celery.events.state

In-memory representation of cluster state.

This module implements a data-structure used to keep track of the state of a cluster of workers and the tasks it is working on (by consuming events).

For every event consumed the state is updated, so the state represents the state of the cluster at the time of the last event.

Snapshots (celery.events.snapshot) can be used to take “pictures” of this state at regular intervals to for example, store that in a database.

classcelery.events.state.State(callback=None, workers=None, tasks=None, taskheap=None, max_workers_in_memory=5000, max_tasks_in_memory=10000, on_node_join=None, on_node_leave=None, tasks_by_type=None, tasks_by_worker=None)

Records clusters state.

classTask(uuid=None, cluster_state=None, children=None, **kwargs)

Task State.

args=None

as_dict()

client=None

clock=0

eta=None

event(type_, timestamp=None, local_received=None, fields=None, precedence=<function precedence>, setattr=<built-in function setattr>, task_event_to_state=<built-in method get of dict object>, RETRY='RETRY')

exception=None

exchange=None

expires=None

failed=None

propertyid

info(fields=None, extra=None)

Information about this task suitable for on-screen display.

kwargs=None

merge_rules={'RECEIVED': ('name', 'args', 'kwargs', 'parent_id', 'root_id', 'retries', 'eta', 'expires')}

How to merge out of order events. Disorder is detected by logical ordering (e.g., task-received must’ve happened before a task-failed event).

A merge rule consists of a state and a list of fields to keep from that state. (RECEIVED, ('name', 'args'), means the name and args fields are always taken from the RECEIVED state, and any values for these fields received before or after is simply ignored.

name=None

propertyorigin

propertyparent

parent_id=None

propertyready

received=None

rejected=None

result=None

retried=None

retries=None

revoked=None

propertyroot

root_id=None

routing_key=None

runtime=None

sent=None

started=None

state='PENDING'

succeeded=None

timestamp=None

traceback=None

worker=None

classWorker(hostname=None, pid=None, freq=60, heartbeats=None, clock=0, active=None, processed=None, loadavg=None, sw_ident=None, sw_ver=None, sw_sys=None)

Worker State.

active

propertyalive

clock

event

expire_window=200

freq

propertyheartbeat_expires

heartbeat_max=4

heartbeats

hostname

propertyid

loadavg

pid

processed

propertystatus_string

sw_ident

sw_sys

sw_ver

update(f, **kw)

alive_workers()

Return a list of (seemingly) alive workers.

clear(ready:bool=True)

clear_tasks(ready=True)

event(event)

event_count=0

freeze_while(fun, *args, **kwargs)

get_or_create_task(uuid)

Get or create task by uuid.

get_or_create_worker(hostname, **kwargs)

Get or create worker by hostname.

Returns:

of (worker, was_created) pairs.

Return type:

Tuple

heap_multiplier=4

itertasks(limit:int|None=None)

rebuild_taskheap(timetuple=<class 'kombu.clocks.timetuple'>)

task_count=0

task_event(type_, fields)

Deprecated, use event().

task_types()

Return a list of all seen task types.

tasks_by_time(limit=None, reverse:bool=True)

Generator yielding tasks ordered by time.

Yields:

Tuples of (uuid, Task).

tasks_by_timestamp(limit=None, reverse:bool=True)

Generator yielding tasks ordered by time.

Yields:

Tuples of (uuid, Task).

worker_event(type_, fields)

Deprecated, use event().

classcelery.events.state.Task(uuid=None, cluster_state=None, children=None, **kwargs)

Task State.

args=None

as_dict()

client=None

clock=0

eta=None

event(type_, timestamp=None, local_received=None, fields=None, precedence=<function precedence>, setattr=<built-in function setattr>, task_event_to_state=<built-in method get of dict object>, RETRY='RETRY')

exception=None

exchange=None

expires=None

failed=None

propertyid

info(fields=None, extra=None)

Information about this task suitable for on-screen display.

kwargs=None

merge_rules={'RECEIVED': ('name', 'args', 'kwargs', 'parent_id', 'root_id', 'retries', 'eta', 'expires')}

How to merge out of order events. Disorder is detected by logical ordering (e.g., task-received must’ve happened before a task-failed event).

A merge rule consists of a state and a list of fields to keep from that state. (RECEIVED, ('name', 'args'), means the name and args fields are always taken from the RECEIVED state, and any values for these fields received before or after is simply ignored.

name=None

propertyorigin

propertyparent

parent_id=None

propertyready

received=None

rejected=None

result=None

retried=None

retries=None

revoked=None

propertyroot

root_id=None

routing_key=None

runtime=None

sent=None

started=None

state='PENDING'

succeeded=None

timestamp=None

traceback=None

worker=None

classcelery.events.state.Worker(hostname=None, pid=None, freq=60, heartbeats=None, clock=0, active=None, processed=None, loadavg=None, sw_ident=None, sw_ver=None, sw_sys=None)

Worker State.

active

propertyalive

clock

event

expire_window=200

freq

propertyheartbeat_expires

heartbeat_max=4

heartbeats

hostname

propertyid

loadavg

pid

processed

propertystatus_string

sw_ident

sw_sys

sw_ver

update(f, **kw)

celery.events.state.heartbeat_expires(timestamp, freq=60, expire_window=200, Decimal=<class 'decimal.Decimal'>, float=<class 'float'>, isinstance=<built-in function isinstance>)

Return time when heartbeat expires.