celery.app.control

Worker Remote Control Client.

Client for worker remote control commands. Server implementation is in celery.worker.control. There are two types of remote control commands:

• Inspect commands: Does not have side effects, will usually just return some value found in the worker, like the list of currently registered tasks, the list of active tasks, etc. Commands are accessible via Inspect class.

• Control commands: Performs side effects, like adding a new queue to consume from. Commands are accessible via Control class.

class celery.app.control.Control(app=None)

Worker remote control client.

class Mailbox(namespace, type='direct', connection=None, clock=None, accept=None, serializer=None, producer_pool=None, queue_ttl=None, queue_expires=None, reply_queue_ttl=None, reply_queue_expires=10.0)

Process Mailbox.

Node(hostname=None, state=None, channel=None, handlers=None)

abcast(command, kwargs=None)

accept = ['json']

Only accepts json messages by default.

call(destination, command, kwargs=None, timeout=None, callback=None, channel=None)

cast(destination, command, kwargs=None)

connection = None

Connection (if bound).

exchange = None

mailbox exchange (init by constructor).

exchange_fmt = '%s.pidbox'

get_queue(hostname)

get_reply_queue()

multi_call(command, kwargs=None, timeout=1, limit=None, callback=None, channel=None)

namespace = None

Name of application.

node_cls

Node 的别名

property oid

producer_or_acquire(producer=None, channel=None)

property producer_pool

reply_exchange = None

exchange to send replies to.

reply_exchange_fmt = 'reply.%s.pidbox'

property reply_queue

serializer = None

Message serializer

type = 'direct'

Exchange type (usually direct, or fanout for broadcast).

add_consumer(queue, exchange=None, exchange_type='direct', routing_key=None, options=None, destination=None, **kwargs)

Tell all (or specific) workers to start consuming from a new queue.

Only the queue name is required as if only the queue is specified then the exchange/routing key will be set to the same name ( like automatic queues do).

备注

This command does not respect the default queue/exchange options in the configuration.

参数:

• queue (str) -- Name of queue to start consuming from.

• exchange (str) -- Optional name of exchange.

• exchange_type (str) -- Type of exchange (defaults to 'direct') command to, when empty broadcast to all workers.

• routing_key (str) -- Optional routing key.

• options (Dict) -- Additional options as supported by kombu.entity.Queue.from_dict().

参见

broadcast() for supported keyword arguments.

autoscale(max, min, destination=None, **kwargs)

Change worker(s) autoscale setting.

参见

Supports the same arguments as broadcast().

broadcast(command, arguments=None, destination=None, connection=None, reply=False, timeout=1.0, limit=None, callback=None, channel=None, pattern=None, matcher=None, **extra_kwargs)

Broadcast a control command to the celery workers.

参数:

• command (str) -- Name of command to send.

• arguments (Dict) -- Keyword arguments for the command.

• destination (List) -- If set, a list of the hosts to send the command to, when empty broadcast to all workers.

• connection (kombu.Connection) -- Custom broker connection to use, if not set, a connection will be acquired from the pool.

• reply (bool) -- Wait for and return the reply.

• timeout (float) -- Timeout in seconds to wait for the reply.

• limit (int) -- Limit number of replies.

• callback (Callable) -- Callback called immediately for each reply received.

• pattern (str) -- Custom pattern string to match

• matcher (Callable) -- Custom matcher to run the pattern to match

cancel_consumer(queue, destination=None, **kwargs)

Tell all (or specific) workers to stop consuming from queue.

参见

Supports the same arguments as broadcast().

disable_events(destination=None, **kwargs)

Tell all (or specific) workers to disable events.

参见

Supports the same arguments as broadcast().

discard_all(connection=None)

Discard all waiting tasks.

This will ignore all tasks waiting for execution, and they will be deleted from the messaging server.

参数:

connection (kombu.Connection) -- Optional specific connection instance to use. If not provided a connection will be acquired from the connection pool.

返回:

the number of tasks discarded.

返回类型:

int

election(id, topic, action=None, connection=None)

enable_events(destination=None, **kwargs)

Tell all (or specific) workers to enable events.

参见

Supports the same arguments as broadcast().

heartbeat(destination=None, **kwargs)

Tell worker(s) to send a heartbeat immediately.

参见

Supports the same arguments as broadcast()

property inspect

Create new Inspect instance.

ping(destination=None, timeout=1.0, **kwargs)

Ping all (or specific) workers.

>>> app.control.ping()
[{'celery@node1': {'ok': 'pong'}}, {'celery@node2': {'ok': 'pong'}}]
>>> app.control.ping(destination=['celery@node2'])
[{'celery@node2': {'ok': 'pong'}}]

返回:

List of {HOSTNAME: {'ok': 'pong'}} dictionaries.

返回类型:

List[Dict]

参见

broadcast() for supported keyword arguments.

pool_grow(n=1, destination=None, **kwargs)

Tell all (or specific) workers to grow the pool by n.

参见

Supports the same arguments as broadcast().

pool_restart(modules=None, reload=False, reloader=None, destination=None, **kwargs)

Restart the execution pools of all or specific workers.

关键字参数:

• modules (Sequence[str]) -- List of modules to reload.

• reload (bool) -- Flag to enable module reloading. Default is False.

• reloader (Any) -- Function to reload a module.

• destination (Sequence[str]) -- List of worker names to send this command to.

参见

Supports the same arguments as broadcast()

pool_shrink(n=1, destination=None, **kwargs)

Tell all (or specific) workers to shrink the pool by n.

参见

Supports the same arguments as broadcast().

purge(connection=None)

Discard all waiting tasks.

This will ignore all tasks waiting for execution, and they will be deleted from the messaging server.

参数:

connection (kombu.Connection) -- Optional specific connection instance to use. If not provided a connection will be acquired from the connection pool.

返回:

the number of tasks discarded.

返回类型:

int

rate_limit(task_name, rate_limit, destination=None, **kwargs)

Tell workers to set a new rate limit for task by type.

参数:

• task_name (str) -- Name of task to change rate limit for.

• rate_limit (int, str) -- The rate limit as tasks per second, or a rate limit string ('100/m', etc. see celery.app.task.Task.rate_limit for more information).

参见

broadcast() for supported keyword arguments.

revoke(task_id, destination=None, terminate=False, signal='SIGTERM', **kwargs)

Tell all (or specific) workers to revoke a task by id (or list of ids).

If a task is revoked, the workers will ignore the task and not execute it after all.

参数:

• task_id (Union(str, list)) -- Id of the task to revoke (or list of ids).

• terminate (bool) -- Also terminate the process currently working on the task (if any).

• signal (str) -- Name of signal to send to process if terminate. Default is TERM.

参见

broadcast() for supported keyword arguments.

revoke_by_stamped_headers(headers, destination=None, terminate=False, signal='SIGTERM', **kwargs)

Tell all (or specific) workers to revoke a task by headers.

If a task is revoked, the workers will ignore the task and not execute it after all.

参数:

• headers (dict[str, Union(str, list)]) -- Headers to match when revoking tasks.

• terminate (bool) -- Also terminate the process currently working on the task (if any).

• signal (str) -- Name of signal to send to process if terminate. Default is TERM.

参见

broadcast() for supported keyword arguments.

shutdown(destination=None, **kwargs)

Shutdown worker(s).

参见

Supports the same arguments as broadcast()

terminate(task_id, destination=None, signal='SIGTERM', **kwargs)

Tell all (or specific) workers to terminate a task by id (or list of ids).

参见

This is just a shortcut to revoke() with the terminate argument enabled.

time_limit(task_name, soft=None, hard=None, destination=None, **kwargs)

Tell workers to set time limits for a task by type.

参数:

• task_name (str) -- Name of task to change time limits for.

• soft (float) -- New soft time limit (in seconds).

• hard (float) -- New hard time limit (in seconds).

• **kwargs (Any) -- arguments passed on to broadcast().

class celery.app.control.Inspect(destination=None, timeout=1.0, callback=None, connection=None, app=None, limit=None, pattern=None, matcher=None)

API for inspecting workers.

This class provides proxy for accessing Inspect API of workers. The API is defined in celery.worker.control

active(safe=None)

Return list of tasks currently executed by workers.

参数:

safe (Boolean) -- Set to True to disable deserialization.

返回:

Dictionary {HOSTNAME: [TASK_INFO,...]}.

返回类型:

Dict

参见

For TASK_INFO details see query_task() return value.

active_queues()

Return information about queues from which worker consumes tasks.

返回:

Dictionary {HOSTNAME: [QUEUE_INFO, QUEUE_INFO,...]}.

返回类型:

Dict

Here is the list of QUEUE_INFO fields:

• name

• exchange

  • name

  • type

  • arguments

  • durable

  • passive

  • auto_delete

  • delivery_mode

  • no_declare

• routing_key

• queue_arguments

• binding_arguments

• consumer_arguments

• durable

• exclusive

• auto_delete

• no_ack

• alias

• bindings

• no_declare

• expires

• message_ttl

• max_length

• max_length_bytes

• max_priority

参见

See the RabbitMQ/AMQP documentation for more details about queue_info fields.

备注

The queue_info fields are RabbitMQ/AMQP oriented. Not all fields applies for other transports.

app = None

clock()

Get the Clock value on workers.

>>> app.control.inspect().clock()
{'celery@node1': {'clock': 12}}

返回:

Dictionary {HOSTNAME: CLOCK_VALUE}.

返回类型:

Dict

conf(with_defaults=False)

Return configuration of each worker.

参数:

with_defaults (bool) -- if set to True, method returns also configuration options with default values.

返回:

Dictionary {HOSTNAME: WORKER_CONFIGURATION}.

返回类型:

Dict

参见

WORKER_CONFIGURATION is a dictionary containing current configuration options. See 配置和默认值 for possible values.

hello(from_node, revoked=None)

memdump(samples=10)

Dump statistics of previous memsample requests.

备注

Requires the psutils library.

memsample()

Return sample current RSS memory usage.

备注

Requires the psutils library.

objgraph(type='Request', n=200, max_depth=10)

Create graph of uncollected objects (memory-leak debugging).

参数:

• n (int) -- Max number of objects to graph.

• max_depth (int) -- Traverse at most n levels deep.

• type (str) -- Name of object to graph. Default is "Request".

返回:

Dictionary {'filename': FILENAME}

返回类型:

Dict

备注

Requires the objgraph library.

ping(destination=None)

Ping all (or specific) workers.

>>> app.control.inspect().ping()
{'celery@node1': {'ok': 'pong'}, 'celery@node2': {'ok': 'pong'}}
>>> app.control.inspect().ping(destination=['celery@node1'])
{'celery@node1': {'ok': 'pong'}}

参数:

destination (List) -- If set, a list of the hosts to send the command to, when empty broadcast to all workers.

返回:

Dictionary {HOSTNAME: {'ok': 'pong'}}.

返回类型:

Dict

参见

broadcast() for supported keyword arguments.

query_task(*ids)

Return detail of tasks currently executed by workers.

参数:

*ids (str) -- IDs of tasks to be queried.

返回:

Dictionary {HOSTNAME: {TASK_ID: [STATE, TASK_INFO]}}.

返回类型:

Dict

Here is the list of TASK_INFO fields:

• id - ID of the task

• name - Name of the task

• args - Positinal arguments passed to the task

• kwargs - Keyword arguments passed to the task

• type - Type of the task

• hostname - Hostname of the worker processing the task

• time_start - Time of processing start

• acknowledged - True when task was acknowledged to broker

• delivery_info - Dictionary containing delivery information

  • exchange - Name of exchange where task was published

  • routing_key - Routing key used when task was published

  • priority - Priority used when task was published

  • redelivered - True if the task was redelivered

• worker_pid - PID of worker processing the task

registered(*taskinfoitems)

Return all registered tasks per worker.

>>> app.control.inspect().registered()
{'celery@node1': ['task1', 'task1']}
>>> app.control.inspect().registered('serializer', 'max_retries')
{'celery@node1': ['task_foo [serializer=json max_retries=3]', 'tasb_bar [serializer=json max_retries=3]']}

参数:

taskinfoitems (Sequence[str]) -- List of Task attributes to include.

返回:

Dictionary {HOSTNAME: [TASK1_INFO, ...]}.

返回类型:

Dict

registered_tasks(*taskinfoitems)

Return all registered tasks per worker.

>>> app.control.inspect().registered()
{'celery@node1': ['task1', 'task1']}
>>> app.control.inspect().registered('serializer', 'max_retries')
{'celery@node1': ['task_foo [serializer=json max_retries=3]', 'tasb_bar [serializer=json max_retries=3]']}

参数:

taskinfoitems (Sequence[str]) -- List of Task attributes to include.

返回:

Dictionary {HOSTNAME: [TASK1_INFO, ...]}.

返回类型:

Dict

report()

Return human readable report for each worker.

返回:

Dictionary {HOSTNAME: {'ok': REPORT_STRING}}.

返回类型:

Dict

reserved(safe=None)

Return list of currently reserved tasks, not including scheduled/active.

返回:

Dictionary {HOSTNAME: [TASK_INFO,...]}.

返回类型:

Dict

参见

For TASK_INFO details see query_task() return value.

revoked()

Return list of revoked tasks.

>>> app.control.inspect().revoked()
{'celery@node1': ['16f527de-1c72-47a6-b477-c472b92fef7a']}

返回:

Dictionary {HOSTNAME: [TASK_ID, ...]}.

返回类型:

Dict

scheduled(safe=None)

Return list of scheduled tasks with details.

返回:

Dictionary {HOSTNAME: [TASK_SCHEDULED_INFO,...]}.

返回类型:

Dict

Here is the list of TASK_SCHEDULED_INFO fields:

• eta - scheduled time for task execution as string in ISO 8601 format

• priority - priority of the task

• request - field containing TASK_INFO value.

参见

For more details about TASK_INFO see query_task() return value.

stats()

Return statistics of worker.

返回:

Dictionary {HOSTNAME: STAT_INFO}.

返回类型:

Dict

Here is the list of STAT_INFO fields:

• broker - Section for broker information.

  • connect_timeout - Timeout in seconds (int/float) for establishing a new connection.

  • heartbeat - Current heartbeat value (set by client).

  • hostname - Node name of the remote broker.

  • insist - No longer used.

  • login_method - Login method used to connect to the broker.

  • port - Port of the remote broker.

  • ssl - SSL enabled/disabled.

  • transport - Name of transport used (e.g., amqp or redis)

  • transport_options - Options passed to transport.

  • uri_prefix - Some transports expects the host name to be a URL. E.g. redis+socket:///tmp/redis.sock. In this example the URI-prefix will be redis.

  • userid - User id used to connect to the broker with.

  • virtual_host - Virtual host used.

• clock - Value of the workers logical clock. This is a positive integer and should be increasing every time you receive statistics.

• uptime - Numbers of seconds since the worker controller was started

• pid - Process id of the worker instance (Main process).

• pool - Pool-specific section.

  • max-concurrency - Max number of processes/threads/green threads.

  • max-tasks-per-child - Max number of tasks a thread may execute before being recycled.

  • processes - List of PIDs (or thread-id’s).

  • put-guarded-by-semaphore - Internal

  • timeouts - Default values for time limits.

  • writes - Specific to the prefork pool, this shows the distribution of writes to each process in the pool when using async I/O.

• prefetch_count - Current prefetch count value for the task consumer.

• rusage - System usage statistics. The fields available may be different on your platform. From getrusage(2):

  • stime - Time spent in operating system code on behalf of this process.

  • utime - Time spent executing user instructions.

  • maxrss - The maximum resident size used by this process (in kilobytes).

  • idrss - Amount of non-shared memory used for data (in kilobytes times ticks of execution)

  • isrss - Amount of non-shared memory used for stack space (in kilobytes times ticks of execution)

  • ixrss - Amount of memory shared with other processes (in kilobytes times ticks of execution).

  • inblock - Number of times the file system had to read from the disk on behalf of this process.

  • oublock - Number of times the file system has to write to disk on behalf of this process.

  • majflt - Number of page faults that were serviced by doing I/O.

  • minflt - Number of page faults that were serviced without doing I/O.

  • msgrcv - Number of IPC messages received.

  • msgsnd - Number of IPC messages sent.

  • nvcsw - Number of times this process voluntarily invoked a context switch.

  • nivcsw - Number of times an involuntary context switch took place.

  • nsignals - Number of signals received.

  • nswap - The number of times this process was swapped entirely out of memory.

• total - Map of task names and the total number of tasks with that type the worker has accepted since start-up.

celery.app.control.flatten_reply(reply)

Flatten node replies.

Convert from a list of replies in this format:

[{'a@example.com': reply},
 {'b@example.com': reply}]

into this format:

{'a@example.com': reply,
 'b@example.com': reply}