Django huey extension which supports multiple huey queues.


Keywords
django, django-extensions, huey
License
MIT
Install
pip install hueyx==1.0.3

Documentation

hueyx

PyPI version

A django extension to run huey with multiple queues. Multiple queues allow tasks to not block each other and to scale independently. Only the redis storage is supported.


Important
  • If you use huey 1.x then install hueyx 0.1.2. Checkout the git tag huey1.x.
  • If you use huey 2.x then install hueyx >= 1.0.

Usage

Install it with

pip install hueyx

Add hueyx in your installed apps.

INSTALLED_APPS = [
    'django.contrib.admin',
    'django.contrib.auth',
    'django.contrib.contenttypes',
    'django.contrib.sessions',
    'django.contrib.messages',
    'django.contrib.staticfiles',
    'hueyx',
]
settings.py

Compared to djhuey, hueyx allows several queues to be defined in the settings.py.

HUEYX = {
    'queue_name1': {
        'connection': {
            'host': 'localhost',
            'port': 6379,
            'db': 0,
        },
        'consumer': {
            'workers': 1,
            'worker_type': 'process',
        }
    },
    'queue_name2': {
        'connection': {
            'connection_pool': ConnectionPool(host='localhost', port=6379, db=1)
        },
        'consumer': {
            'multiple_scheduler_locking': True,
            'prometheus_metrics': True,
            'workers': 2,
            'worker_type': 'thread',
        }
    },
    'priority_queue_name3': {
        'huey_class': 'huey.PriorityRedisHuey',
        'connection': {
            'connection_pool': ConnectionPool(host='localhost', port=6379, db=1)
        },
        'consumer': {
            'multiple_scheduler_locking': True,
            'prometheus_metrics': True,
            'workers': 2,
            'worker_type': 'thread',
        }
    },
}

The settings are almost the same as in djhuey. Have a look at the huey documentation to see the exact parameter usage.

Exceptions:

  • You can only configure redis as storage engine by configure huey_class to huey.RedisHuey, huey.PriorityRedisHuey, huey.RedisExpireHuey or huey.PriorityRedisExpireHuey.
  • The name and backend_class parameters are not supported.
  • The options multiple_scheduler_locking and prometheus_metrics_enabled have been added. See below.
  • The parameters heartbeat_timeout for db_task has been added. See below.
tasks.py
from hueyx.queues import hueyx

"""
Define which queue you want to use.
They are predefined in settings.py.
"""
HUEY_Q1 = hueyx('queue_name1')
HUEY_Q2 = hueyx('queue_name2')


@HUEY_Q1.task()
def my_task1():
    print('my_task1 called')
    
@HUEY_Q1.db_task()
def my_db_task1():
    print('my_db_task1 called')
    
@HUEY_Q2.task()
def my_task2():
    print('my_task2 called')

@HUEY_Q2.periodic_task(crontab(minute='0', hour='3'))
def my_periodic_task2():
    print('my_periodic_task2 called')
    return 1
    
@HUEY_Q2.db_task(heartbeat_timeout=120)
def my_heartbeat_task(heartbeat: Heartbeat):
    with heartbeat.long_running_operation():
        print('This operation can take a while -> don\'t check for heartbeats')
    print('Now we check for heartbeats -> call heartbeat() periodically')
    heartbeat()
Push task to queue
from example.tasks import my_task1, my_db_task1, my_task2


my_task1()  # Task for queue_name1
my_db_task1()  # Task for queue_name1
my_task2()  # Task for queue_name2
Run consumer

Consumers are started with the queue_name.

./manage.py run_hueyx queue_name1
Heartbeat tasks

Heartbeat tasks are tasks with the parameter heartbeat_timeout. It defines the timeout in seconds. They get a Heartbeat object which needs to be called in order to send a heartbeat to redis. If no heartbeat occurs in set timeout the task is presumed to be dead and will automatically get restarted. heartbeat_timeout needs to be at least 120 seconds. It does not work together with the parameter include_task.

Additional settings

multiple_scheduler_locking

multiple_scheduler_locking has been added to support multiple huey schedulers. If you run huey in a cloud environment, you will end up running multiple huey instances which each will schedule the periodic task. multiple_scheduler_locking prevents periodic tasks to be scheduled multiple times. It is false by default.

Huey signals

Optionally hueyx pushes all huey signals to the redis pubsub hueyx.huey2.signaling if enabled.

HUEYX_SIGNALS = {
    'enabled': True,
    'environment': 'your environment'
}

The format of the message is

{
    'environment': settings.HUEYX_SIGNALS['environment'],
    'queue': queue,
    'pid': pid,
    'signal': signal_name,
    'task': task_name
}

The environment parameter is a optional variable.

Prometheus

The huey-exporter project takes the signals und reports it to prometheus.

Collaborators