dotnet/orleans

EventHub pulling agent checkpointer should detect out of bounds offset and recover

Open

#7,256 创建于 2021年9月2日

在 GitHub 查看
 (9 评论) (0 反应) (1 负责人)C# (10,777 star) (2,123 fork)batch import
help wanted

描述

It's possible for checkpoint offset(s) to become invalid when the pulling agents are offline and partition data in the event hub is expunged/expired. When this happens, one or more pulling agent(s) is/are suspended - and exceptions repeatedly thrown - until their data in the storage account table is cleared.

I would suggest that the pulling agent(s) reset their position, following the pattern used by EventProcessorOptions.InitialOffsetProvider -- e.g.

image

This would give optional per-partition recovery control and maximum flexibility while reusing available classes in the event hub package.

Example issue:

The supplied offset '10402431925616' is invalid. The last offset in the system is '489637839544' TrackingId:d1526603-0359-4a2a-a397-e1a14fb32774_B14, SystemTracker:180-UsageMessagingComponentManager, Timestamp:2021-08-31T21:16:42 Reference:24b9ac31-3526-42e9-831e-f4e16f99210b, TrackingId:3cc193a5-6919-475c-81a6-babc8b8796b0_B14, SystemTracker:eh-digitaltwin-dev-01:eventhub:ayla-in~21503|test_oisin, Timestamp:2021-08-31T21:16:43 TrackingId:673510f56c824f93ac79c22d4df41d65_G19, SystemTracker:gateway5, Timestamp:2021-08-31T21:16:43

System.ArgumentException:
   at Microsoft.Azure.EventHubs.Amqp.AmqpPartitionReceiver+<OnReceiveAsync>d__13.MoveNext (Microsoft.Azure.EventHubs, Version=4.3.2.0, Culture=neutral, PublicKeyToken=7e34167dcc6d6d8c)
   at System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at Microsoft.Azure.EventHubs.Amqp.AmqpPartitionReceiver+<OnReceiveAsync>d__13.MoveNext (Microsoft.Azure.EventHubs, Version=4.3.2.0, Culture=neutral, PublicKeyToken=7e34167dcc6d6d8c)
   at System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at System.Runtime.CompilerServices.ConfiguredTaskAwaitable`1+ConfiguredTaskAwaiter.GetResult (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at Microsoft.Azure.EventHubs.PartitionReceiver+<ReceiveAsync>d__30.MoveNext (Microsoft.Azure.EventHubs, Version=4.3.2.0, Culture=neutral, PublicKeyToken=7e34167dcc6d6d8c)
   at System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at System.Runtime.CompilerServices.TaskAwaiter`1.GetResult (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at Orleans.ServiceBus.Providers.EventHubAdapterReceiver+<GetQueueMessagesAsync>d__22.MoveNext (Orleans.Streaming.EventHubs, Version=2.0.0.0, Culture=neutral, PublicKeyToken=null)
   at System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification (System.Private.CoreLib, 
Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at System.Runtime.CompilerServices.TaskAwaiter`1.GetResult (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at Orleans.Streams.PersistentStreamPullingAgent+<ReadFromQueue>d__37.MoveNext (Orleans.Runtime, Version=2.0.0.0, Culture=neutral, PublicKeyToken=null)
   at System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at System.Runtime.CompilerServices.TaskAwaiter.ThrowForNonSuccess (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at System.Runtime.CompilerServices.TaskAwaiter`1.GetResult (System.Private.CoreLib, Version=5.0.0.0, Culture=neutral, PublicKeyToken=7cec85d7bea7798e)
   at Orleans.Internal.AsyncExecutorWithRetries+<ExecuteWithRetriesHelper>d__4`1.MoveNext (Orleans.Core, Version=2.0.0.0, Culture=neutral, PublicKeyToken=null)

贡献者指南

技术栈
csharp
领域
backenddata
议题类型
feature
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
4
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
3-5 days
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
stale
清晰度议题是否清楚说明期望改动、验收标准和下一步。
clear
前置要求
Familiarity with Azure EventHubsOrleans streaming concepts
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
30
研究方向
Examine the existing EventHubAdapterReceiver and PersistentStreamPullingAgent classes to understand current offset handling. Review the suggested pattern from EventProcessorOptions.InitialOffsetProvider in the Azure SDK. Investigate how to detect an out of bounds offset and reset it to a valid position, possibly by storing a flag or using a newer offset. Coordinate with the assignee (if any) to ensure the approach aligns with existing design.