SqlException 事务在通信缓冲区资源上死锁

Question

我有一个必须执行大量数据库查询的服务器进程，它使用 TPL 并行运行东西。今年全年它一直运行良好，直到今天它在 30 分钟内崩溃了两次，但以下情况除外：

事务（进程 ID 89）在与另一个进程的通信缓冲区资源上死锁，并已被选为死锁牺牲品。重新运行事务。

数据库被配置为记录任何死锁，但它没有记录任何东西，所以看起来这个死锁似乎只发生在客户端？

除了一个 msdn forum post that doesn't provide any information 之外，我找不到对此异常的任何引用。

以前有人见过这个异常吗？或者知道我可以做些什么来了解更多信息？

---> System.AggregateException: One or more errors occurred. 

---> System.Data.SqlClient.SqlException: Transaction (Process ID 89) was deadlocked on communication buffer resources with another process and has been chosen as the deadlock victim. Rerun the transaction.
   at System.Data.SqlClient.SqlConnection.OnError(SqlException exception, Boolean breakConnection)
   at System.Data.SqlClient.TdsParser.ThrowExceptionAndWarning()
   at System.Data.SqlClient.TdsParser.Run(RunBehavior runBehavior, SqlCommand cmdHandler, SqlDataReader dataStream, BulkCopySimpleResultSet bulkCopyHandler, TdsParserStateObject stateObj)
   at System.Data.SqlClient.SqlDataReader.HasMoreRows()
   at System.Data.SqlClient.SqlDataReader.ReadInternal(Boolean setTimeout)
   at App.CoreEngine.V5.DataAccess.SqlReader.Read(String readerDescription) in C:\SourceCode\AppV1\Releases\Libraries\CoreEngine\CoreEngine.V5\DataAccess\SqlReader.cs:line 121
   at App.CoreEngine.V5.DataAccess.DataContext.ExecuteQuery(PtQuery query, ValueStore`1 store, String readerDescription) in C:\SourceCode\AppV1\Releases\Libraries\CoreEngine\CoreEngine.V5\DataAccess\DataContext.cs:line 328
   at App.CoreEngine.V5.DataAccess.DataContext.<>c__DisplayClass12.<GetCalculatedDataForCompare>b__f(Object _) in C:\SourceCode\AppV1\Releases\Libraries\CoreEngine\CoreEngine.V5\DataAccess\DataContext.cs:line 267
   at System.Threading.Tasks.Task.Execute()
   --- End of inner exception stack trace ---
   at System.Threading.Tasks.Task.WaitAll(Task[] tasks, Int32 millisecondsTimeout, CancellationToken cancellationToken)
   at App.CoreEngine.V5.DataAccess.DataContext.GetCalculatedDataForCompare() in C:\SourceCode\AppV1\Releases\Libraries\CoreEngine\CoreEngine.V5\DataAccess\DataContext.cs:line 276
   at System.Threading.Tasks.Task`1.InvokeFuture(Object futureAsObj)
   at System.Threading.Tasks.Task.Execute()
   --- End of inner exception stack trace ---
   at System.Threading.Tasks.Task.Wait(Int32 millisecondsTimeout, CancellationToken cancellationToken)
   at System.Threading.Tasks.Task`1.get_Result()
   at App.CoreEngine.V5.DataAccess.DataContext.get_CalcCompareData() in C:\SourceCode\AppV1\Releases\Libraries\CoreEngine\CoreEngine.V5\DataAccess\DataContext.cs:line 389
   at App.CoreEngine.V5.Calculation.CalculationEngine.Run() in C:\SourceCode\AppV1\Releases\Libraries\CoreEngine\CoreEngine.V5\Calculation\CalculationEngine.cs:line 243
   at App.CoreEngine.V5.Processor.Milestone.BatchRunner.Run() in C:\SourceCode\AppV1\Releases\Libraries\CoreEngine\CoreEngine.V5\Processor\Milestone\BatchRunner.cs:line 171

---> (Inner Exception #0) System.AggregateException: One or more errors occurred. 
---> System.Data.SqlClient.SqlException: Transaction (Process ID 89) was deadlocked on communication buffer resources with another process and has been chosen as the deadlock victim. Rerun the transaction.
   at System.Data.SqlClient.SqlConnection.OnError(SqlException exception, Boolean breakConnection)
   at System.Data.SqlClient.TdsParser.ThrowExceptionAndWarning()
   at System.Data.SqlClient.TdsParser.Run(RunBehavior runBehavior, SqlCommand cmdHandler, SqlDataReader dataStream, BulkCopySimpleResultSet bulkCopyHandler, TdsParserStateObject stateObj)
   at System.Data.SqlClient.SqlDataReader.HasMoreRows()
   at System.Data.SqlClient.SqlDataReader.ReadInternal(Boolean setTimeout)
   at App.CoreEngine.V5.DataAccess.SqlReader.Read(String readerDescription) in C:\SourceCode\Releases\Libraries\CoreEngine\CoreEngine.V5\DataAccess\SqlReader.cs:line 121
   at App.CoreEngine.V5.DataAccess.DataContext.ExecuteQuery(PtQuery query, ValueStore`1 store, String readerDescription) in C:\SourceCode\AppV1\Releases\Libraries\CoreEngine\CoreEngine.V5\DataAccess\DataContext.cs:line 328
   at App.CoreEngine.V5.DataAccess.DataContext.<>c__DisplayClass12.<GetCalculatedDataForCompare>b__f(Object _) in C:\SourceCode\AppV1\Releases\Libraries\CoreEngine\CoreEngine.V5\DataAccess\DataContext.cs:line 267
   at System.Threading.Tasks.Task.Execute()
   --- End of inner exception stack trace ---
   at System.Threading.Tasks.Task.WaitAll(Task[] tasks, Int32 millisecondsTimeout, CancellationToken cancellationToken)
   at CoreEngine.V5.DataAccess.DataContext.GetCalculatedDataForCompare() in C:\SourceCode\AppV1\Releases\Libraries\CoreEngine\CoreEngine.V5\DataAccess\DataContext.cs:line 276
   at System.Threading.Tasks.Task`1.InvokeFuture(Object futureAsObj)
   at System.Threading.Tasks.Task.Execute()

编辑 - 我刚刚再次检查了发生这种情况的服务器是否已打开 dbcc 1222，我运行 dbcc tracestatus 并得到：

TraceFlag   Status  Global  Session
1222    1   1   0
3605    1   1   0

日志中没有报告死锁

Answer 1

我的猜测是执行计划现在使用并行性，而之前它没有达到成本阈值。

在查询中尝试 MAXDOP 1

评论后编辑

您还需要跟踪标志 1204。

TF 1222 给出了死锁图，但是对于这种“通信缓冲区资源”死锁，可能不会涉及 2 个对象（我猜这里不是索引/表冲突）。见http://msdn.microsoft.com/en-us/library/ms178104.aspx

还有未记录的 TF 1205，它在错误日志中提供了更多信息

Answer 2

您是否可以尝试以下任何一种方法：

• 让服务器端跟踪运行以捕获死锁事件
• 应用CU1并尝试Retrieving Deadlock Graphs with SQL Server 2008 Extended Events
• 打开另一个死锁捕获跟踪标志，1205

Answer 3

在我的例子中，这个过程是删除一些记录，然后在表中插入记录，它是在多线程环境中发生的。

我必须在用于删除的列上的表上添加一个非聚集索引，这解决了我的问题。