploy

    • Execution # deploy。每次排程ExecutionVertex,都會有一個Execution,在此階段會將Execution的狀態變更為DEPLOYING狀態,並且為該ExecutionVertex生成對應的部署描述資訊,然後從對應的slot中獲取對應的TaskManagerGateway,以便向對應的TaskManager提交Task。其中,ExecutionVertex.createDeploymentDescriptor方法中,包含了從Execution Graph到真正物理執行圖的轉換。如將IntermediateResultPartition轉化成ResultPartition,ExecutionEdge轉成InputChannelDeploymentDescriptor(最終會在執行時轉化成InputGate)。

      • // 這裡一個關鍵點是:Execution 部署時候,是 從 SingleLogicalSlot ---> AllocatedSlot ---> TaskManagerGateway 這個順序獲取了 TaskManager 的 RPC 閘道器,然後通過 taskManagerGateway.submitTask 才能提交任務的。這樣就把 Execution 部署階段和執行階段聯絡起來了
        public void deploy() throws JobException {
        			final TaskDeploymentDescriptor deployment = TaskDeploymentDescriptorFactory
        			.fromExecutionVertex(vertex, attemptNumber)
        				.createDeploymentDescriptor(
        				slot.getAllocationId(),
        					slot.getPhysicalSlotNumber(),
        					taskRestore,
        					producedPartitions.values());
          
            	// 這裡就是關鍵點
          		final TaskManagerGateway taskManagerGateway = slot.getTaskManagerGateway();
          
              // 在這裡通過RPC提交task給了TaskManager
          		CompletableFuture.supplyAsync(() -> taskManagerGateway.submitTask(deployment, rpcTimeout), executor).thenCompose(Function.identity())
        }
        
        
  • TaskExecutor # submitTask, 程式執行到 TE,這就是正式執行了。TaskManager(TaskExecutor)在接收到提交Task的請求後,會經過一些初始化(如從BlobServer拉取檔案,反序列化作業和Task資訊、LibaryCacheManager等),然後這些初始化的資訊會用於生成Task(Runnable物件),然後啟動該Task,其程式碼呼叫路徑如下 Task#startTaskThread(啟動Task執行緒)-> Task#run(將ExecutionVertex狀態變更為RUNNING狀態,此時在FLINK web前臺檢視頂點狀態會變更為RUNNING狀態,另外還會生成了一個AbstractInvokable物件,該物件是FLINK銜接執行使用者程式碼的關鍵。

    • // 這個方法會建立真正的Task,然後呼叫task.startTaskThread();開始task的執行。
      public CompletableFuture<Acknowledge> submitTask(
            TaskDeploymentDescriptor tdd, JobMasterId jobMasterId, Time timeout) {
        		// taskSlot.getMemoryManager(); 會獲取slot的記憶體管理器,這裡就是分割記憶體的部分功能
        		memoryManager = taskSlotTable.getTaskMemoryManager(tdd.getAllocationId());
        		// 在Task建構函式中,會根據輸入的引數,建立InputGate, ResultPartition, ResultPartitionWriter等。
      			Task task = new Task(
      				jobInformation,
      				taskInformation,
      				tdd.getExecutionAttemptId(),
      				tdd.getAllocationId(),
      				tdd.getSubtaskIndex(),
      				tdd.getAttemptNumber(),
      				tdd.getProducedPartitions(),
      				tdd.getInputGates(),
      				tdd.getTargetSlotNumber(),
      				memoryManager,
      				taskExecutorServices.getIOManager(),
      				taskExecutorServices.getShuffleEnvironment(),
      				taskExecutorServices.getKvStateService(),
      				taskExecutorServices.getBroadcastVariableManager(),
      				taskExecutorServices.getTaskEventDispatcher(),
      				taskStateManager,
      				taskManagerActions,
      				inputSplitProvider,
      				checkpointResponder,
      				aggregateManager,
      				blobCacheService,
      				libraryCache,
      				fileCache,
      				taskManagerConfiguration,
      				taskMetricGroup,
      				resultPartitionConsumableNotifier,
      				partitionStateChecker,
      				getRpcService().getExecutor());
        
            taskAdded = taskSlotTable.addTask(task);
        		task.startTaskThread();
      }
      
      
    • 開始了執行緒了。而startTaskThread方法,則會執行executingThread.start,從而呼叫Task.run方法。

      • public void startTaskThread() {
           executingThread.start();
        }
        
        
  • 最後會執行到 Task,就是呼叫使用者程式碼。這裡的invokable即為operator物件例項,通過反射建立。具體地,即為OneInputStreamTask,或者SourceStreamTask等。以OneInputStreamTask為例,Task的核心執行程式碼即為OneInputStreamTask.invoke方法,它會呼叫StreamTask.run方法,這是個抽象方法,最終會呼叫其派生類的run方法,即OneInputStreamTask, SourceStreamTask等。

    • // 這裡的invokable即為operator物件例項,通過反射建立。
      private void doRun() {
      	AbstractInvokable invokable = null;
      	invokable = loadAndInstantiateInvokable(userCodeClassLoader, nameOfInvokableClass, env);
      	// run the invokable
        invokable.invoke();
      }
      
      
  • tryFulfillSlotRequestOrMakeAvailable

0x09 Slot發揮作用

有人可能有一個疑問:Slot分配之後,在執行時候怎麼發揮作用呢?

這裡我們就用WordCount示例來看看。

示例程式碼就是WordCount。只不過做了一些配置:

  • taskmanager.numberOfTaskSlots 是為了設定有幾個taskmanager。
  • 其他是為了除錯,加長了心跳時間或者超時時間。
public class WordCount {

    public static void main(String[] args) throws Exception {

        Configuration conf = new Configuration();
        conf.setString("heartbeat.timeout", "18000000");
        conf.setString("resourcemanager.job.timeout", "18000000");
        conf.setString("resourcemanager.taskmanager-timeout", "18000000");
        conf.setString("slotmanager.request-timeout", "18000000");
        conf.setString("slotmanager.taskmanager-timeout", "18000000");
        conf.setString("slot.request.timeout", "18000000");
        conf.setString("slot.idle.timeout", "18000000");
        conf.setString("akka.ask.timeout", "18000000");
        conf.setString("taskmanager.numberOfTaskSlots", "1");

        final LocalEnvironment env = ExecutionEnvironment.createLocalEnvironment(conf);
        final MultipleParameterTool params = MultipleParameterTool.fromArgs(args);
        env.getConfig().setGlobalJobParameters(params);

        // get input data
        DataSet<String> text = null;
        if (params.has("input")) {
            // union all the inputs from text files
            for (String input : params.getMultiParameterRequired("input")) {
                if (text == null) {
                    text = env.readTextFile(input);
                } else {
                    text = text.union(env.readTextFile(input));
                }
            }
        } else {
            // get default test text data
            text = WordCountData.getDefaultTextLineDataSet(env);
        }

        DataSet<Tuple2<String, Integer>> counts =
                // split up the lines in pairs (2-tuples) containing: (word,1)
                text.flatMap(new Tokenizer())
                        // group by the tuple field "0" and sum up tuple field "1"
                        .groupBy(0)
                        .sum(1);

        // emit result
        if (params.has("output")) {
            counts.writeAsCsv(params.get("output"), "\n", " ");
            env.execute("WordCount Example");
        } else {
            counts.print();
        }
    }

    // *************************************************************************
    //     USER FUNCTIONS
    // *************************************************************************
    public static final class Tokenizer implements FlatMapFunction<String, Tuple2<String, Integer>> {

        @Override
        public void flatMap(String value, Collector<Tuple2<String, Integer>> out) {
            // normalize and split the line
            String[] tokens = value.toLowerCase().split("\\W+");

            // emit the pairs
            for (String token : tokens) {
                if (token.length() > 0) {
                    out.collect(new Tuple2<>(token, 1));
                }
            }
        }
    }
}

9.1 部署階段

這裡 Slot 起到了一個承接作用,把具體提交部署和執行階段聯絡起來

前面提到,當TE 提交一個Slot之後,RM會在這個Slot上提交Task。具體邏輯如下:

每次排程ExecutionVertex,都會有一個Execution。在 Execution # deploy 函式中。

  • 會將Execution的狀態變更為DEPLOYING狀態,並且為該ExecutionVertex生成對應的部署描述資訊。其中,ExecutionVertex.createDeploymentDescriptor方法中,包含了從Execution Graph到真正物理執行圖的轉換。
    • 如將IntermediateResultPartition轉化成ResultPartition
    • ExecutionEdge轉成InputChannelDeploymentDescriptor(最終會在執行時轉化成InputGate)。
  • 然後從對應的slot中獲取對應的TaskManagerGateway,以便向對應的TaskManager提交Task。這裡一個關鍵點是:Execution 部署時候,是 從 SingleLogicalSlot ---> AllocatedSlot ---> TaskManagerGateway 這個順序獲取了 TaskManager 的 RPC 閘道器。
  • 最後通過 taskManagerGateway.submitTask 提交 Task。

具體程式碼如下:

// 這裡一個關鍵點是:Execution 部署時候,是 從 SingleLogicalSlot ---> AllocatedSlot ---> TaskManagerGateway 這個順序獲取了 TaskManager 的 RPC 閘道器,然後通過 taskManagerGateway.submitTask 才能提交任務的。這樣就把 Execution 部署階段和執行階段聯絡起來了
public void deploy() throws JobException {
			final TaskDeploymentDescriptor deployment = TaskDeploymentDescriptorFactory
			.fromExecutionVertex(vertex, attemptNumber)
				.createDeploymentDescriptor(
				slot.getAllocationId(),
					slot.getPhysicalSlotNumber(),
					taskRestore,
					producedPartitions.values());

   	// 這裡就是關鍵點
  		final TaskManagerGateway taskManagerGateway = slot.getTaskManagerGateway();

      // 在這裡通過RPC提交task給了TaskManager
  		CompletableFuture.supplyAsync(() -> taskManagerGateway.submitTask(deployment, rpcTimeout), executor).thenCompose(Function.identity())
}

9.2 執行階段

這裡僅以Split為例子說明,Slot在其中也起到了連線作用,使用者從Slot中可以得到其 TaskManager 的host,然後Split會根據這個host繼續操作

當 Source 讀取輸入之後,可能涉及到分割輸入,Flink就會進行輸入分片的切分。

9.2.1 FileInputSplit 的由來

Flink 一般把檔案按並行度拆分成FileInputSplit的個數,當然並不是完全有幾個並行度就生成幾個FileInputSplit物件,根據具體演算法得到,但是FileInputSplit個數,一定是(並行度個數,或者並行度個數+1)。因為計算FileInputSplit個數時,參照物是檔案大小 / 並行度 ,如果沒有餘數,剛好整除,那麼FileInputSplit個數一定是並行度,如果有餘數,FileInputSplit個數就為是(並行度個數,或者並行度個數+1)。

Flink在生成階段,會把JobVertex 轉化為ExecutionJobVertex,呼叫new ExecutionJobVertex(),ExecutionJobVertex中存了inputSplits,所以會根據並行並來計算inputSplits的個數。

ExecutionJobVertex 建構函式中有如下程式碼,這些程式碼作用是生成 InputSplit,賦值到 ExecutionJobVertex 的成員變數 inputSplits 中,這樣就知道了從哪裡得倒 Split:

// set up the input splits, if the vertex has any
try {
			InputSplitSource<InputSplit> splitSource = (InputSplitSource<InputSplit>) jobVertex.getInputSplitSource();

			if (splitSource != null) {
				try {
					inputSplits = splitSource.createInputSplits(numTaskVertices);
					if (inputSplits != null) {
						splitAssigner = splitSource.getInputSplitAssigner(inputSplits);
					}
				} 
}
            
// 此時splitSource如下:
splitSource = {CollectionInputFormat@7603} "[To be, or not to be,--that is the question:--, Whether 'tis nobler in the mind to suffer, The slings and arrows of outrageous fortune, ...]"
 serializer = {StringSerializer@7856} 
 dataSet = {ArrayList@7857}  size = 35
 iterator = null
 partitionNumber = 0
 runtimeContext = null    

9.2.2 File Split

這裡以網上文章Flink-1.10.0中的readTextFile解讀內容為例,給大家看看檔案切片大致流程。當然他介紹的是Stream型別。

readTextFile分成兩個階段,一個Source,一個Split Reader。這兩個階段可以分為多個執行緒,不一定是2個執行緒。因為Split Reader的並行度時根據配置檔案或者啟動引數來決定的。

Source的執行流程如下,Source的是用來構建輸入切片的,不做資料的讀取操作。這裡是按照本地執行模式整理的。

Task.run()
 |-- invokable.invoke()
 |    |-- StreamTask.invoke()
 |    |    |-- beforeInvoke()
 |    |    |    |-- init()
 |    |    |    |    |-- SourceStreamTask.init()
 |    |    |    |-- initializeStateAndOpen()
 |    |    |    |    |-- operator.initializeState()
 |    |    |    |    |-- operator.open()
 |    |    |    |    |    |-- SourceStreamTask.LegacySourceFunctionThread.run()
 |    |    |    |    |    |    |-- StreamSource.run()
 |    |    |    |    |    |    |    |-- userFunction.run(ctx)
 |    |    |    |    |    |    |    |    |-- ContinuousFileMonitoringFunction.run()
 |    |    |    |    |    |    |    |    |    |-- RebalancePartitioner.selectChannel()
 |    |    |    |    |    |    |    |    |    |-- RecordWriter.emit()

Split Reader的程式碼執行流程如下:

Task.run()
 |-- invokable.invoke()
 |    |-- StreamTask.invoke()
 |    |    |-- beforeInvoke()
 |    |    |    |-- init()
 |    |    |    |    |--OneInputStreamTask.init()
 |    |    |    |-- initializeStateAndOpen()
 |    |    |    |    |-- operator.initializeState()
 |    |    |    |    |    |-- ContinuousFileReaderOperator.initializeState()
 |    |    |    |    |-- operator.open()
 |    |    |    |    |    |-- ContinuousFileReaderOperator.open()
 |    |    |    |    |    |    |-- ContinuousFileReaderOperator.SplitReader.run()
 |    |    |-- runMailboxLoop()
 |    |    |    |-- StreamTask.processInput()
 |    |    |    |    |-- StreamOneInputProcessor.processInput()
 |    |    |    |    |    |-- StreamTaskNetworkInput.emitNext() while迴圈不停的處理輸入資料
 |    |    |    |    |    |    |-- ContinuousFileReaderOperator.processElement()
 |    |    |-- afterInvoke()    

9.2.3 Slot的使用

針對本文示例,我們重點介紹Slot在其中的使用。

呼叫路徑如下:

  • DataSourceTask # invoke,此時執行在 TE

  • DataSourceTask # hasNext

    • while (!this.taskCanceled && splitIterator.hasNext())
      
  • RpcInputSplitProvider # getNextInputSplit

    • CompletableFuture<SerializedInputSplit> futureInputSplit = jobMasterGateway.requestNextInputSplit(   jobVertexID,   executionAttemptID);
      
  • RPC

  • 來到 JM

  • JobMaster # requestNextInputSplit

  • SchedulerBase # requestNextInputSplit,這裡會從 executionGraph 獲取 Execution,然後從 Execution 獲取 InputSplit

    • public SerializedInputSplit requestNextInputSplit(JobVertexID vertexID, ExecutionAttemptID executionAttempt) throws IOException {
      
      		final Execution execution = executionGraph.getRegisteredExecutions().get(executionAttempt);
      
      		final ExecutionJobVertex vertex = executionGraph.getJobVertex(vertexID);
      
      		final InputSplit nextInputSplit = execution.getNextInputSplit();
      
      		final byte[] serializedInputSplit = InstantiationUtil.serializeObject(nextInputSplit);
              
      		return new SerializedInputSplit(serializedInputSplit);
      }
      
    • 這裡 execution.getNextInputSplit() 就會呼叫 Slot,可以看到,這裡先獲取Slot,然後從Slot獲取其 TaskManager 的host。再從 Vertiex 獲取 InputSplit

      • public InputSplit getNextInputSplit() {
        		final LogicalSlot slot = this.getAssignedResource();
        		final String host = slot != null ? slot.getTaskManagerLocation().getHostname() : null;
        		return this.vertex.getNextInputSplit(host);
        }
        
      • public InputSplit getNextInputSplit(String host) {
        		final int taskId = getParallelSubtaskIndex();
        		synchronized (inputSplits) {
        			final InputSplit nextInputSplit = jobVertex.getSplitAssigner().getNextInputSplit(host, taskId);
        			if (nextInputSplit != null) {
        				inputSplits.add(nextInputSplit);
        			}
        			return nextInputSplit;
        		}
        }
        
        // runtime 資訊如下
        inputSplits = {GenericInputSplit[1]@13113} 
         0 = {GenericInputSplit@13121} "GenericSplit (0/1)"
          partitionNumber = 0
          totalNumberOfPartitions = 1
        
  • 回到 SchedulerBase # requestNextInputSplit,返回 return new SerializedInputSplit(serializedInputSplit);

  • RPC

  • 返回 運算元 Task,TE,獲取到了 InputSplit,就可以繼續處理輸入。

    • final InputSplit split = splitIterator.next();
      final InputFormat<OT, InputSplit> format = this.format;			
      // open input format
      // open還沒開始真正的讀資料,只是定位,設定當前切片資訊(切片的開始位置,切片長度),和定位開始位置。把第一個換行符,分到前一個分片,自己從第二個換行符開始讀取資料
      format.open(split);
      

0xFF 參考

一文了解 Apache Flink 的資源管理機制

Flink5:Flink執行架構(Slot和並行度)

Flink Slot詳解與Job Execution Graph優化

聊聊flink的slot.request.timeout配置

Apache Flink 原始碼解析(三)Flink on Yarn (2) Resource Manager

Flink on Yarn模式下的TaskManager個數

Flink on YARN時,如何確定TaskManager數

Flink】Flink作業排程流程分析

Flink原理與實現:如何生成ExecutionGraph及物理執行圖

Flink原始碼走讀(一):Flink工程目錄

flink分析使用之七任務的啟動

flink原始碼解析3 ExecutionGraph的形成與物理執行

Flink 內部原理之作業與排程

Flink之使用者程式碼生成排程層圖結構

3. Flink Slot申請

Flink 任務和排程

Flink的Slot是如何做到平均劃分TM記憶體的?

Flink-1.10.0中的readTextFile解讀

Flink1.7.2 Dataset 檔案切片計算方式和切片資料讀取原始碼分析

flink任務提交流程分析

Flink Parallelism和Slot理解

相關文章