Skip to main content

Python Client for Google Cloud Pub/Sub

Project description

This is a shared codebase for gcloud-rest-pubsub and gcloud-rest-pubsub

Latest PyPI Version Python Version Support (gcloud-rest-pubsub) Python Version Support (gcloud-rest-pubsub)


$ pip install --upgrade gcloud-{aio,rest}-pubsub



gcloud-{aio,rest}-pubsub provides SubscriberClient as an interface to call pubsub’s HTTP API:

from import SubscriberClient
from import SubscriberMessage

client = SubscriberClient()
# create subscription
await client.create_subscription(

# pull messages
messages: List[SubscriberMessage] = await client.pull(

There’s also helper function you can use to setup a pubsub processing pipeline. It is built with asyncio and thus only available in gcloud-rest-pubsub package. The usage is fairly simple:

from import SubscriberClient
from import subscribe
from import MetricsAgent

subscriber_client = SubscriberClient()

async def handler(message):

await subscribe(

While defaults are somewhat sensible, it is highly recommended to performance test your application and tweak function parameter to your specific needs. Here’s a few hints:


an async function that will be called for each message. It should accept an instance of SubscriberMessage as its only argument and return None if the message should be acked. An exception raised within the handler will result in the message being left to expire, and thus it will be redelivered according to your subscription’s ack deadline.


number of workers that will be making pull requests to pubsub. Please note that a worker will only fetch new batch once the handler was called for each message from the previous batch. This means that running only a single worker will most likely make your application IO bound. If you notice this being an issue don’t hesitate to bump this parameter.


number of pubsub messages a worker will try to fetch in a single batch. This value is passed to pull endpoint as maxMessages parameter. A rule of thumb here is the faster your handler is the bigger this value should be.


ack requests are handled separately and are done in batches. This parameters specifies how often ack requests will be made. Setting it to 0.0 will effectively disable batching.


how many handle calls a worker can make until it blocks to wait for them to return. If you process messages independently from each other you should be good with the default value of 1. If you do something fancy (e.g. aggregate messages before processing them), you’ll want a higher pool here. You can think of num_producers * num_tasks_per_consumer as an upper limit of how many messages can possibly be within your application state at any given moment.


if enabled messages for which callback raised an exception will be explicitly nacked using modifyAckDeadline endpoint so they can be retried immediately.


same as ack_window but for nack requests

subscribe has also an optional metrics_client argument. You can provide any metrics agent that implements the same interface as MetricsAgent (Datadog client will do ;) ) and get the following metrics:

  • pubsub.producer.batch - [histogram] actual size of a batch retrieved from pubsub.

  • pubsub.consumer.failfast - [increment] a message was dropped due to its lease being expired.

  • pubsub.consumer.latency.receive - [histogram] how many seconds it took for a message to reach handler after it was published.

  • pubsub.consumer.succeeded - [increment] handler call was successfull.

  • pubsub.consumer.failed - [increment] handler call raised an exception.

  • pubsub.consumer.latency.runtime - [histogram] handler execution time in seconds.

  • pubsub.acker.batch.failed - [increment] ack request failed.

  • pubsub.acker.batch - [histogram] actual number of messages that was acked in a single request.


The PublisherClient is a dead-simple alternative to the official Google Cloud Pub/Sub publisher client. The main design goal was to eliminate all the additional gRPC overhead implemented by the upstream client.

If migrating between this library and the official one, the main difference is this: the gcloud-{aio,rest}-pubsub publisher’s .publish() method immediately publishes the messages you’ve provided, rather than maintaining our own publishing queue, implementing batching and flow control, etc. If you’re looking for a full-featured publishing library with all the bells and whistles built in, you may be interested in the upstream provider. If you’re looking to manage your own batching / timeouts / retry / threads / etc, this library should be a bit easier to work with.

Sample usage:

from import PubsubMessage
from import PublisherClient

async with aiohttp.ClientSession() as session:
    client = PublisherClient(session=session)

    topic = client.topic_path('my-gcp-project', 'my-topic-name')

    messages = [
        PubsubMessage(b'payload', attribute='value'),
        PubsubMessage(b'other payload', other_attribute='whatever',
                      more_attributes='something else'),
    response = await client.publish(topic, messages)
    # response == {'messageIds': ['1', '2']}


For testing purposes, you may want to use gcloud-rest-pubsub along with a local GCS emulator. Setting the $PUBSUB_EMULATOR_HOST environment variable to the local address of your emulator should be enough to do the trick.

For example, using the official Google Pubsub emulator:

gcloud beta emulators pubsub start --host-port=

Any gcloud-rest-pubsub Publisher requests made with that environment variable set will query the emulator instead of the official GCS APIs.

For easier ergonomics, you may be interested in messagebird/gcloud-pubsub-emulator.


This library mostly tries to stay agnostic of potential use-cases; as such, we do not implement any sort of retrying or other policies under the assumption that we wouldn’t get things right for every user’s situation.

As such, we recommend configuring your own policies on an as-needed basis. The backoff library can make this quite straightforward! For example, you may find it useful to configure something like:

class SubscriberClientWithBackoff(SubscriberClient):
    @backoff.on_exception(backoff.expo, aiohttp.ClientResponseError,
                          max_tries=5, jitter=backoff.full_jitter)
    async def pull(self, *args: Any, **kwargs: Any):
        return await super().pull(*args, **kwargs)


Please see our contributing guide.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

gcloud-rest-pubsub-5.0.1.tar.gz (15.4 kB view hashes)

Uploaded Source

Built Distribution

gcloud_rest_pubsub-5.0.1-py2.py3-none-any.whl (15.7 kB view hashes)

Uploaded Python 2 Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page