[V2] DirectMethod from cloud becomes unavailable after device side internet reconnection

[yfakariya]

Environment

• Service Client: Microsoft.Azure.Devices v1.16.0-preview1
• Docker images:
  • edgeHub: microsoft/azureiotedge-hub:1.0-preview
  • edgeAgent: microsoft/azureiotedge-agent:1.0-preview
  • tempSensor: microsoft/azureiotedge-simulated-temperature-sensor:1.0-preview

Symptom

I tested preview IoT Edge for internet connectivity scenario with AMQP protocol. For D2C telemetry messaging, Edge Hub's "storeAndForward" feature worked perfect, it buffered unsent message while it was disconnected. However, after reconnection, I could not call direct method for the device with error code 404103 and following stack trace.

Timed out waiting for device to connect.\",\"info\":{\"timeout\":\"00:00:00\"},\"timestampUtc\":\"2018-03-29T10:41:59.9932398Z\"}","ExceptionMessage":""} not registered
   at Microsoft.Azure.Devices.HttpClientHelper.<ExecuteAsync>d__35.MoveNext()
--- End of stack trace from previous location where exception was thrown ---
   at System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw()
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at Microsoft.Azure.Devices.HttpClientHelper.<PostAsync>d__26`2.MoveNext()
--- End of stack trace from previous location where exception was thrown ---
   at System.Runtime.ExceptionServices.ExceptionDispatchInfo.Throw()
   at System.Runtime.CompilerServices.TaskAwaiter.HandleNonSuccessAndDebuggerNotification(Task task)
   at System.Runtime.CompilerServices.ConfiguredTaskAwaitable`1.ConfiguredTaskAwaiter.GetResult()
   at InvokeDirectMethod.Program.<Main>d__0.MoveNext() in MYDIRECTORY\InvokeDirectMethod\InvokeDirectMethod\Program.cs:line

Expected behavior

• DirectMethod is successful after device side internet reconnection.

Actual behavior

• 404103 error until device reboot.

Sidenote

It looks same problem as this. I think this issue is more critical than vanilla device client because

• We cannot detect internet(AMQP) disconnection because an IoT Edge hides underlying internet connectivity, so we cannot react such condition.
• Modules (docker container) hard to initiate host device reboot.

[darobs]

Thank you for reporting this, @yfakariya,

I added to our backlog to track this bug.

[dtjensen]

I am getting this error when attempting to send a direct method. Is there a workaround or a fix? It happens when sending from the portal or from my own service console app.

[trbenning]

Any updates on this? We're seeing the same issue in our environment as well.

[avranju]

I believe this should get addressed by GA. @varunpuranik can confirm.

[varunpuranik]

Yes, that is correct - this issue should be fixed by GA.