[ServiceBus] AMQP LinkDetach to be considered as retryable error

[yunhaoling]

10 mins idle timeout (no activity on a link) would make the service proactively shutdown a link. This seems to be a common scenario where we should not expect there's always messaging activity happening.

In current condition based error handling, LinkDetach (triggered by the 10mins idle timeout) would fall into the category of ServiceBusError.
However, to recovery from the error is pretty straight forward by re-establishing the link.
We could do this internally to simplify user code.

• need cross team check

---

Error logging

Traceback (most recent call last):#12
File "/usr/local/lib/python3.9/site-packages/azure/servicebus/_base_handler.py", line 293, in _do_retryable_operation#012
return operation(**kwargs)#12
File "/usr/local/lib/python3.9/site-packages/azure/servicebus/_servicebus_sender.py", line 238, in _send#012
self._handler.send_message(message.message)#12
File "/usr/local/lib/python3.9/site-packages/uamqp/client.py", line 725, in send_message#012
running = self.do_work()#12
File "/usr/local/lib/python3.9/site-packages/uamqp/client.py", line 395, in do_work#012
if not self.client_ready():#12
File "/usr/local/lib/python3.9/site-packages/uamqp/client.py", line 379, in client_ready#012
if not self._client_ready():#12
File "/usr/local/lib/python3.9/site-packages/uamqp/client.py", line 527, in _client_ready#012
if self.message_handler.get_state() == constants.MessageSenderState.Error:#12
File "/usr/local/lib/python3.9/site-packages/uamqp/sender.py", line 184, in get_state#012
raise self._error#012uamqp.errors.LinkDetach: ErrorCodes.LinkDetachForced: The link 'G6:3354839:sender-link-4e63dd77-7eaf-4d6d-8448-299b5161661b' is force detached. Code: publisher(link11766). Details: AmqpMessagePublisher.IdleTimerExpired: Idle timeout: 00:10:00.#12#012

[AndreCimander]

Was this already fixed in 7.0.1, by chance? We encountered this issue today with the current master of Celery on servicebus 7.0.0

[yunhaoling]

hey @AndreCimander , sorry I didn't have the time to work on this one recently, will start once I get some free cycles.
btw, could you give me more context on master of Celery, I actually don't know what it is.

and before the internal improvement is made, I recommend you to wrap your code with your own retry logic.
thanks for you patience in advance😊

[yunhaoling]

PR merged, this is gonna be included in our 7.0.2 release.