使用Polly包重试时发生NullReferenceException错误

Question

我们有.net 6服务，我们使用polly包(Polly，Version=7.0.0.0)重试瞬态api调用。我们看到System.NullReferenceException: Object没有设置为一个对象错误的实例。尽管重试似乎有效，但我认为代码中遗漏了一些东西。

我们能够看到用于重试的日志：

NotExtended delaying for 1000 ms, then making retry 1 {...data..}
NotExtended delaying for 1000 ms, then making retry 2 {...data..}
NotExtended delaying for 1000 ms, then making retry 3 {...data..}

因为我对Polly的工作方式很陌生，所以不知道我们在下面的代码中到底遗漏了什么。任何建议都会很有帮助。提前谢谢。

P.S.：我在粘贴代码片段，在这里我们使用polly包，而不是与polly实现无关的代码行，以避免混淆。

不确定此错误是否发生在很少的api调用或所有这些调用中。

using Polly;
using Polly.Extensions.Http;
using Polly.Wrap;

services.AddHttpClient<IApiService, ApiService>()
        .AddPolicyHandler(GetPollyRetryPolicy(builder.services));


private static AsyncPolicyWrap<HttpResponseMessage> GetPollyRetryPolicy(IServiceCollection services)
{
    var asyncPolicy = HttpPolicyExtensions
                        .HandleTransientHttpError()
                        .WaitAndRetryAsync(3, retryAttempt => TimeSpan.FromSeconds(retryAttempt),
                            onRetryAsync: async (outcome, timespan, retryAttempt, context) => 
                            {
                                var serviceProvider = services.BuildServiceProvider();
                                var data = new                                                      
                                {
                                    url = outcome.Result?.RequestMessage?.RequestUri?.ToString(),
                                    method = outcome.Result?.RequestMessage?.Method?.Method,
                                    resultContent = await outcome.Result?.Content?.ReadAsStringAsync()
                                };
                                serviceProvider.GetService<ILogger>()
                                ?.Warn($"{outcome.Result.StatusCode} delaying for {timespan.TotalMilliseconds} ms, " +
                                   $"then making retry {retryAttempt}", data: data, ex: outcome.Exception);
                            });

    return Policy.TimeoutAsync(60).WrapAsync(asyncPolicy);
}

我们正在接收的错误如下：

System.NullReferenceException: Object reference not set to an instance of an object.

--- End of stack trace from previous location ---
   at Polly.Retry.AsyncRetryEngine.ImplementationAsync[TResult](Func`3 action, Context context, CancellationToken cancellationToken, ExceptionPredicates shouldRetryExceptionPredicates, ResultPredicates`1 shouldRetryResultPredicates, Func`5 onRetryAsync, Int32 permittedRetryCount, IEnumerable`1 sleepDurationsEnumerable, Func`4 sleepDurationProvider, Boolean continueOnCapturedContext)
   at Polly.AsyncPolicy`1.ExecuteAsync(Func`3 action, Context context, CancellationToken cancellationToken, Boolean continueOnCapturedContext)
   at Polly.Wrap.AsyncPolicyWrapEngine.<>c__DisplayClass2_0`1.<<ImplementationAsync>b__0>d.MoveNext()
--- End of stack trace from previous location ---
   at Polly.Timeout.AsyncTimeoutEngine.ImplementationAsync[TResult](Func`3 action, Context context, CancellationToken cancellationToken, Func`2 timeoutProvider, TimeoutStrategy timeoutStrategy, Func`5 onTimeoutAsync, Boolean continueOnCapturedContext)
   at Polly.Timeout.AsyncTimeoutEngine.ImplementationAsync[TResult](Func`3 action, Context context, CancellationToken cancellationToken, Func`2 timeoutProvider, TimeoutStrategy timeoutStrategy, Func`5 onTimeoutAsync, Boolean continueOnCapturedContext)
   at Polly.AsyncPolicy.ExecuteAsync[TResult](Func`3 action, Context context, CancellationToken cancellationToken, Boolean continueOnCapturedContext)
   at Polly.Wrap.AsyncPolicyWrapEngine.ImplementationAsync[TResult](Func`3 func, Context context, CancellationToken cancellationToken, Boolean continueOnCapturedContext, IAsyncPolicy outerPolicy, IAsyncPolicy`1 innerPolicy)
   at Polly.AsyncPolicy`1.ExecuteAsync(Func`3 action, Context context, CancellationToken cancellationToken, Boolean continueOnCapturedContext)
   at Microsoft.Extensions.Http.PolicyHttpMessageHandler.SendAsync(HttpRequestMessage request, CancellationToken cancellationToken)
   at Microsoft.Extensions.Http.Logging.LoggingScopeHttpMessageHandler.SendAsync(HttpRequestMessage request, CancellationToken cancellationToken)
   at System.Net.Http.HttpClient.<SendAsync>g__Core|83_0(HttpRequestMessage request, HttpCompletionOption completionOption, CancellationTokenSource cts, Boolean disposeCts, CancellationTokenSource pendingRequestsCts, CancellationToken originalCancellationToken)

Answer 1

正如Stephen Cleary在他的评论中提到的:在每次重试中构建一个全新的服务提供商是非常奇怪的。这不仅是奇怪的，而且根本没有必要。

提供对AddPolicyHandler has an overload的访问的IServiceProvider

services.AddHttpClient<IApiService, ApiService>()
        .AddPolicyHandler((provider, _) => GetPollyRetryPolicy(provider))

有了这一技术，您只需在ILogger委托中请求这样的onRetryAsync：

var logger = provider.GetRequiredService<ILogger>();

我还建议使用静态Wrap方法而不是实例方法：

Policy.WrapAsync(Policy.TimeoutAsync(60), asyncPolicy);

Here I have detailed the differences between the alternatives。

还请注意，订购事项：

• 超时值外部，重试内部：
  • 超时操作为全局的、超疼痛的
  • 外部，超时为本地约束，因此它为每次重试尝试

  重新设置自身。