#
# 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.
import json
from sqlalchemy import BigInteger, Column, Index, Integer, String, Text
from airflow.configuration import conf
from airflow.exceptions import AirflowException
from airflow.models.base import ID_LEN, Base
from airflow.utils import timezone
from airflow.utils.session import provide_session
from airflow.utils.sqlalchemy import UtcDateTime
from airflow.utils.state import State
[docs]class SensorInstance(Base):
"""
SensorInstance support the smart sensor service. It stores the sensor task states
and context that required for poking include poke context and execution context.
In sensor_instance table we also save the sensor operator classpath so that inside
smart sensor there is no need to import the dagbag and create task object for each
sensor task.
SensorInstance include another set of columns to support the smart sensor shard on
large number of sensor instance. The key idea is to generate the hash code from the
poke context and use it to map to a shorter shard code which can be used as an index.
Every smart sensor process takes care of tasks whose `shardcode` are in a certain range.
"""
[docs] __tablename__ = "sensor_instance"
[docs] id = Column(Integer, primary_key=True)
[docs] task_id = Column(String(ID_LEN), nullable=False)
[docs] dag_id = Column(String(ID_LEN), nullable=False)
[docs] execution_date = Column(UtcDateTime, nullable=False)
[docs] state = Column(String(20))
_try_number = Column('try_number', Integer, default=0)
[docs] start_date = Column(UtcDateTime)
[docs] operator = Column(String(1000), nullable=False)
[docs] op_classpath = Column(String(1000), nullable=False)
[docs] hashcode = Column(BigInteger, nullable=False)
[docs] shardcode = Column(Integer, nullable=False)
[docs] poke_context = Column(Text, nullable=False)
[docs] execution_context = Column(Text)
[docs] created_at = Column(UtcDateTime, default=timezone.utcnow, nullable=False)
[docs] updated_at = Column(UtcDateTime, default=timezone.utcnow, onupdate=timezone.utcnow, nullable=False)
# SmartSensor doesn't support mapped operators, but this is needed for compatibly with the
# log_filename_template of TaskInstances
[docs] __table_args__ = (
Index('ti_primary_key', dag_id, task_id, execution_date, unique=True),
Index('si_hashcode', hashcode),
Index('si_shardcode', shardcode),
Index('si_state_shard', state, shardcode),
Index('si_updated_at', updated_at),
)
def __init__(self, ti):
self.dag_id = ti.dag_id
self.task_id = ti.task_id
self.execution_date = ti.execution_date
@staticmethod
[docs] def get_classpath(obj):
"""
Get the object dotted class path. Used for getting operator classpath.
:param obj:
:return: The class path of input object
:rtype: str
"""
module_name, class_name = obj.__module__, obj.__class__.__name__
return module_name + "." + class_name
@classmethod
@provide_session
[docs] def register(cls, ti, poke_context, execution_context, session=None):
"""
Register task instance ti for a sensor in sensor_instance table. Persist the
context used for a sensor and set the sensor_instance table state to sensing.
:param ti: The task instance for the sensor to be registered.
:param poke_context: Context used for sensor poke function.
:param execution_context: Context used for execute sensor such as timeout
setting and email configuration.
:param session: SQLAlchemy ORM Session
:return: True if the ti was registered successfully.
:rtype: Boolean
"""
if poke_context is None:
raise AirflowException('poke_context should not be None')
encoded_poke = json.dumps(poke_context)
encoded_execution_context = json.dumps(execution_context)
sensor = (
session.query(SensorInstance)
.filter(
SensorInstance.dag_id == ti.dag_id,
SensorInstance.task_id == ti.task_id,
SensorInstance.execution_date == ti.execution_date,
)
.with_for_update()
.first()
)
if sensor is None:
sensor = SensorInstance(ti=ti)
sensor.operator = ti.operator
sensor.op_classpath = SensorInstance.get_classpath(ti.task)
sensor.poke_context = encoded_poke
sensor.execution_context = encoded_execution_context
sensor.hashcode = hash(encoded_poke)
sensor.shardcode = sensor.hashcode % conf.getint('smart_sensor', 'shard_code_upper_limit')
sensor.try_number = ti.try_number
sensor.state = State.SENSING
sensor.start_date = timezone.utcnow()
session.add(sensor)
session.commit()
return True
@property
[docs] def try_number(self):
"""
Return the try number that this task number will be when it is actually
run.
If the TI is currently running, this will match the column in the
database, in all other cases this will be incremented.
"""
# This is designed so that task logs end up in the right file.
if self.state in State.running:
return self._try_number
return self._try_number + 1
@try_number.setter
def try_number(self, value):
self._try_number = value
[docs] def __repr__(self):
return (
"<{self.__class__.__name__}: id: {self.id} poke_context: {self.poke_context} "
"execution_context: {self.execution_context} state: {self.state}>".format(self=self)
)
@provide_session
[docs] def get_dagrun(self, session):
"""
Returns the DagRun for this SensorInstance
:param session: SQLAlchemy ORM Session
:return: DagRun
"""
from airflow.models.dagrun import DagRun # Avoid circular import
dr = (
session.query(DagRun)
.filter(DagRun.dag_id == self.dag_id, DagRun.execution_date == self.execution_date)
.one()
)
return dr
