Celery consumer (only) with an external producer - redis

I'm using Celery 4.4.7 with Redis as my message broker.
I want to use celery as a consumer only, as the external producer is a java application.
The java application pushes messages to a channel on redis. But my celery application is not picking up the messages.
I have simulated the java producer in python using redis-py (redis_producer.py) to publish to a channel. The redis_consumer.py is able to pickup the messages from the producer.
But my celery_consumer.py seems to be blind to these messages.
Messages from redis_producer.py is picked up by the redis_consumer.py, but not from celery.
Messages from kombu_producer.py is picked up by the celery worker, but not from my redis_consumer.py
redis_producer.py
import json
import redis
r = redis.Redis(host='localhost', port=6379)
for i in range(10):
body = {
'id': i,
'message': f'Hello {i}',
}
r.publish(channel='redis.test.topic', message=json.dumps(body))
redis_consumer.py
import json
import multiprocessing
import os
import signal
import redis
redis_conn = redis.Redis(charset='utf-8', decode_responses=True)
def sub(name: str):
pubsub = redis_conn.pubsub()
pubsub.subscribe('redis.test.topic')
for message in pubsub.listen():
print(message)
if message.get('type') == 'message':
data = message.get('data')
print('%s: %s' % (name, data))
def on_terminate(signum, stack):
wait_for_current_scp_operation()
if __name__ == '__main__':
multiprocessing.Process(target=sub, args=('consumer',)).start()
signal.signal(signal.SIGTERM, on_terminate)
celery_consumer.py
import os
import ssl
from celery import Celery
from celery import shared_task
from celery.utils.log import get_task_logger
from kombu import Exchange
from kombu import Queue
logger = get_task_logger(__name__)
# Celery broker Url
broker_url = os.environ.get('CELERY_BROKER_URL', None)
if broker_url is None:
broker_url = 'redis://localhost:6379/0'
# store to use to store task results, default=None
result_backend = os.environ.get('CELERY_RESULT_BACKEND', None)
if result_backend is None:
result_backend = 'redis://localhost:6379/0'
config = dict(
broker_url=broker_url,
result_backend=result_backend,
# maximum number of connections that can be open in the connection pool
broker_pool_limit=20,
broker_transport_options={
'visibility_timeout': 3600,
'confirm_publish': True,
},
# Serializer method
task_serializer='json',
result_serializer='json',
accept_content=['json'],
# Below two settings just reserves one task at a time
# Task acknowledgement mode, default=False
task_acks_late=True,
# How many message to prefetch, default=4, value=1 disable prefetch
worker_prefetch_multiplier=1,
# Dates and times in messages will be converted to use the UTC timezone
timezone='UTC',
enable_utc=True,
# if true, store the task return values
task_ignore_result=False,
# if true, result messages will be persistent, messages won't be lost after a broker restart
result_persistent=False,
# Celery tasks expiry (in secs), default=1d, value=0/None never expire
result_expires=900,
# Message compression setting
task_compression='gzip',
# Task execution marker, default=False
task_track_started=True,
# rate limits on tasks (Disable all rate limits, even if tasks has explicit rate limits set)
worker_disable_rate_limits=True,
# is True, all tasks will be executed locally by blocking until the task returns
task_always_eager=False,
# Send events so the worker can be monitored by tools like celerymon
worker_send_task_events=False,
# Expiry time in seconds for when a monitor clients event queue will be deleted, default=never
event_queue_expires=60,
# Default queue, exchange, routing keys configuration
# task_default_queue = 'default.queue',
# task_default_exchange = 'default.exchange',
# task_default_exchange_type='topic',
# task_default_routing_key = 'default.route',
# task_create_missing_queues = True
task_queues=(
# Default configuration
Queue('redis.test.topic',
Exchange('redis.test.topic'),
routing_key='redis.test.topic'),
),
)
def create_celery_app() -> Celery:
logger.info('Initializing Celery...')
celery_app = Celery(name=__name__)
celery_app.config_from_object(config)
return celery_app
# create a celery app
app = create_celery_app()
#app.task(name='task_process_message', bind=True, max_retries=3)
def task_process_message(self, message):
try:
logger.info(f'{message}: Triggered task: task_process_message')
except Exception as e:
logger.exception(
f'Error executing task_process_message({message}')
# We do not have to reset timestamp since the job always looks back by 1 hr
self.retry(exc=e, countdown=utils.get_retry_delay(self.request.retries))
#shared_task(name='shared_task_process_message', bind=True, max_retries=3)
def shared_task_process_message(self, message):
try:
logger.info(f'{message}: Triggered task: shared_task_process_message')
except Exception as e:
logger.exception(
f'Error executing shared_task_process_message({message}')
# We do not have to reset timestamp since the job always looks back by 1 hr
self.retry(exc=e, countdown=utils.get_retry_delay(self.request.retries))
kombu_producer.py
from kombu import Producer, Consumer, Queue, Connection
import json
redis_url = 'redis://localhost:6379/0'
conn = Connection(redis_url)
producer = Producer(conn.channel())
channel = 'redis.test.topic'
for i in range(10):
body = {
'task': 'task_process_message',
'id': f'{i}',
'kwargs': {'message': f'Hello {i}',
}
}
producer.publish(body=body, routing_key='redis.test.topic')
The below picture shows activity on redis using a regular redis producer/consumer
Below picture shows activity on redis while running kombu producer and celery consumer.

Related

Disable global qos in rabbitmq from celery

I have created a RabbitMQ 3-node cluster using three aws ec2 servers. I'm trying to access the quorum queue I created using celery. When I connect it gives the error
raise error_for_code(reply_code, reply_text,
amqp.exceptions.AMQPNotImplementedError: Basic.consume: (540) NOT_IMPLEMENTED - queue 'Replica_que' in vhost '/' does not support global qos
I suppose it will work if I disabled global qos but I couldn't find where I can do it. How do I disable global qos in celery?
my celery code
from celery import Celery
from time import sleep
import kombu
broker_uri=['amqp://xxxx:5672/', 'amqp://xxxx:5672/','amqp://xxx:5672/']
backend_uri="mongodb+srv://xxxxx"
app = Celery('TestApp', broker=broker_uri,backend=backend_uri)
app.config_from_object('celeryconfig')
app.conf.task_default_exchange='Replica_que'
app.conf.task_default_routing_key='Replica'
#app.task
def reverse(text):
sleep(10)
return text[:-1]
and the config code
from kombu import Queue
task_queues = [Queue(name="Replica_que", queue_arguments={"x-queue-type": "quorum"})]
task_routes = {
'tasks.add': 'Replica_que',
}
This was possible by adding a celeryconfig.py file,
from kombu import Queue
task_queues = [Queue(name="Replica_que", queue_arguments={"x-queue-type": "quorum"})]
task_routes = {
'tasks.add': 'Replica_que',
}
and creating a custom QoS class: https://github.com/celery/celery/issues/6067
So I added the QoS class
class NoChannelGlobalQoS(bootsteps.StartStopStep):
requires = {'celery.worker.consumer.tasks:Tasks'}
def start(self, c):
qos_global = False
c.connection.default_channel.basic_qos(0, c.initial_prefetch_count, qos_global)
def set_prefetch_count(prefetch_count):
return c.task_consumer.qos(
prefetch_count=prefetch_count,
apply_global=qos_global,
)
c.qos = QoS(set_prefetch_count, c.initial_prefetch_count)
app.steps['consumer'].add(NoChannelGlobalQoS)
Currently this is an issue in celery related to quorum queue but this works.

Client not receiving events from Flask-SocketIO server with Redis message queue

I want to add multiprocessing to my Flask-SocketIO server so I am trying to add a Redis message queue as per the Flask-SocketIO docs. Even without adding multiprocessing, the client is not receiving any events. Everything else is working fine (e.g. the web page is being served, HTTP requests are being made, database calls are being made). There are no error messages on the front or back end. Before I added the Redis queue it was working. I verified that the 'addname' SocketIO route is being hit and that the request.sid looks right. What am I doing wrong?
Very simplified server code:
external_sio = SocketIO(message_queue='redis://')
def requester(user, sid):
global external_sio
external_sio.emit('addname', {'data': 'hello'}, room=sid)
# do some stuff with requests and databases
external_sio.emit('addname', {'data': 'goodbye'}, room=sid)
def main():
app = Flask(__name__,
static_url_path='',
static_folder='dist',
template_folder='dist')
socketio = SocketIO(app)
#socketio.on('addname')
def add_name(user):
global external_sio
external_sio.emit('addname', {'data': 'test'}, room=request.sid)
requester(user.data, request.sid)
socketio.run(app, host='0.0.0.0', port=8000)
if __name__ == '__main__':
main()
Simplified client code (React Javascript):
const socket = SocketIOClient('ipaddress:8000')
socket.emit('addname', {data: 'somename'})
socket.on('addname', ({data}) => console.log(data))
The main server also needs to be connected to the message queue. In your main server do this:
socketio = SocketIO(app, message_queue='redis://')
In your external process do this:
external_sio = SocketIO(message_queue='redis://') # <--- no app on this one

Google Cloud IoT - Multiple config messages

In my code I'm subscribing to 3 different topics:
/devices/{}/config
/devices/{}/events
/devices/{}/state
If I'm not subscribed in /devices/{}/config I don't receive any config message, and that's ok. But if I'm subscribed in /devices/{}/config I get one message for each subscribe I have.
Example:
Subscribed in "/devices/{}/config" and "/devices/{}/events" I get 2 config messages.
Subscribed in "/devices/{}/config" and "/devices/{}/state" I get 2 config messages.
Subscribed in "/devices/{}/config", "/devices/{}/state" and "/devices/{}/events" I get 3 config messages.
Subscribed in "/devices/{}/events" and "/devices/{}/state" I get 0 config messages.
This is causing error in IoT core: mqtt:
The device "xxxxxxxx" could not be updated. Device state can be updated only once every 1s.
In fact i want and need just one config message. What am I doing wrong?
This is my code:
# [START iot_mqtt_includes]
import argparse
import datetime
import os
import random
import ssl
import time
import log
import updateConfig
import jwt
import paho.mqtt.client as mqtt
import payload
# [END iot_mqtt_includes]
# The initial backoff time after a disconnection occurs, in seconds.
minimum_backoff_time = 1
# The maximum backoff time before giving up, in seconds.
MAXIMUM_BACKOFF_TIME = 32
# Whether to wait with exponential backoff before publishing.
should_backoff = False
# [START iot_mqtt_jwt]
def create_jwt(project_id, private_key_file, algorithm):
token = {
# The time that the token was issued at
'iat': datetime.datetime.utcnow(),
# The time the token expires.
'exp': datetime.datetime.utcnow() + datetime.timedelta(minutes=60),
# The audience field should always be set to the GCP project id.
'aud': project_id
}
# Read the private key file.
with open(private_key_file, 'r') as f:
private_key = f.read()
f.close()
return jwt.encode(token, private_key, algorithm=algorithm)
# [END iot_mqtt_jwt]
# [START iot_mqtt_config]
def error_str(rc):
"""Convert a Paho error to a human readable string."""
return '{}: {}'.format(rc, mqtt.error_string(rc))
def on_connect(unused_client, unused_userdata, unused_flags, rc):
"""Callback for when a device connects."""
log.append_log('ao_conectar - ' + mqtt.connack_string(rc))
# After a successful connect, reset backoff time and stop backing off.
global should_backoff
global minimum_backoff_time
should_backoff = False
minimum_backoff_time = 1
def on_disconnect(unused_client, unused_userdata, rc):
"""Paho callback for when a device disconnects."""
append_log('ao_desconectar - '+ error_str(rc))
# Since a disconnect occurred, the next loop iteration will wait with
# exponential backoff.
global should_backoff
should_backoff = True
def on_publish(unused_client, unused_userdata, unused_mid):
"""Paho callback when a message is sent to the broker."""
def on_message(unused_client, unused_userdata, message):
"""Callback when the device receives a message on a subscription."""
config = str(message.payload)
retorno = updateConfig.update(config)
def get_client(
project_id, cloud_region, registry_id, device_id, private_key_file,
algorithm, ca_certs, mqtt_bridge_hostname, mqtt_bridge_port):
"""Create our MQTT client. The client_id is a unique string that identifies
this device. For Google Cloud IoT Core, it must be in the format below."""
client = mqtt.Client(
client_id=('projects/{}/locations/{}/registries/{}/devices/{}'
.format(
project_id,
cloud_region,
registry_id,
device_id)))
# With Google Cloud IoT Core, the username field is ignored, and the
# password field is used to transmit a JWT to authorize the device.
client.username_pw_set(
username='unused',
password=create_jwt(
project_id, private_key_file, algorithm))
# Enable SSL/TLS support.
client.tls_set(ca_certs=ca_certs, tls_version=ssl.PROTOCOL_TLSv1_2)
# Register message callbacks. https://eclipse.org/paho/clients/python/docs/
# describes additional callbacks that Paho supports. In this example, the
# callbacks just print to standard out.
client.on_connect = on_connect
client.on_publish = on_publish
client.on_disconnect = on_disconnect
client.on_message = on_message
# Connect to the Google MQTT bridge.
client.connect(mqtt_bridge_hostname, mqtt_bridge_port)
# This is the topic that the device will receive configuration updates on.
mqtt_config_topic = '/devices/{}/config'.format(device_id)
# Subscribe to the config topic.
client.subscribe(mqtt_config_topic, qos=1)
return client
# [END iot_mqtt_config]
def parse_command_line_args():
"""Parse command line arguments."""
parser = argparse.ArgumentParser(description=(
'Example Google Cloud IoT Core MQTT device connection code.'))
parser.add_argument(
'--project_id',
default=os.environ.get('GOOGLE_CLOUD_PROJECT'),
help='GCP cloud project name')
parser.add_argument(
'--registry_id', required=True, help='Cloud IoT Core registry id')
parser.add_argument(
'--device_id', required=True, help='Cloud IoT Core device id')
parser.add_argument(
'--private_key_file',
required=True, help='Path to private key file.')
parser.add_argument(
'--algorithm',
choices=('RS256', 'ES256'),
required=True,
help='Which encryption algorithm to use to generate the JWT.')
parser.add_argument(
'--cloud_region', default='us-central1', help='GCP cloud region')
parser.add_argument(
'--ca_certs',
default='roots.pem',
help=('CA root from https://pki.google.com/roots.pem'))
parser.add_argument(
'--message_type',
choices=('event', 'state'),
default='event',
help=('Indicates whether the message to be published is a '
'telemetry event or a device state message.'))
parser.add_argument(
'--mqtt_bridge_hostname',
default='mqtt.googleapis.com',
help='MQTT bridge hostname.')
parser.add_argument(
'--mqtt_bridge_port',
choices=(8883, 443),
default=8883,
type=int,
help='MQTT bridge port.')
parser.add_argument(
'--jwt_expires_minutes',
default=20,
type=int,
help=('Expiration time, in minutes, for JWT tokens.'))
return parser.parse_args()
# [START iot_mqtt_run]
def main():
log.append_log("Iniciando uma nova conexao com o Google IoT.")
global minimum_backoff_time
args = parse_command_line_args()
# Publish to the events or state topic based on the flag.
jwt_iat = datetime.datetime.utcnow()
jwt_exp_mins = args.jwt_expires_minutes
client = get_client(
args.project_id, args.cloud_region, args.registry_id, args.device_id,
args.private_key_file, args.algorithm, args.ca_certs,
args.mqtt_bridge_hostname, args.mqtt_bridge_port)
# Publish num_messages mesages to the MQTT bridge once per second.
while True:
# Process network events.
client.loop()
# Wait if backoff is required.
if should_backoff:
# If backoff time is too large, give up.
if minimum_backoff_time > MAXIMUM_BACKOFF_TIME:
log.append_log('Tempo maximo de backoff excedido. Desistindo.')
break
# Otherwise, wait and connect again.
delay = minimum_backoff_time + random.randint(0, 1000) / 1000.0
log.append_log('Esperando {} segundos antes de reconectar.'.format(delay))
time.sleep(delay)
minimum_backoff_time *= 2
client.connect(args.mqtt_bridge_hostname, args.mqtt_bridge_port)
# [START iot_mqtt_jwt_refresh]
seconds_since_issue = (datetime.datetime.utcnow() - jwt_iat).seconds
if seconds_since_issue > 60 * jwt_exp_mins:
log.append_log('Atualizando token de acesso depois de {} segundos'.format(seconds_since_issue))
client.loop_stop()
jwt_iat = datetime.datetime.utcnow()
client = get_client(
args.project_id, args.cloud_region,
args.registry_id, args.device_id, args.private_key_file,
args.algorithm, args.ca_certs, args.mqtt_bridge_hostname,
args.mqtt_bridge_port)
# [END iot_mqtt_jwt_refresh]
# Publish "payload" to the MQTT topic. qos=1 means at least once
# delivery. Cloud IoT Core also supports qos=0 for at most once
# delivery.
payloadToPublish = payload.lerPayload()
if payloadToPublish != 'sem payload':
if payloadToPublish[0] == 'event':
mqtt_topic = '/devices/{}/{}'.format(args.device_id, 'events')
log.append_log('publicando [' + payloadToPublish[1] + ']')
else:
mqtt_topic = '/devices/{}/{}'.format(args.device_id, 'state')
client.publish(mqtt_topic, payloadToPublish[1], qos=1)
# [END iot_mqtt_run]
if __name__ == '__main__':
main()
This is the expected behavior:
The event and state topic are used for device to cloud communication and the config topic is used for sending configuration data to an IoT device. Subscribing to the events / state topics is effectively a NOOP.
Cloud IoT Core only supports QoS 1, which is "at least once" for message transmission, you may want to try QoS 0, which does not retry message transmission until ACK but I don't think this is what you want
STATE and CONFIG transmission is limited to a single message per second; transmitting configuration changes faster than this will cause the error you're seeing

Extract Runtime and Time of Completion from Celery Task:

After running a celery task, I get the following message piped to stdout:
[2015-11-22 21:08:52,158: INFO/MainProcess] Task jobs.tasks.some_task[9c37f17b-dec4-4cb0-ab2b-fb5e997e430a]
succeeded in 2.9128301960008685s: 0.9731072908536255
I would like to programmatically extract the completion time (here: 2015-11-22 21:08:52) and the runtime (2.9128301960008685s) and display these to the user.
How can these values be extracted from the AsyncResult?
Broker is RabbitMQ.
Use celery task events to get the task runtime and time of completion. For more details about task events, refer this link.
from celery import Celery
import datetime
taskId_startTime = {}
taskId_addedTime = {}
def my_monitor():
app = Celery('vwadaptor', broker='redis://workerdb:6379/0',backend='redis://workerdb:6379/0')
state = app.events.State()
def announce_task_succeeded(event):
state.event(event)
task = state.tasks.get(event['uuid'])
print "runtime: ", task.runtime
print "time of completion: ", datetime.datetime.now()
with app.connection() as connection:
recv = app.events.Receiver(connection, handlers={
'task-succeeded': announce_task_succeeded,
})
recv.capture(limit=None, timeout=None, wakeup=True)
my_monitor()

Can Kombu do publish and sucscribe to multiple consumers

Using Kombu with RabbitMQ to implement a classic publish/subscribe design pattern. I have created a producer that creates a topic:
from kombu import Connection, Exchange, Queue
media_exchange = Exchange('media', 'topic', durable=False)
video_queue = Queue('video', exchange=media_exchange, routing_key='video')
with Connection('amqp://guest:guest#localhost//') as conn:
producer = conn.Producer(serializer='json')
producer.publish('Hello World!',
exchange=media_exchange, routing_key='video',
declare=[video_queue])
I then created a consumer to consume from the publisher:
from kombu import Connection, Exchange, Queue
media_exchange = Exchange('media', type='topic', durable=False)
video_queue = Queue('video', exchange=media_exchange, routing_key='video')
def process_media(body, message):
print(body)
#message.ack()
with Connection('amqp://guest:guest#localhost//') as conn:
with conn.Consumer(video_queue, callbacks=[process_media]) as consumer:
# Process messages and handle events on all channels
while True:
conn.drain_events()
In then launch two consumers, each one in a separate terminal; both wait for a message:
terminal 1: python consumer.py
terminal 2: python consumer.py
When I run the producer, only one consumer receives the message.
The producer publishes in an exchange, not in a queue. The queues are defined by the consumers. When using different queue name for each consumer then all will get the message.
When using many consumers for the same queue then it is load balancing, that's why only one of your consumers gets the message.
To clarify, the messages in the queue are 'consumed' i.e. the first consumer consumes it, and the message is no more in the queue, that's why the second consumer isn't getting anything.
To have 2 separate consumers for same message - use 2 separate queues i.e.
video_queue1 and video_queue2, declared and bound to the exchange media_exchange, using same key video.
producer.py
from kombu import Connection, Exchange, Queue
media_exchange = Exchange('media', 'topic', durable=False)
video_queue1 = Queue('video1', exchange=media_exchange, routing_key='video')
video_queue2 = Queue('video2', exchange=media_exchange, routing_key='video')
with Connection('amqp://guest:guest#localhost//') as conn:
producer = conn.Producer(serializer='json')
producer.publish('Hello World!',
exchange=media_exchange, routing_key='video',
declare=[video_queue1, video_queue2])
consumer1.py
from kombu import Connection, Exchange, Queue
media_exchange = Exchange('media', type='topic', durable=False)
video_queue = Queue('video1', exchange=media_exchange, routing_key='video')
def process_media(body, message):
print(body)
#message.ack()
with Connection('amqp://guest:guest#localhost//') as conn:
with conn.Consumer(video_queue, callbacks=[process_media]) as consumer:
# Process messages and handle events on all channels
while True:
conn.drain_events()
consumer2.py
from kombu import Connection, Exchange, Queue
media_exchange = Exchange('media', type='topic', durable=False)
video_queue = Queue('video2', exchange=media_exchange, routing_key='video')
def process_media(body, message):
print(body)
#message.ack()
with Connection('amqp://guest:guest#localhost//') as conn:
with conn.Consumer(video_queue, callbacks=[process_media]) as consumer:
# Process messages and handle events on all channels
while True:
conn.drain_events()