服务结构无状态标识服务器3服务在负载下失败

Question

我们正在将基于身份服务器3 (https://github.com/IdentityServer/IdentityServer3)的无状态服务迁移到服务结构中。

该项目在本地开发框和正常开发负载下的生产集群上运行良好，但当以每秒20-30个请求投入生产时，它很快就会停止响应请求，ARR中配置的健康检查也会变得不健康。

该服务由IIS (应用程序请求路由)群集负责SSL卸载。

标识服务器日志输出以下两个错误，一个似乎与入站请求有关，另一个与用于身份服务器持久性的出站请求有关。

Microsoft.WindowsAzure.Storage.StorageException: The client could not finish the operation within specified timeout. ---> System.TimeoutException: The client could not finish the operation within specified timeout.
   --- End of inner exception stack trace ---
   at Microsoft.WindowsAzure.Storage.Core.Executor.Executor.EndExecuteAsync[T](IAsyncResult result) in c:\Program Files (x86)\Jenkins\workspace\release_dotnet_master\Lib\ClassLibraryCommon\Core\Executor\Executor.cs:line 50
   at Microsoft.WindowsAzure.Storage.Core.Util.AsyncExtensions.<>c__DisplayClass1`1.<CreateCallback>b__0(IAsyncResult ar) in c:\Program Files (x86)\Jenkins\workspace\release_dotnet_master\Lib\ClassLibraryCommon\Core\Util\AsyncExtensions.cs:line 66
--- End of stack trace from previous location where exception was thrown ---
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter`1.GetResult()
   at Collabco.Myday.Identity.IdSvr.BaseStore`1.<ExecuteQueryAsync>d__21.MoveNext() in C:\Dev\myday-identity\IdentityServer\IdSvr\BaseStore.cs:line 258
--- End of stack trace from previous location where exception was thrown ---
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter`1.GetResult()
   at Collabco.Myday.Identity.IdSvr.ScopeStore.<GetScopesAsync>d__3.MoveNext() in C:\Dev\myday-identity\IdentityServer\IdSvr\ScopeStore.cs:line 43
--- End of stack trace from previous location where exception was thrown ---
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at IdentityServer3.Core.Endpoints.DiscoveryEndpointController.<GetConfiguration>d__11.MoveNext() in c:\local\identity\server3\IdentityServer3\source\Core\Endpoints\Connect\DiscoveryEndpointController.cs:line 73
--- End of stack trace from previous location where exception was thrown ---
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at System.Threading.Tasks.System.Web.Http910180.TaskHelpersExtensions.<CastToObject>d__3`1.MoveNext() in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
--- End of stack trace from previous location where exception was thrown ---
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at System.Web.Http.Controllers.ApiControllerActionInvoker.<InvokeActionAsyncCore>d__0.MoveNext() in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
--- End of stack trace from previous location where exception was thrown ---
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at System.Web.Http.Controllers.ActionFilterResult.<ExecuteAsync>d__2.MoveNext() in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
--- End of stack trace from previous location where exception was thrown ---
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at System.Web.Http.Dispatcher.HttpControllerDispatcher.<SendAsync>d__1.MoveNext() in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
Request Information
RequestID:
RequestDate:
StatusMessage:

我们看到的另一个例外是：

System.ObjectDisposedException: Cannot access a disposed object.
Object name: 'System.Net.HttpListenerRequest'.
   at System.Net.HttpListenerRequest.CheckDisposed()
   at System.Net.HttpListenerRequest.get_LocalEndPoint()
   at System.Net.HttpListenerRequest.get_IsLocal()
   at Microsoft.Owin.Host.HttpListener.RequestProcessing.OwinHttpListenerContext.GetServerIsLocal()
   at Microsoft.Owin.Host.HttpListener.RequestProcessing.CallEnvironment.get_ServerIsLocal()
   at Microsoft.Owin.Host.HttpListener.RequestProcessing.CallEnvironment.PropertiesTryGetValue(String key, Object& value)
   at Microsoft.Owin.Host.HttpListener.RequestProcessing.CallEnvironment.TryGetValue(String key, Object& value)
   at Microsoft.Owin.OwinContext.Get[T](String key) in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
   at System.Web.Http.Owin.OwinHttpRequestContext.get_IsLocal() in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
   at System.Web.Http.Owin.OwinHttpRequestContext.get_IncludeErrorDetail() in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
   at System.Net.Http.System.Web.Http910180.HttpRequestMessageExtensions.CreateErrorResponse(HttpRequestMessage request, HttpStatusCode statusCode, Func`2 errorCreator) in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
   at System.Web.Http.ExceptionHandling.DefaultExceptionHandler.Handle(ExceptionHandlerContext context) in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
   at System.Web.Http.ExceptionHandling.DefaultExceptionHandler.HandleAsync(ExceptionHandlerContext context, CancellationToken cancellationToken) in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
   at System.Web.Http.ExceptionHandling.LastChanceExceptionHandler.HandleAsync(ExceptionHandlerContext context, CancellationToken cancellationToken) in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
   at System.Web.Http.ExceptionHandling.ExceptionHandlerExtensions.<HandleAsyncCore>d__0.MoveNext() in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
--- End of stack trace from previous location where exception was thrown ---
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at System.Web.Http.Dispatcher.HttpControllerDispatcher.<SendAsync>d__1.MoveNext() in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
--- End of stack trace from previous location where exception was thrown ---
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at System.Web.Http.Owin.PassiveAuthenticationMessageHandler.<SendAsync>d__0.MoveNext() in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0
--- End of stack trace from previous location where exception was thrown ---
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess(Task task)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at System.Web.Http.HttpServer.<SendAsync>d__0.MoveNext() in c:\local\identity\server3\IdentityServer3\source\Core\Configuration\DiscoveryOptions.cs:line 0

经过大约5-10分钟后，清除负载，没有回收任何东西，服务恢复活力。服务结构也不会在不稳定期间检测到任何故障。

有什么想法吗？

Answer 1

解决办法是：

1. 确保超时异常被捕获、处理和记录，在某些情况下没有。
2. 调整重试策略，使重试更加紧密(延迟100 10 )，并且限制在不超过10秒。还实现了重试日志记录。实际上，到目前为止我还没有再试过一次。
3. 将表服务端点上的连接限制提高到1000。

最重要的因素似乎是数字3，虽然我首先做的实现注意到了一些改进，这是很明显的。

为此，我在存储库类(标识服务器术语中的基本存储类)上使用了一个静态构造函数，其代码依赖于StorageAccount实例。除了最后一行之外，所有的一切都已经存在。

        var tableServicePoint = System.Net.ServicePointManager.FindServicePoint(storageAccount.TableEndpoint);
        tableServicePoint.UseNagleAlgorithm = false;
        tableServicePoint.Expect100Continue = false;
        tableServicePoint.ConnectionLimit = 1000;

以下文章对此有所帮助：https://azure.microsoft.com/en-gb/documentation/articles/storage-performance-checklist/

总之，我的结论是默认的连接限制(2或10个不同的文档冲突)导致对表存储的请求排队并最终超时，最终导致服务崩溃。

要确认设置连接限制在使用蔚蓝web应用程序时是不必要的，因此服务fabric / owin自宿主项目的工作方式与使用影响蔚蓝存储的连接的方式有所不同。