incubator-airflow/airflow/executors/celery_executor.py

610 строки
25 KiB
Python

#
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.
"""CeleryExecutor
.. seealso::
For more information on how the CeleryExecutor works, take a look at the guide:
:ref:`executor:CeleryExecutor`
"""
import datetime
import logging
import math
import operator
import os
import subprocess
import time
import traceback
from collections import OrderedDict
from multiprocessing import Pool, cpu_count
from typing import Any, Dict, List, Mapping, MutableMapping, Optional, Set, Tuple, Union
from celery import Celery, Task, states as celery_states
from celery.backends.base import BaseKeyValueStoreBackend
from celery.backends.database import DatabaseBackend, session_cleanup
from celery.result import AsyncResult
from celery.signals import import_modules as celery_import_modules
from setproctitle import setproctitle # pylint: disable=no-name-in-module
import airflow.settings as settings
from airflow.config_templates.default_celery import DEFAULT_CELERY_CONFIG
from airflow.configuration import conf
from airflow.exceptions import AirflowException, AirflowTaskTimeout
from airflow.executors.base_executor import BaseExecutor, CommandType, EventBufferValueType
from airflow.models.taskinstance import SimpleTaskInstance, TaskInstance, TaskInstanceKey
from airflow.stats import Stats
from airflow.utils.log.logging_mixin import LoggingMixin
from airflow.utils.net import get_hostname
from airflow.utils.state import State
from airflow.utils.timeout import timeout
from airflow.utils.timezone import utcnow
log = logging.getLogger(__name__)
# Make it constant for unit test.
CELERY_FETCH_ERR_MSG_HEADER = 'Error fetching Celery task state'
CELERY_SEND_ERR_MSG_HEADER = 'Error sending Celery task'
OPERATION_TIMEOUT = conf.getfloat('celery', 'operation_timeout', fallback=1.0)
'''
To start the celery worker, run the command:
airflow celery worker
'''
if conf.has_option('celery', 'celery_config_options'):
celery_configuration = conf.getimport('celery', 'celery_config_options')
else:
celery_configuration = DEFAULT_CELERY_CONFIG
app = Celery(conf.get('celery', 'CELERY_APP_NAME'), config_source=celery_configuration)
@app.task
def execute_command(command_to_exec: CommandType) -> None:
"""Executes command."""
BaseExecutor.validate_command(command_to_exec)
log.info("Executing command in Celery: %s", command_to_exec)
if settings.EXECUTE_TASKS_NEW_PYTHON_INTERPRETER:
_execute_in_subprocess(command_to_exec)
else:
_execute_in_fork(command_to_exec)
def _execute_in_fork(command_to_exec: CommandType) -> None:
pid = os.fork()
if pid:
# In parent, wait for the child
pid, ret = os.waitpid(pid, 0)
if ret == 0:
return
raise AirflowException('Celery command failed on host: ' + get_hostname())
from airflow.sentry import Sentry
ret = 1
try:
from airflow.cli.cli_parser import get_parser
parser = get_parser()
# [1:] - remove "airflow" from the start of the command
args = parser.parse_args(command_to_exec[1:])
setproctitle(f"airflow task supervisor: {command_to_exec}")
args.func(args)
ret = 0
except Exception as e: # pylint: disable=broad-except
log.error("Failed to execute task %s.", str(e))
ret = 1
finally:
Sentry.flush()
os._exit(ret) # pylint: disable=protected-access
def _execute_in_subprocess(command_to_exec: CommandType) -> None:
env = os.environ.copy()
try:
# pylint: disable=unexpected-keyword-arg
subprocess.check_output(command_to_exec, stderr=subprocess.STDOUT, close_fds=True, env=env)
# pylint: disable=unexpected-keyword-arg
except subprocess.CalledProcessError as e:
log.exception('execute_command encountered a CalledProcessError')
log.error(e.output)
msg = 'Celery command failed on host: ' + get_hostname()
raise AirflowException(msg)
class ExceptionWithTraceback:
"""
Wrapper class used to propagate exceptions to parent processes from subprocesses.
:param exception: The exception to wrap
:type exception: Exception
:param exception_traceback: The stacktrace to wrap
:type exception_traceback: str
"""
def __init__(self, exception: Exception, exception_traceback: str):
self.exception = exception
self.traceback = exception_traceback
# Task instance that is sent over Celery queues
# TaskInstanceKey, SimpleTaskInstance, Command, queue_name, CallableTask
TaskInstanceInCelery = Tuple[TaskInstanceKey, SimpleTaskInstance, CommandType, Optional[str], Task]
def send_task_to_executor(
task_tuple: TaskInstanceInCelery,
) -> Tuple[TaskInstanceKey, CommandType, Union[AsyncResult, ExceptionWithTraceback]]:
"""Sends task to executor."""
key, _, command, queue, task_to_run = task_tuple
try:
with timeout(seconds=OPERATION_TIMEOUT):
result = task_to_run.apply_async(args=[command], queue=queue)
except Exception as e: # pylint: disable=broad-except
exception_traceback = f"Celery Task ID: {key}\n{traceback.format_exc()}"
result = ExceptionWithTraceback(e, exception_traceback)
return key, command, result
# pylint: disable=unused-import
@celery_import_modules.connect
def on_celery_import_modules(*args, **kwargs):
"""
Preload some "expensive" airflow modules so that every task process doesn't have to import it again and
again.
Loading these for each task adds 0.3-0.5s *per task* before the task can run. For long running tasks this
doesn't matter, but for short tasks this starts to be a noticeable impact.
"""
import jinja2.ext # noqa: F401
import numpy # noqa: F401
import airflow.jobs.local_task_job
import airflow.macros
import airflow.operators.bash
import airflow.operators.python
import airflow.operators.subdag # noqa: F401
try:
import kubernetes.client # noqa: F401
except ImportError:
pass
# pylint: enable=unused-import
class CeleryExecutor(BaseExecutor):
"""
CeleryExecutor is recommended for production use of Airflow. It allows
distributing the execution of task instances to multiple worker nodes.
Celery is a simple, flexible and reliable distributed system to process
vast amounts of messages, while providing operations with the tools
required to maintain such a system.
"""
def __init__(self):
super().__init__()
# Celery doesn't support bulk sending the tasks (which can become a bottleneck on bigger clusters)
# so we use a multiprocessing pool to speed this up.
# How many worker processes are created for checking celery task state.
self._sync_parallelism = conf.getint('celery', 'SYNC_PARALLELISM')
if self._sync_parallelism == 0:
self._sync_parallelism = max(1, cpu_count() - 1)
self.bulk_state_fetcher = BulkStateFetcher(self._sync_parallelism)
self.tasks = {}
# Mapping of tasks we've adopted, ordered by the earliest date they timeout
self.adopted_task_timeouts: Dict[TaskInstanceKey, datetime.datetime] = OrderedDict()
self.task_adoption_timeout = datetime.timedelta(
seconds=conf.getint('celery', 'task_adoption_timeout', fallback=600)
)
self.task_publish_retries: Dict[TaskInstanceKey, int] = OrderedDict()
self.task_publish_max_retries = conf.getint('celery', 'task_publish_max_retries', fallback=3)
def start(self) -> None:
self.log.debug('Starting Celery Executor using %s processes for syncing', self._sync_parallelism)
def _num_tasks_per_send_process(self, to_send_count: int) -> int:
"""
How many Celery tasks should each worker process send.
:return: Number of tasks that should be sent per process
:rtype: int
"""
return max(1, int(math.ceil(1.0 * to_send_count / self._sync_parallelism)))
def trigger_tasks(self, open_slots: int) -> None:
"""
Overwrite trigger_tasks function from BaseExecutor
:param open_slots: Number of open slots
:return:
"""
sorted_queue = self.order_queued_tasks_by_priority()
task_tuples_to_send: List[TaskInstanceInCelery] = []
for _ in range(min(open_slots, len(self.queued_tasks))):
key, (command, _, queue, simple_ti) = sorted_queue.pop(0)
task_tuple = (key, simple_ti, command, queue, execute_command)
task_tuples_to_send.append(task_tuple)
if key not in self.task_publish_retries:
self.task_publish_retries[key] = 1
if task_tuples_to_send:
self._process_tasks(task_tuples_to_send)
def _process_tasks(self, task_tuples_to_send: List[TaskInstanceInCelery]) -> None:
first_task = next(t[4] for t in task_tuples_to_send)
# Celery state queries will stuck if we do not use one same backend
# for all tasks.
cached_celery_backend = first_task.backend
key_and_async_results = self._send_tasks_to_celery(task_tuples_to_send)
self.log.debug('Sent all tasks.')
for key, _, result in key_and_async_results:
if isinstance(result, ExceptionWithTraceback) and isinstance(
result.exception, AirflowTaskTimeout
):
if key in self.task_publish_retries and (
self.task_publish_retries.get(key) <= self.task_publish_max_retries
):
Stats.incr("celery.task_timeout_error")
self.log.info(
"[Try %s of %s] Task Timeout Error for Task: (%s).",
self.task_publish_retries[key],
self.task_publish_max_retries,
key,
)
self.task_publish_retries[key] += 1
continue
self.queued_tasks.pop(key)
self.task_publish_retries.pop(key)
if isinstance(result, ExceptionWithTraceback):
self.log.error( # pylint: disable=logging-not-lazy
CELERY_SEND_ERR_MSG_HEADER + ": %s\n%s\n", result.exception, result.traceback
)
self.event_buffer[key] = (State.FAILED, None)
elif result is not None:
result.backend = cached_celery_backend
self.running.add(key)
self.tasks[key] = result
# Store the Celery task_id in the event buffer. This will get "overwritten" if the task
# has another event, but that is fine, because the only other events are success/failed at
# which point we don't need the ID anymore anyway
self.event_buffer[key] = (State.QUEUED, result.task_id)
# If the task runs _really quickly_ we may already have a result!
self.update_task_state(key, result.state, getattr(result, 'info', None))
def _send_tasks_to_celery(self, task_tuples_to_send: List[TaskInstanceInCelery]):
if len(task_tuples_to_send) == 1 or self._sync_parallelism == 1:
# One tuple, or max one process -> send it in the main thread.
return list(map(send_task_to_executor, task_tuples_to_send))
# Use chunks instead of a work queue to reduce context switching
# since tasks are roughly uniform in size
chunksize = self._num_tasks_per_send_process(len(task_tuples_to_send))
num_processes = min(len(task_tuples_to_send), self._sync_parallelism)
def reset_signals():
# Since we are run from inside the SchedulerJob, we don't to
# inherit the signal handlers that we registered there.
import signal
signal.signal(signal.SIGINT, signal.SIG_DFL)
signal.signal(signal.SIGTERM, signal.SIG_DFL)
signal.signal(signal.SIGUSR2, signal.SIG_DFL)
with Pool(processes=num_processes, initializer=reset_signals) as send_pool:
key_and_async_results = send_pool.map(
send_task_to_executor, task_tuples_to_send, chunksize=chunksize
)
return key_and_async_results
def sync(self) -> None:
if not self.tasks:
self.log.debug("No task to query celery, skipping sync")
return
self.update_all_task_states()
if self.adopted_task_timeouts:
self._check_for_stalled_adopted_tasks()
def _check_for_stalled_adopted_tasks(self):
"""
See if any of the tasks we adopted from another Executor run have not
progressed after the configured timeout.
If they haven't, they likely never made it to Celery, and we should
just resend them. We do that by clearing the state and letting the
normal scheduler loop deal with that
"""
now = utcnow()
timedout_keys = []
for key, stalled_after in self.adopted_task_timeouts.items():
if stalled_after > now:
# Since items are stored sorted, if we get to a stalled_after
# in the future then we can stop
break
# If the task gets updated to STARTED (which Celery does) or has
# already finished, then it will be removed from this list -- so
# the only time it's still in this list is when it a) never made it
# to celery in the first place (i.e. race condition somehwere in
# the dying executor) or b) a really long celery queue and it just
# hasn't started yet -- better cancel it and let the scheduler
# re-queue rather than have this task risk stalling for ever
timedout_keys.append(key)
if timedout_keys:
self.log.error(
"Adopted tasks were still pending after %s, assuming they never made it to celery and "
"clearing:\n\t%s",
self.task_adoption_timeout,
"\n\t".join([repr(x) for x in timedout_keys]),
)
for key in timedout_keys:
self.event_buffer[key] = (State.FAILED, None)
del self.tasks[key]
del self.adopted_task_timeouts[key]
def debug_dump(self) -> None:
"""Called in response to SIGUSR2 by the scheduler"""
super().debug_dump()
self.log.info(
"executor.tasks (%d)\n\t%s", len(self.tasks), "\n\t".join(map(repr, self.tasks.items()))
)
self.log.info(
"executor.adopted_task_timeouts (%d)\n\t%s",
len(self.adopted_task_timeouts),
"\n\t".join(map(repr, self.adopted_task_timeouts.items())),
)
def update_all_task_states(self) -> None:
"""Updates states of the tasks."""
self.log.debug("Inquiring about %s celery task(s)", len(self.tasks))
state_and_info_by_celery_task_id = self.bulk_state_fetcher.get_many(self.tasks.values())
self.log.debug("Inquiries completed.")
for key, async_result in list(self.tasks.items()):
state, info = state_and_info_by_celery_task_id.get(async_result.task_id)
if state:
self.update_task_state(key, state, info)
def change_state(self, key: TaskInstanceKey, state: str, info=None) -> None:
super().change_state(key, state, info)
self.tasks.pop(key, None)
self.adopted_task_timeouts.pop(key, None)
def update_task_state(self, key: TaskInstanceKey, state: str, info: Any) -> None:
"""Updates state of a single task."""
try:
if state == celery_states.SUCCESS:
self.success(key, info)
elif state in (celery_states.FAILURE, celery_states.REVOKED):
self.fail(key, info)
elif state == celery_states.STARTED:
# It's now actually running, so know it made it to celery okay!
self.adopted_task_timeouts.pop(key, None)
elif state == celery_states.PENDING:
pass
else:
self.log.info("Unexpected state for %s: %s", key, state)
except Exception: # noqa pylint: disable=broad-except
self.log.exception("Error syncing the Celery executor, ignoring it.")
def end(self, synchronous: bool = False) -> None:
if synchronous:
while any(task.state not in celery_states.READY_STATES for task in self.tasks.values()):
time.sleep(5)
self.sync()
def execute_async(
self,
key: TaskInstanceKey,
command: CommandType,
queue: Optional[str] = None,
executor_config: Optional[Any] = None,
):
"""Do not allow async execution for Celery executor."""
raise AirflowException("No Async execution for Celery executor.")
def terminate(self):
pass
def try_adopt_task_instances(self, tis: List[TaskInstance]) -> List[TaskInstance]:
# See which of the TIs are still alive (or have finished even!)
#
# Since Celery doesn't store "SENT" state for queued commands (if we create an AsyncResult with a made
# up id it just returns PENDING state for it), we have to store Celery's task_id against the TI row to
# look at in future.
#
# This process is not perfect -- we could have sent the task to celery, and crashed before we were
# able to record the AsyncResult.task_id in the TaskInstance table, in which case we won't adopt the
# task (it'll either run and update the TI state, or the scheduler will clear and re-queue it. Either
# way it won't get executed more than once)
#
# (If we swapped it around, and generated a task_id for Celery, stored that in TI and enqueued that
# there is also still a race condition where we could generate and store the task_id, but die before
# we managed to enqueue the command. Since neither way is perfect we always have to deal with this
# process not being perfect.)
celery_tasks = {}
not_adopted_tis = []
for ti in tis:
if ti.external_executor_id is not None:
celery_tasks[ti.external_executor_id] = (AsyncResult(ti.external_executor_id), ti)
else:
not_adopted_tis.append(ti)
if not celery_tasks:
# Nothing to adopt
return tis
states_by_celery_task_id = self.bulk_state_fetcher.get_many(
map(operator.itemgetter(0), celery_tasks.values())
)
adopted = []
cached_celery_backend = next(iter(celery_tasks.values()))[0].backend
for celery_task_id, (state, info) in states_by_celery_task_id.items():
result, ti = celery_tasks[celery_task_id]
result.backend = cached_celery_backend
# Set the correct elements of the state dicts, then update this
# like we just queried it.
self.adopted_task_timeouts[ti.key] = ti.queued_dttm + self.task_adoption_timeout
self.tasks[ti.key] = result
self.running.add(ti.key)
self.update_task_state(ti.key, state, info)
adopted.append(f"{ti} in state {state}")
if adopted:
task_instance_str = '\n\t'.join(adopted)
self.log.info(
"Adopted the following %d tasks from a dead executor\n\t%s", len(adopted), task_instance_str
)
return not_adopted_tis
def fetch_celery_task_state(async_result: AsyncResult) -> Tuple[str, Union[str, ExceptionWithTraceback], Any]:
"""
Fetch and return the state of the given celery task. The scope of this function is
global so that it can be called by subprocesses in the pool.
:param async_result: a tuple of the Celery task key and the async Celery object used
to fetch the task's state
:type async_result: tuple(str, celery.result.AsyncResult)
:return: a tuple of the Celery task key and the Celery state and the celery info
of the task
:rtype: tuple[str, str, str]
"""
try:
with timeout(seconds=OPERATION_TIMEOUT):
# Accessing state property of celery task will make actual network request
# to get the current state of the task
info = async_result.info if hasattr(async_result, 'info') else None
return async_result.task_id, async_result.state, info
except Exception as e: # pylint: disable=broad-except
exception_traceback = f"Celery Task ID: {async_result}\n{traceback.format_exc()}"
return async_result.task_id, ExceptionWithTraceback(e, exception_traceback), None
def _tasks_list_to_task_ids(async_tasks) -> Set[str]:
return {a.task_id for a in async_tasks}
class BulkStateFetcher(LoggingMixin):
"""
Gets status for many Celery tasks using the best method available
If BaseKeyValueStoreBackend is used as result backend, the mget method is used.
If DatabaseBackend is used as result backend, the SELECT ...WHERE task_id IN (...) query is used
Otherwise, multiprocessing.Pool will be used. Each task status will be downloaded individually.
"""
def __init__(self, sync_parralelism=None):
super().__init__()
self._sync_parallelism = sync_parralelism
def get_many(self, async_results) -> Mapping[str, EventBufferValueType]:
"""Gets status for many Celery tasks using the best method available."""
if isinstance(app.backend, BaseKeyValueStoreBackend):
result = self._get_many_from_kv_backend(async_results)
return result
if isinstance(app.backend, DatabaseBackend):
result = self._get_many_from_db_backend(async_results)
return result
result = self._get_many_using_multiprocessing(async_results)
self.log.debug("Fetched %d states for %d task", len(result), len(async_results))
return result
def _get_many_from_kv_backend(self, async_tasks) -> Mapping[str, EventBufferValueType]:
task_ids = _tasks_list_to_task_ids(async_tasks)
keys = [app.backend.get_key_for_task(k) for k in task_ids]
values = app.backend.mget(keys)
task_results = [app.backend.decode_result(v) for v in values if v]
task_results_by_task_id = {task_result["task_id"]: task_result for task_result in task_results}
return self._prepare_state_and_info_by_task_dict(task_ids, task_results_by_task_id)
def _get_many_from_db_backend(self, async_tasks) -> Mapping[str, EventBufferValueType]:
task_ids = _tasks_list_to_task_ids(async_tasks)
session = app.backend.ResultSession()
task_cls = app.backend.task_cls
with session_cleanup(session):
tasks = session.query(task_cls).filter(task_cls.task_id.in_(task_ids)).all()
task_results = [app.backend.meta_from_decoded(task.to_dict()) for task in tasks]
task_results_by_task_id = {task_result["task_id"]: task_result for task_result in task_results}
return self._prepare_state_and_info_by_task_dict(task_ids, task_results_by_task_id)
@staticmethod
def _prepare_state_and_info_by_task_dict(
task_ids, task_results_by_task_id
) -> Mapping[str, EventBufferValueType]:
state_info: MutableMapping[str, EventBufferValueType] = {}
for task_id in task_ids:
task_result = task_results_by_task_id.get(task_id)
if task_result:
state = task_result["status"]
info = None if not hasattr(task_result, "info") else task_result["info"]
else:
state = celery_states.PENDING
info = None
state_info[task_id] = state, info
return state_info
def _get_many_using_multiprocessing(self, async_results) -> Mapping[str, EventBufferValueType]:
num_process = min(len(async_results), self._sync_parallelism)
with Pool(processes=num_process) as sync_pool:
chunksize = max(1, math.floor(math.ceil(1.0 * len(async_results) / self._sync_parallelism)))
task_id_to_states_and_info = sync_pool.map(
fetch_celery_task_state, async_results, chunksize=chunksize
)
states_and_info_by_task_id: MutableMapping[str, EventBufferValueType] = {}
for task_id, state_or_exception, info in task_id_to_states_and_info:
if isinstance(state_or_exception, ExceptionWithTraceback):
self.log.error( # pylint: disable=logging-not-lazy
CELERY_FETCH_ERR_MSG_HEADER + ":%s\n%s\n",
state_or_exception.exception,
state_or_exception.traceback,
)
else:
states_and_info_by_task_id[task_id] = state_or_exception, info
return states_and_info_by_task_id