重试HttpClient不成功的请求

Question

我正在构建一个给出HttpContent对象的函数,它将发出请求并在失败时重试.但是我得到异常,说HttpContent对象在发出请求后被处理掉.无论如何都要复制或复制HttpContent对象,以便我可以发出多个请求.

 public HttpResponseMessage ExecuteWithRetry(string url, HttpContent content)
 {
  HttpResponseMessage result = null;
  bool success = false;
  do
  {
      using (var client = new HttpClient())
      {
          result = client.PostAsync(url, content).Result;
          success = result.IsSuccessStatusCode;
      }
  }
  while (!success);

 return result;
} 

// Works with no exception if first request is successful
ExecuteWithRetry("http://www.requestb.in/xfxcva" /*valid url*/, new StringContent("Hello World"));
// Throws if request has to be retried ...
ExecuteWithRetry("http://www.requestb.in/badurl" /*invalid url*/, new StringContent("Hello World"));

(显然我不会无限期地尝试,但上面的代码基本上就是我想要的).

它产生了这个例外

System.AggregateException: One or more errors occurred. ---> System.ObjectDisposedException: Cannot access a disposed object.
Object name: 'System.Net.Http.StringContent'.
   at System.Net.Http.HttpContent.CheckDisposed()
   at System.Net.Http.HttpContent.CopyToAsync(Stream stream, TransportContext context)
   at System.Net.Http.HttpClientHandler.GetRequestStreamCallback(IAsyncResult ar)
   --- End of inner exception stack trace ---
   at System.Threading.Tasks.Task.ThrowIfExceptional(Boolean includeTaskCanceledExceptions)
   at System.Threading.Tasks.Task`1.GetResultCore(Boolean waitCompletionNotification)
   at System.Threading.Tasks.Task`1.get_Result()
   at Submission#8.ExecuteWithRetry(String url, HttpContent content)

反正有没有复制HttpContent对象或重用它？

Answer 1

不考虑实现包装的重试功能,而是HttpClient考虑HttpClient使用在HttpMessageHandler内部执行重试逻辑的方法来构造.例如:

public class RetryHandler : DelegatingHandler
{
    // Strongly consider limiting the number of retries - "retry forever" is
    // probably not the most user friendly way you could respond to "the
    // network cable got pulled out."
    private const int MaxRetries = 3;

    public RetryHandler(HttpMessageHandler innerHandler)
        : base(innerHandler)
    { }

    protected override async Task<HttpResponseMessage> SendAsync(
        HttpRequestMessage request,
        CancellationToken cancellationToken)
    {
        HttpResponseMessage response = null;
        for (int i = 0; i < MaxRetries; i++)
        {
            response = await base.SendAsync(request, cancellationToken);
            if (response.IsSuccessStatusCode) {
                return response;
            }
        }

        return response;
    }
}

public class BusinessLogic
{
    public void FetchSomeThingsSynchronously()
    {
        // ...

        // Consider abstracting this construction to a factory or IoC container
        using (var client = new HttpClient(new RetryHandler(new HttpClientHandler())))
        {
            myResult = client.PostAsync(yourUri, yourHttpContent).Result;
        }

        // ...
    }
}

Answer 2

ASP.NET Core 2.1答案

ASP.NET Core 2.1直接添加了对Polly的支持.这UnreliableEndpointCallerService是一个HttpClient在其构造函数中接受a的类.失败的请求将以指数退避重试,以便下一次重试发生在前一次重试之后的指数级更长的时间内:

services
    .AddHttpClient<UnreliableEndpointCallerService>()
    .AddTransientHttpErrorPolicy(
        x => x.WaitAndRetryAsync(3, retryAttempt => TimeSpan.FromSeconds(Math.Pow(3, retryAttempt)));

另外,请考虑阅读我的博客文章"Optimally Configure HttpClientFactory".

其他平台答案

此实现使用Polly以指数退避重试,以便下一次重试发生在前一次重试之后的指数级更长的时间内.如果它也重试HttpRequestException或TaskCanceledException因超时异常.Polly比Topaz更容易使用.

public class HttpRetryMessageHandler : DelegatingHandler
{
    public HttpRetryMessageHandler(HttpClientHandler handler) : base(handler) {}

    protected override Task<HttpResponseMessage> SendAsync(
        HttpRequestMessage request,
        CancellationToken cancellationToken) =>
        Policy
            .Handle<HttpRequestException>()
            .Or<TaskCanceledException>()
            .OrResult<HttpResponseMessage>(x => !x.IsSuccessStatusCode)
            .WaitAndRetryAsync(3, retryAttempt => TimeSpan.FromSeconds(Math.Pow(3, retryAttempt)))
            .ExecuteAsync(() => base.SendAsync(request, cancellationToken));
}

using (var client = new HttpClient(new HttpRetryMessageHandler(new HttpClientHandler())))
{
    var result = await client.GetAsync("http://example.com");
}

Answer 3

The current answers won't work as expected in all cases, specifically in the very common case of request timeout (see my comments there).

In addition, they implement a very naive retry strategy - many times you'd want something a bit more sophosticated, such as exponential backoff (which is the default in the Azure Storage Client API).

I stumbled upon TOPAZ while reading a related blog post (also offering the misguided internal retry approach). Here's what I came up with:

// sample usage: var response = await RequestAsync(() => httpClient.GetAsync(url));
Task<HttpResponseMessage> RequestAsync(Func<Task<HttpResponseMessage>> requester)
{
    var retryPolicy = new RetryPolicy(transientErrorDetectionStrategy, retryStrategy);
    //you can subscribe to the RetryPolicy.Retrying event here to be notified 
    //of retry attempts (e.g. for logging purposes)
    return retryPolicy.ExecuteAsync(async () =>
    {
        HttpResponseMessage response;
        try
        {
            response = await requester().ConfigureAwait(false);
        }
        catch (TaskCanceledException e) //HttpClient throws this on timeout
        {
            //we need to convert it to a different exception
            //otherwise ExecuteAsync will think we requested cancellation
            throw new HttpRequestException("Request timed out", e);
        }
        //assuming you treat an unsuccessful status code as an error
        //otherwise just return the respone here
        return response.EnsureSuccessStatusCode(); 
    });
}

Note the requester delegate parameter. It should not be an HttpRequestMessage since you can't send the same request multiple times. As for the strategies, that depends on your use case. For example, a transient error detection strategy could be as simple as:

private sealed class TransientErrorCatchAllStrategy : ITransientErrorDetectionStrategy
{
    public bool IsTransient(Exception ex)
    {
        return true;
    }
}

As for the retry strategy, TOPAZ offers three options:

1. FixedInterval
2. Incremental
3. ExponentialBackoff

For example, here's the TOPAZ equivalent of what the Azure Client Storage Library uses for default:

int retries = 3;
var minBackoff = TimeSpan.FromSeconds(3.0);
var maxBackoff = TimeSpan.FromSeconds(120.0);
var deltaBackoff= TimeSpan.FromSeconds(4.0);
var strategy = new ExponentialBackoff(retries, minBackoff, maxBackoff, deltaBackoff);

For more information see http://msdn.microsoft.com/en-us/library/hh680901(v=pandp.50).aspx

EDIT Note that if your request contains an HttpContent object, you'll have to regenerate it every time as that will be disposed by HttpClient as well (thanks for catching that Alexandre Pepin). For example () => httpClient.PostAsync(url, new StringContent("foo"))).

Answer 4

复制StringContent可能不是最好的主意.但简单的修改可以解决问题.只需修改函数并在循环内创建StringContent对象,例如:

public HttpResponseMessage ExecuteWithRetry(string url, string contentString)
{
   HttpResponseMessage result = null;
   bool success = false;
   using (var client = new HttpClient())
   {
      do
      {
         result = client.PostAsync(url, new StringContent(contentString)).Result;
         success = result.IsSuccessStatusCode;
      }
      while (!success);
  }    

  return result;
} 

然后打电话给它

ExecuteWithRetry("http://www.requestb.in/xfxcva" /*valid url*/, "Hello World");

Answer 5

这就是我使用 polly 实现的目标。

努盖特

https://www.nuget.org/packages/Microsoft.Extensions.Http.Polly

https://www.nuget.org/packages/Polly

using Polly;
using Polly.Extensions.Http;

//// inside configure service
services.AddHttpClient("RetryHttpClient", c =>
{
    c.BaseAddress = new Uri($"{configuration["ExternalApis:MyApi"]}/");
    c.DefaultRequestHeaders.Add("Accept", "application/json");
    c.Timeout = TimeSpan.FromMinutes(5);
    c.DefaultRequestHeaders.ConnectionClose = true;

}).AddPolicyHandler(GetRetryPolicy());

//// add this method to give retry policy
private static IAsyncPolicy<HttpResponseMessage> GetRetryPolicy()
{
    return HttpPolicyExtensions
        //// 408,5xx
        .HandleTransientHttpError()
        //// 404
        .OrResult(msg => msg.StatusCode == HttpStatusCode.NotFound)
        //// 401
        .OrResult(msg => msg.StatusCode == HttpStatusCode.Unauthorized)
        //// Retry 3 times, with wait 1,2 and 4 seconds.
        .WaitAndRetryAsync(3, retryAttempt => TimeSpan.FromSeconds(Math.Pow(2, retryAttempt)));
}

Answer 6

这建立在接受的答案的基础上，但增加了传递重试次数的能力，并增加了为每个请求添加非阻塞延迟/等待时间的能力。它还使用 try catch 来确保在发生异常后重试继续发生。最后，我添加了代码以在 BadRequests 的情况下跳出循环，您不想多次重新发送相同的错误请求。

public class HttpRetryHandler : DelegatingHandler
{
    private int MaxRetries;
    private int WaitTime;

    public HttpRetryHandler(HttpMessageHandler innerHandler, int maxRetries = 3, int waitSeconds = 0)
        : base(innerHandler)
    {
        MaxRetries = maxRetries;
        WaitTime = waitSeconds * 1000; 
    }

    protected override async Task<HttpResponseMessage> SendAsync(HttpRequestMessage request, CancellationToken cancellationToken)
    {
        HttpResponseMessage response = null;
        for (int i = 0; i < MaxRetries; i++)
        {
            try
            {
                response = await base.SendAsync(request, cancellationToken);
                if (response.IsSuccessStatusCode)
                {
                    return response;
                }
                else if(response.StatusCode == HttpStatusCode.BadRequest)
                {
                    // Don't reattempt a bad request
                    break; 
                }
            }
            catch
            {
                // Ignore Error As We Will Attempt Again
            }
            finally
            {
                response.Dispose(); 
            }

            if(WaitTime > 0)
            {
                await Task.Delay(WaitTime);
            }
        }

        return response;
    }
}

}