A colleague of mine noticed that I had these peppered through some consumer daemons:
for m in self.iterconsume():
if self.closed:
break

To handle signals for a graceful daemon close, we close the channel asynchronously. Which will cause an exception in the iterconsume or when it tries to fetch the next message. I put the flag there to prevent some wild exception from bleeding through. As a test, we modified the pyamqplib backend to raise a StopIteration when the channel is closed (sorry for the lack of a diff):
def consume(self, limit=None):
"""Returns an iterator that waits for one message at a time."""
for total_message_count in count():
if limit and total_message_count >= limit:
raise StopIteration

        if not self.channel.is_open:
            raise StopIteration

        self.channel.wait()
        yield True

There is also the case where an amqp exception can be raised by the channel.wait call, but I defer the handling of that up to you (catch AMQPException and raise StopIteration or let the Exception bleed through; I am for the latter).

Let me know if you need any other information on this.