This document describes the current stable version of Celery (4.1). For development docs, go here.

Task implementation: request context and the task base class.


Task base class.


When called tasks apply the run() method. This method must be defined by all tasks (that is unless the __call__() method is overridden).

AsyncResult(task_id, **kwargs)[source]

Get AsyncResult instance for this kind of task.

Parameters:task_id (str) – Task id to get result for.
exception MaxRetriesExceededError

The tasks max restart limit has been exceeded.

exception Task.OperationalError

Recoverable message transport connection error.

Task.Strategy = u'celery.worker.strategy:default'

Execution strategy used, or the qualified name of one.

Task.abstract = True

Deprecated attribute abstract here for compatibility.

Task.acks_late = False

When enabled messages for this task will be acknowledged after the task has been executed, and not just before (the default behavior).

Please note that this means the task may be executed twice if the worker crashes mid execution.

The application default can be overridden with the task_acks_late setting.

Task.add_to_chord(sig, lazy=False)[source]

Add signature to the chord the current task is a member of.

New in version 4.0.

Currently only supported by the Redis result backend.

  • sig (~@Signature) – Signature to extend chord with.
  • lazy (bool) – If enabled the new task won’t actually be called, and sig.delay() must be called manually.
Task.after_return(status, retval, task_id, args, kwargs, einfo)[source]

Handler called after the task returns.

  • status (str) – Current task state.
  • retval (Any) – Task return value/exception.
  • task_id (str) – Unique id of the task.
  • args (Tuple) – Original arguments for the task.
  • kwargs (Dict) – Original keyword arguments for the task.
  • einfo (ExceptionInfo) – Exception information.

The return value of this handler is ignored.

Return type:


Task.apply(args=None, kwargs=None, link=None, link_error=None, task_id=None, retries=None, throw=None, logfile=None, loglevel=None, headers=None, **options)[source]

Execute this task locally, by blocking until the task returns.

  • args (Tuple) – positional arguments passed on to the task.
  • kwargs (Dict) – keyword arguments passed on to the task.
  • throw (bool) – Re-raise task exceptions. Defaults to the task_eager_propagates setting.

pre-evaluated result.

Return type:


Task.apply_async(args=None, kwargs=None, task_id=None, producer=None, link=None, link_error=None, shadow=None, **options)[source]

Apply tasks asynchronously by sending a message.

  • args (Tuple) – The positional arguments to pass on to the task.
  • kwargs (Dict) – The keyword arguments to pass on to the task.
  • countdown (float) – Number of seconds into the future that the task should execute. Defaults to immediate execution.
  • eta (datetime) – Absolute time and date of when the task should be executed. May not be specified if countdown is also supplied.
  • expires (float, datetime) – Datetime or seconds in the future for the task should expire. The task won’t be executed after the expiration time.
  • shadow (str) – Override task name used in logs/monitoring. Default is retrieved from shadow_name().
  • connection (kombu.Connection) – Re-use existing broker connection instead of acquiring one from the connection pool.
  • retry (bool) – If enabled sending of the task message will be retried in the event of connection loss or failure. Default is taken from the task_publish_retry setting. Note that you need to handle the producer/connection manually for this to work.
  • retry_policy (Mapping) – Override the retry policy used. See the task_publish_retry_policy setting.
  • queue (str, kombu.Queue) – The queue to route the task to. This must be a key present in task_queues, or task_create_missing_queues must be enabled. See Routing Tasks for more information.
  • exchange (str, kombu.Exchange) – Named custom exchange to send the task to. Usually not used in combination with the queue argument.
  • routing_key (str) – Custom routing key used to route the task to a worker server. If in combination with a queue argument only used to specify custom routing keys to topic exchanges.
  • priority (int) – The task priority, a number between 0 and 9. Defaults to the priority attribute.
  • serializer (str) – Serialization method to use. Can be pickle, json, yaml, msgpack or any custom serialization method that’s been registered with kombu.serialization.registry. Defaults to the serializer attribute.
  • compression (str) – Optional compression method to use. Can be one of zlib, bzip2, or any custom compression methods registered with kombu.compression.register(). Defaults to the task_compression setting.
  • link (~@Signature) – A single, or a list of tasks signatures to apply if the task returns successfully.
  • link_error (~@Signature) – A single, or a list of task signatures to apply if an error occurs while executing the task.
  • producer (kombu.Producer) – custom producer to use when publishing the task.
  • add_to_parent (bool) – If set to True (default) and the task is applied while executing another task, then the result will be appended to the parent tasks request.children attribute. Trailing can also be disabled by default using the trail attribute
  • publisher (kombu.Producer) – Deprecated alias to producer.
  • headers (Dict) – Message headers to be included in the message.

Promise of future evaluation.

Return type:


  • TypeError – If not enough arguments are passed, or too many arguments are passed. Note that signature checks may be disabled by specifying @task(typing=False).
  • kombu.exceptions.OperationalError – If a connection to the transport cannot be made, or if the connection is lost.


Also supports all keyword arguments supported by kombu.Producer.publish().

Task.autoregister = True

If disabled this task won’t be registered automatically.


The result store backend used for this task.

Task.chunks(it, n)[source]

Create a chunks task for this task.

Task.default_retry_delay = 180

Default time in seconds before a retry of the task should be executed. 3 minutes by default.

Task.delay(*args, **kwargs)[source]

Star argument version of apply_async().

Does not support the extra options enabled by apply_async().

  • *args (Any) – Positional arguments passed on to the task.
  • **kwargs (Any) – Keyword arguments passed on to the task.

Future promise.

Return type:


Task.expires = None

Default task expiry time.

Task.ignore_result = False

If enabled the worker won’t store task state and return values for this task. Defaults to the task_ignore_result setting.[source]

Create a xmap task from it.

Task.max_retries = 3

Maximum number of retries before giving up. If set to None, it will never stop retrying. = None

Name of the task.

classmethod Task.on_bound(app)[source]

Called when the task is bound to an app.


This class method can be defined to do additional actions when the task class is bound to an app.

Task.on_failure(exc, task_id, args, kwargs, einfo)[source]

Error handler.

This is run by the worker when the task fails.

  • exc (Exception) – The exception raised by the task.
  • task_id (str) – Unique id of the failed task.
  • args (Tuple) – Original arguments for the task that failed.
  • kwargs (Dict) – Original keyword arguments for the task that failed.
  • einfo (ExceptionInfo) – Exception information.

The return value of this handler is ignored.

Return type:


Task.on_retry(exc, task_id, args, kwargs, einfo)[source]

Retry handler.

This is run by the worker when the task is to be retried.

  • exc (Exception) – The exception sent to retry().
  • task_id (str) – Unique id of the retried task.
  • args (Tuple) – Original arguments for the retried task.
  • kwargs (Dict) – Original keyword arguments for the retried task.
  • einfo (ExceptionInfo) – Exception information.

The return value of this handler is ignored.

Return type:


Task.on_success(retval, task_id, args, kwargs)[source]

Success handler.

Run by the worker if the task executes successfully.

  • retval (Any) – The return value of the task.
  • task_id (str) – Unique id of the executed task.
  • args (Tuple) – Original arguments for the executed task.
  • kwargs (Dict) – Original keyword arguments for the executed task.

The return value of this handler is ignored.

Return type:


Task.rate_limit = None

Rate limit for this task type. ExamplesNone (no rate limit), ‘100/s’ (hundred tasks a second), ‘100/m’ (hundred tasks a minute),`‘100/h’` (hundred tasks an hour)

Task.reject_on_worker_lost = None

Even if acks_late is enabled, the worker will acknowledge tasks when the worker process executing them abruptly exits or is signaled (e.g., KILL/INT, etc).

Setting this to true allows the message to be re-queued instead, so that the task will execute again by the same worker, or another worker.

Warning: Enabling this can cause message loops; make sure you know what you’re doing.


Replace this task, with a new task inheriting the task id.

New in version 4.0.


sig (~@Signature) – signature to replace with.

  • Ignore – This is always raised, so the best practice
  • is to always use raise self.replace(...) to convey
  • to the reader that the task won’t continue after being replaced.

Get current request object.

Task.request_stack = <celery.utils.threads._LocalStack object>

Task request stack, the current request will be the topmost.

Task.resultrepr_maxsize = 1024

Max length of result representation used in logs and events.

Task.retry(args=None, kwargs=None, exc=None, throw=True, eta=None, countdown=None, max_retries=None, **options)[source]

Retry the task.


>>> from imaginary_twitter_lib import Twitter
>>> from proj.celery import app
>>> @app.task(bind=True)
... def tweet(self, auth, message):
...     twitter = Twitter(oauth=auth)
...     try:
...         twitter.post_status_update(message)
...     except twitter.FailWhale as exc:
...         # Retry in 5 minutes.
...         raise self.retry(countdown=60 * 5, exc=exc)


Although the task will never return above as retry raises an exception to notify the worker, we use raise in front of the retry to convey that the rest of the block won’t be executed.

  • args (Tuple) – Positional arguments to retry with.
  • kwargs (Dict) – Keyword arguments to retry with.
  • exc (Exception) –

    Custom exception to report when the max retry limit has been exceeded (default: MaxRetriesExceededError).

    If this argument is set and retry is called while an exception was raised (sys.exc_info() is set) it will attempt to re-raise the current exception.

    If no exception was raised it will raise the exc argument provided.

  • countdown (float) – Time in seconds to delay the retry for.
  • eta (dateime) – Explicit time and date to run the retry at.
  • max_retries (int) – If set, overrides the default retry limit for this execution. Changes to this parameter don’t propagate to subsequent task retry attempts. A value of None, means “use the default”, so if you want infinite retries you’d have to set the max_retries attribute of the task to None first.
  • time_limit (int) – If set, overrides the default time limit.
  • soft_time_limit (int) – If set, overrides the default soft time limit.
  • throw (bool) – If this is False, don’t raise the Retry exception, that tells the worker to mark the task as being retried. Note that this means the task will be marked as failed if the task raises an exception, or successful if it returns after the retry call.
  • **options (Any) – Extra options to pass on to apply_async().

celery.exceptions.Retry – To tell the worker that the task has been re-sent for retry. This always happens, unless the throw keyword argument has been explicitly set to False, and is considered normal operation.*args, **kwargs)[source]

The body of the task executed by workers.

Task.s(*args, **kwargs)[source]

Create signature.

Shortcut for .s(*a, **k) -> .signature(a, k).

Task.send_event(type_, retry=True, retry_policy=None, **fields)[source]

Send monitoring event message.

This can be used to add custom event types in Flower and other monitors.


type (str) – Type of event, e.g. "task-failed".

Keyword Arguments:
  • retry (bool) – Retry sending the message if the connection is lost. Default is taken from the task_publish_retry setting.
  • retry_policy (Mapping) – Retry settings. Default is taken from the task_publish_retry_policy setting.
  • **fields (Any) – Map containing information about the event. Must be JSON serializable.
Task.send_events = True

If enabled the worker will send monitoring events related to this task (but only if the worker is configured to send task related events). Note that this has no effect on the task-failure event case where a task is not registered (as it will have no task class to check this flag).

Task.serializer = u'json'

The name of a serializer that are registered with kombu.serialization.registry. Default is ‘pickle’.

Task.shadow_name(args, kwargs, options)[source]

Override for custom task name in worker logs/monitoring.


from celery.utils.imports import qualname

def shadow_name(task, args, kwargs, options):
    return qualname(args[0])

@app.task(shadow_name=shadow_name, serializer='pickle')
def apply_function_async(fun, *args, **kwargs):
    return fun(*args, **kwargs)
  • args (Tuple) – Task positional arguments.
  • kwargs (Dict) – Task keyword arguments.
  • options (Dict) – Task execution options.*args, **kwargs)[source]

Create immutable signature.

Shortcut for .si(*a, **k) -> .signature(a, k, immutable=True).

Task.signature(args=None, *starargs, **starkwargs)[source]

Create signature.

object for
this task, wrapping arguments and execution options for a single task invocation.
Return type:signature
Task.soft_time_limit = None

Soft time limit. Defaults to the task_soft_time_limit setting.


Create a xstarmap task from it.

Task.store_errors_even_if_ignored = False

When enabled errors will be stored even if the task is otherwise configured to ignore results.

Task.subtask(args=None, *starargs, **starkwargs)

Create signature.

object for
this task, wrapping arguments and execution options for a single task invocation.
Return type:signature
Task.throws = ()

Tuple of expected exceptions.

These are errors that are expected in normal operation and that shouldn’t be regarded as a real error by the worker. Currently this means that the state will be updated to an error state, but the worker won’t log the event as an error.

Task.time_limit = None

Hard time limit. Defaults to the task_time_limit setting.

Task.track_started = False

If enabled the task will report its status as ‘started’ when the task is executed by a worker. Disabled by default as the normal behavior is to not report that level of granularity. Tasks are either pending, finished, or waiting to be retried.

Having a ‘started’ status can be useful for when there are long running tasks and there’s a need to report what task is currently running.

The application default can be overridden using the task_track_started setting.

Task.trail = True

If enabled the request will keep track of subtasks started by this task, and this information will be sent with the result (result.children).

Task.typing = True

Enable argument checking. You can set this to false if you don’t want the signature to be checked when calling the task. Defaults to Celery.strict_typing.

Task.update_state(task_id=None, state=None, meta=None)[source]

Update task state.

  • task_id (str) – Id of the task to update. Defaults to the id of the current task.
  • state (str) – New state.
  • meta (Dict) – State meta-data.
class*args, **kwargs)[source]

Task request variables (Task.request).

Here for backwards compatibility as tasks no longer use a custom meta-class.

alias of type