See PR 39307.

In python our send_messages in client.py is wrapped in an amqp level do_retryable_operation. On the other hand our receive operations are not, they only have the capability to be retried at the sdk level. Due to this we at this time cannot implement link-reattach (it would only work for send) without refactoring more code and extensively testing.

Overall Ideas of implementation:

• Wrap receive in double retry as well to allow for link re-attach on receiving operations

Follow up questions:

• Where should we do this?
• How would servicebus sessions impact this?
• What impact would this have on perf?
• uAMQP cannot do this -- would need to split out receive into the _transport files