How to pause and resume consumption gracefully in rabbitmq, pika python

问题

I'm using basic_consume() for receiving messages and basic_cancel for canceling consuming, but there is a problem.

Here is the code of pika.channel

 def basic_consume(self, consumer_callback, queue='', no_ack=False,
                      exclusive=False, consumer_tag=None):
        """Sends the AMQP command Basic.Consume to the broker and binds messages
        for the consumer_tag to the consumer callback. If you do not pass in
        a consumer_tag, one will be automatically generated for you. Returns
        the consumer tag.

        For more information on basic_consume, see:
        http://www.rabbitmq.com/amqp-0-9-1-reference.html#basic.consume

        :param method consumer_callback: The method to callback when consuming
        :param queue: The queue to consume from
        :type queue: str or unicode
        :param bool no_ack: Tell the broker to not expect a response
        :param bool exclusive: Don't allow other consumers on the queue
        :param consumer_tag: Specify your own consumer tag
        :type consumer_tag: str or unicode
        :rtype: str

        """
        self._validate_channel_and_callback(consumer_callback)

        # If a consumer tag was not passed, create one
        consumer_tag = consumer_tag or 'ctag%i.%s' % (self.channel_number,
                                                      uuid.uuid4().get_hex())

        if consumer_tag in self._consumers or consumer_tag in self._cancelled:
            raise exceptions.DuplicateConsumerTag(consumer_tag)

        self._consumers[consumer_tag] = consumer_callback
        self._pending[consumer_tag] = list()
        self._rpc(spec.Basic.Consume(queue=queue,
                                     consumer_tag=consumer_tag,
                                     no_ack=no_ack,
                                     exclusive=exclusive),
                           self._on_eventok,
                           [(spec.Basic.ConsumeOk,
                             {'consumer_tag': consumer_tag})])

        return consumer_tag

def basic_cancel(self, callback=None, consumer_tag='', nowait=False):
        """This method cancels a consumer. This does not affect already
        delivered messages, but it does mean the server will not send any more
        messages for that consumer. The client may receive an arbitrary number
        of messages in between sending the cancel method and receiving the
        cancel-ok reply. It may also be sent from the server to the client in
        the event of the consumer being unexpectedly cancelled (i.e. cancelled
        for any reason other than the server receiving the corresponding
        basic.cancel from the client). This allows clients to be notified of
        the loss of consumers due to events such as queue deletion.

        :param method callback: Method to call for a Basic.CancelOk response
        :param str consumer_tag: Identifier for the consumer
        :param bool nowait: Do not expect a Basic.CancelOk response
        :raises: ValueError

        """
        self._validate_channel_and_callback(callback)
        if consumer_tag not in self.consumer_tags:
            return
        if callback:
            if nowait is True:
                raise ValueError('Can not pass a callback if nowait is True')
            self.callbacks.add(self.channel_number,
                               spec.Basic.CancelOk,
                               callback)
        self._cancelled.append(consumer_tag)
        self._rpc(spec.Basic.Cancel(consumer_tag=consumer_tag,
                                    nowait=nowait),
                  self._on_cancelok,
                  [(spec.Basic.CancelOk,
                    {'consumer_tag': consumer_tag})] if nowait is False else [])

As you can see every time I'm cancelling consumption consumer_tag is added to _canceled list. And if I would use this tag in basic_consume again the duplicateConsumer exception will be raised. Well, I could use a new consumer_tag every time, but in fact I'm not. Because sooner or later generated tag would be exactly match some of the previous ones.

How should I pause and resume consumption gracefully in pika?

回答1:

Is there a reason why you define your own consumer_tags? You can pass an empty string and let RabbitMQ generate consumer tags for you. The reply from basic.consume, which is basic.consume-ok will return the generated consumer_tag, so you can use it later to stop consuming.

See: http://www.rabbitmq.com/amqp-0-9-1-reference.html#basic.consume-ok

回答2:

That looks like Pika is doing more than it should - it doesn't need to create a consumer tag if one is not supplied (the server will) and it also doesn't need to watch for duplicated consumer tags (resuming with the same tag is supported by the server).

So I'm not sure how to do this with Pika - file a bug I suppose.

来源：https://stackoverflow.com/questions/20651111/how-to-pause-and-resume-consumption-gracefully-in-rabbitmq-pika-python