Files
deb-python-taskflow/taskflow/engines/action_engine/actions/task.py
Joshua Harlow 359cc490bd Create and use a serial retry executor
To make it easily possible to change the retry
atom execution from being in the engine thread this
creates a retry executor (which is similar to the task
executor) and provide that a serial executor (which it will
use to execute with). This makes the retry and task actions
closer to being the same and makes the surrounding code that
much similar (which makes understanding it easier).

Change-Id: I993e938280df3bd97f8076293183ef21989e2dba
2015-07-21 10:10:23 -07:00

147 lines
5.9 KiB
Python

# -*- coding: utf-8 -*-
# Copyright (C) 2012-2013 Yahoo! Inc. All Rights Reserved.
#
# Licensed 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.
import functools
from taskflow.engines.action_engine.actions import base
from taskflow import logging
from taskflow import states
from taskflow import task as task_atom
from taskflow.types import failure
LOG = logging.getLogger(__name__)
class TaskAction(base.Action):
"""An action that handles scheduling, state changes, ... of task atoms."""
def __init__(self, storage, notifier, task_executor):
super(TaskAction, self).__init__(storage, notifier)
self._task_executor = task_executor
def _is_identity_transition(self, old_state, state, task, progress=None):
if state in self.SAVE_RESULT_STATES:
# saving result is never identity transition
return False
if state != old_state:
# changing state is not identity transition by definition
return False
# NOTE(imelnikov): last thing to check is that the progress has
# changed, which means progress is not None and is different from
# what is stored in the database.
if progress is None:
return False
old_progress = self._storage.get_task_progress(task.name)
if old_progress != progress:
return False
return True
def change_state(self, task, state,
progress=None, result=base.Action.NO_RESULT):
old_state = self._storage.get_atom_state(task.name)
if self._is_identity_transition(old_state, state, task,
progress=progress):
# NOTE(imelnikov): ignore identity transitions in order
# to avoid extra write to storage backend and, what's
# more important, extra notifications.
return
if state in self.SAVE_RESULT_STATES:
save_result = None
if result is not self.NO_RESULT:
save_result = result
self._storage.save(task.name, save_result, state)
else:
self._storage.set_atom_state(task.name, state)
if progress is not None:
self._storage.set_task_progress(task.name, progress)
task_uuid = self._storage.get_atom_uuid(task.name)
details = {
'task_name': task.name,
'task_uuid': task_uuid,
'old_state': old_state,
}
if result is not self.NO_RESULT:
details['result'] = result
self._notifier.notify(state, details)
if progress is not None:
task.update_progress(progress)
def _on_update_progress(self, task, event_type, details):
"""Should be called when task updates its progress."""
try:
progress = details.pop('progress')
except KeyError:
pass
else:
try:
self._storage.set_task_progress(task.name, progress,
details=details)
except Exception:
# Update progress callbacks should never fail, so capture and
# log the emitted exception instead of raising it.
LOG.exception("Failed setting task progress for %s to %0.3f",
task, progress)
def schedule_execution(self, task):
self.change_state(task, states.RUNNING, progress=0.0)
arguments = self._storage.fetch_mapped_args(
task.rebind,
atom_name=task.name,
optional_args=task.optional
)
if task.notifier.can_be_registered(task_atom.EVENT_UPDATE_PROGRESS):
progress_callback = functools.partial(self._on_update_progress,
task)
else:
progress_callback = None
task_uuid = self._storage.get_atom_uuid(task.name)
return self._task_executor.execute_task(
task, task_uuid, arguments,
progress_callback=progress_callback)
def complete_execution(self, task, result):
if isinstance(result, failure.Failure):
self.change_state(task, states.FAILURE, result=result)
else:
self.change_state(task, states.SUCCESS,
result=result, progress=1.0)
def schedule_reversion(self, task):
self.change_state(task, states.REVERTING, progress=0.0)
arguments = self._storage.fetch_mapped_args(
task.rebind,
atom_name=task.name,
optional_args=task.optional
)
task_uuid = self._storage.get_atom_uuid(task.name)
task_result = self._storage.get(task.name)
failures = self._storage.get_failures()
if task.notifier.can_be_registered(task_atom.EVENT_UPDATE_PROGRESS):
progress_callback = functools.partial(self._on_update_progress,
task)
else:
progress_callback = None
return self._task_executor.revert_task(
task, task_uuid, arguments, task_result, failures,
progress_callback=progress_callback)
def complete_reversion(self, task, result):
if isinstance(result, failure.Failure):
self.change_state(task, states.REVERT_FAILURE, result=result)
else:
self.change_state(task, states.REVERTED, progress=1.0,
result=result)