如何解决SharePoint CSOM HTTP请求错误429？已执行官方方案仍有请求被拦截

Let’s dig into the nuanced, easy-to-miss factors that might still be triggering throttling even after you’ve followed the core guidance. SharePoint’s throttling logic can be tricky, especially when dealing with desktop add-ins that span multiple Office apps and interact with SharePoint under user credentials.

• Reuse ClientContext Instances Consistently
  A common pitfall is spinning up a new ClientContext for every single request. Each new context creates a fresh connection, adding unnecessary overhead and increasing your odds of hitting throttling limits. Instead, reuse the same context for a session of operations, and properly dispose of it when finished. Example:

  using (var context = new ClientContext(siteUrl))
  {
      context.Credentials = new SharePointOnlineCredentials(username, securePassword);
      // Batch multiple load operations into one ExecuteQuery call
      var web = context.Web;
      var targetList = web.Lists.GetByTitle("Project Documents");
      var items = targetList.GetItems(CamlQuery.CreateAllItemsQuery(50)); // Limit page size
      
      context.Load(web, w => w.Title);
      context.Load(targetList, l => l.Title);
      context.Load(items, i => i.Include(item => item["FileLeafRef"], item => item["Modified"]));
      
      context.ExecuteQuery();
      // Process your data here
  }
  

• Optimize Batch Processing & Pagination
  Even with batching, pulling too much data in one go can trigger throttling. Use pagination in CamlQuery to cap results at 50-100 items per request, and avoid running parallel batches under the same user context. For bulk updates, use ClientContext.Batch to group multiple write operations into a single request instead of sending them one by one.

• Refine Retry Logic with Retry-After Headers
  If you’re already retrying failed requests, make sure you’re honoring the Retry-After header included in the 429 response. Blindly retrying with fixed intervals will only worsen throttling. Combine exponential backoff with this header to give SharePoint time to recover. Example:

  int retryAttempts = 0;
  const int maxRetries = 4;
  bool requestSucceeded = false;
  
  while (!requestSucceeded && retryAttempts < maxRetries)
  {
      try
      {
          context.ExecuteQuery();
          requestSucceeded = true;
      }
      catch (WebException ex)
      {
          var httpResponse = ex.Response as HttpWebResponse;
          if (httpResponse?.StatusCode == HttpStatusCode.TooManyRequests)
          {
              int retryAfter = int.TryParse(httpResponse.Headers["Retry-After"], out var value) ? value : 5;
              // Exponential backoff: multiply wait time by 2 each retry
              int waitSeconds = retryAfter * (int)Math.Pow(2, retryAttempts);
              Thread.Sleep(waitSeconds * 1000);
              retryAttempts++;
          }
          else
          {
              // Not a throttling error, rethrow
              throw;
          }
      }
  }
  

• Check for Cross-App Concurrent Usage
  Your add-in runs across Word, Excel, PowerPoint, and Outlook—if a user has multiple Office apps open with your add-in active, that’s multiple concurrent sessions hitting SharePoint under the same user account. SharePoint enforces per-user concurrency limits, so add logic to restrict concurrent operations across app instances, or warn users to avoid multi-app usage during heavy tasks.

• Cache Static/Infrequent Data Locally
  If your add-in repeatedly fetches the same data (like list metadata, site settings), cache it locally (e.g., in the add-in’s storage or a local database) instead of hitting SharePoint every time. Only refresh the cache when the user triggers a manual refresh or after a set time window.

• Verify Tenant-Level Throttling Policies
  If none of the above fixes work, it’s possible your tenant has custom throttling policies applied, or your add-in’s request pattern is flagged as suspicious. Reach out to Microsoft 365 support to review your tenant’s throttling logs—they can pinpoint exactly which requests are being blocked and why.

内容的提问来源于stack exchange，提问作者Bruno Leitão