[ 
https://issues.apache.org/jira/browse/FLINK-6434?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16236178#comment-16236178
 ] 

ASF GitHub Bot commented on FLINK-6434:
---------------------------------------

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

    https://github.com/apache/flink/pull/4937#discussion_r148596180
  
    --- Diff: 
flink-runtime/src/main/java/org/apache/flink/runtime/instance/SlotPool.java ---
    @@ -262,23 +263,36 @@ public void disconnectResourceManager() {
        // 
------------------------------------------------------------------------
     
        @Override
    -   public CompletableFuture<SimpleSlot> allocateSlot(
    -                   ScheduledUnit task,
    +   public CompletableFuture<SimpleSlot> allocateSlot(AllocationID 
allocationID,
                        ResourceProfile resources,
                        Iterable<TaskManagerLocation> locationPreferences,
                        Time timeout) {
     
    -           return internalAllocateSlot(task, resources, 
locationPreferences);
    +           return internalAllocateSlot(allocationID, resources, 
locationPreferences);
        }
     
        @Override
        public void returnAllocatedSlot(Slot slot) {
                internalReturnAllocatedSlot(slot);
        }
     
    +   @Override
    +   public void cancelSlotAllocation(AllocationID allocationID) {
    +           waitingForResourceManager.remove(allocationID);
    +           
    +           removePendingRequestWithException(allocationID, new 
CancellationException("Allocation " + allocationID + " cancelled"));
    +
    +           if (allocatedSlots.contains(allocationID)) {
    +                   Slot slot = allocatedSlots.get(allocationID);
    --- End diff --
    
    We could avoid the `contains` call by simply calling `get` and then compare 
against `null`.


> There may be allocatedSlots leak in SlotPool
> --------------------------------------------
>
>                 Key: FLINK-6434
>                 URL: https://issues.apache.org/jira/browse/FLINK-6434
>             Project: Flink
>          Issue Type: Bug
>          Components: Cluster Management
>            Reporter: shuai.xu
>            Assignee: shuai.xu
>            Priority: Major
>              Labels: flip-6
>
> If the call allocateSlot() from Execution to Slotpool timeout, the job will 
> begin to failover, but the pending request are still in SlotPool, if then a 
> new slot register to SlotPool, it may be fulfill the outdated pending request 
> and be added to allocatedSlots, but it will never be used and will never be 
> recycled.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to