TimerCallback.PerformTimerCallback内存分配

时间:2010-06-20 07:00:25

标签: c# wcf memory-management memory-profiling

我有一个WCF服务,我正在分析它的内存分配 我看到每个请求分配的 36%内存是通过TimerCallBack.PerformTimerCallback分配的,而后者又调用以下内容:
ExecutionContext.CreateCopy
ExecutionContext.Run
所有这些分配甚至都没有通过我的代码,只分配像Hahtable.bucket和ExecutionContextRunData这样的通用对象。

我在我的服务中使用Timer作为应用原因,我为每个请求创建一个带有更新布尔值的简单回调,然后立即处理它。可能有关系吗?

以下是相关的分配图:

<root> :   41 MB    (100.00%)
 System.Threading._TimerCallback::PerformTimerCallback static void (Object):   41 MB    (100.00%)
  System.Threading.ExecutionContext::CreateCopy System.Threading.ExecutionContext ():   30 MB    (73.96%)
  System.Threading.ExecutionContext::Run static void (System.Threading.ExecutionContext System.Threading.ContextCallback Object):   11 MB    (26.04%)
   System.Threading.ExecutionContext::RunInternal static void (System.Threading.ExecutionContext System.Threading.ContextCallback Object):  7.8 MB    (19.18%)
   System.Data.ProviderBase.DbConnectionFactory::PruneConnectionPoolGroups void (Object):  2.6 MB    (6.42%)
   System.Data.ProviderBase.DbConnectionPool::CleanupCallback void (Object):  183 kB    (0.44%)
    System.Data.ProviderBase.DbConnectionPoolGroup::ClearInternal bool (bool):  1.2 MB    (3.06%)
    System.Collections.Generic.Dictionary<T,U>::.ctor void (int32):  840 kB    (2.02%)
    System.Threading.ThreadPool::QueueUserWorkItem static bool (System.Threading.WaitCallback):  183 kB    (0.44%)
     System.Collections.Generic.Dictionary<T,U>::.ctor void (int32 <UNKNOWN>):  840 kB    (2.02%)
     System.Collections.Specialized.HybridDictionary::Add void (Object Object):  305 kB    (0.73%)
     System.Collections.Specialized.HybridDictionary::GetEnumerator void ():  255 kB    (0.61%)
     System.Collections.Specialized.ListDictionary::Add void (Object Object):  255 kB    (0.61%)
     System.Collections.Specialized.ListDictionary.NodeEnumerator::get_Current Object ():  204 kB    (0.49%)
     System.Threading.ThreadPool::QueueUserWorkItemHelper static bool (System.Threading.WaitCallback Object System.Threading.StackCrawlMark& bool):  183 kB    (0.44%)
      System.Collections.Generic.Dictionary<T,U>::Initialize void (int32):  840 kB    (2.02%)
      System.Threading._ThreadPoolWaitCallback::.ctor void (System.Threading.WaitCallback Object bool System.Threading.StackCrawlMark&):  135 kB    (0.32%)
       System.Threading.ExecutionContext::Capture static System.Threading.ExecutionContext (System.Threading.StackCrawlMark&):  135 kB    (0.32%)
        System.Runtime.Remoting.Messaging.LogicalCallContext::Clone Object ():   26 MB    (63.75%)
         System.Collections.Hashtable::.ctor void (int32 float32):   15 MB    (36.34%)
         System.Collections.Hashtable::GetEnumerator System.Collections.IDictionaryEnumerator ():  2.9 MB    (7.07%)
          System.Collections.Hashtable.bucket [] :   15 MB    (36.34%)
          System.Threading.ExecutionContext.ExecutionContextRunData :  7.8 MB    (19.18%)
          System.Collections.Hashtable :  4.5 MB    (11.10%)
          System.Threading.ExecutionContext :  4.3 MB    (10.53%)
          System.Runtime.Remoting.Messaging.LogicalCallContext :  3.8 MB    (9.24%)
          System.Collections.Hashtable.HashtableEnumerator :  2.9 MB    (7.07%)
          System.Collections.Generic.Dictionary<T,U>.Entry [] :  611 kB    (1.47%)
          System.Collections.Generic.Dictionary<T,U> :  560 kB    (1.34%)
          System.Collections.Specialized.ListDictionary :  305 kB    (0.73%)
          System.Collections.Specialized.ListDictionary.NodeEnumerator :  255 kB    (0.61%)
          System.Collections.Specialized.ListDictionary.DictionaryNode :  255 kB    (0.61%)
          System.Collections.Specialized.HybridDictionary :  255 kB    (0.61%)
          System.Int32 [] :  229 kB    (0.55%)
          System.Collections.DictionaryEntry :  204 kB    (0.49%)
          System.Threading._ThreadPoolWaitCallback :   48 kB    (0.12%)
           <bottom> :   41 MB    (100.00%)

所有这些分配来自哪里?这是正常的吗? 有没有办法降低一些?

1 个答案:

答案 0 :(得分:0)

为了完整性,问题是有人将MaxBufferSize和MaxBufferPoolSize值更改为以前值的1,000倍。