Python UnitOfWorkDao.recover_from_duplicatekeyerror方法代码示例

# 需要导入模块: from synergy.db.dao.unit_of_work_dao import UnitOfWorkDao [as 别名]
# 或者: from synergy.db.dao.unit_of_work_dao.UnitOfWorkDao import recover_from_duplicatekeyerror [as 别名]
class AbstractStateMachine(object):
    """ Abstract state machine used to govern all processes and their states """

    def __init__(self, logger, timetable, name):
        self.name = name
        self.logger = logger
        self.publishers = PublishersPool(self.logger)
        self.timetable = timetable
        self.uow_dao = UnitOfWorkDao(self.logger)
        self.job_dao = JobDao(self.logger)

    def __del__(self):
        try:
            self.logger.info('Closing Flopsy Publishers Pool...')
            self.publishers.close()
        except Exception as e:
            self.logger.error('Exception caught while closing Flopsy Publishers Pool: %s' % str(e))

    def _log_message(self, level, process_name, timeperiod, msg):
        """ method performs logging into log file and Timetable's tree node"""
        self.timetable.add_log_entry(process_name, timeperiod, msg)
        self.logger.log(level, msg)

    @with_reconnect
    def _insert_uow(self, process_name, start_timeperiod, end_timeperiod, start_id, end_id):
        """creates unit_of_work and inserts it into the DB
            :raise DuplicateKeyError: if unit_of_work with given parameters already exists """
        uow = UnitOfWork()
        uow.process_name = process_name
        uow.timeperiod = start_timeperiod
        uow.start_id = str(start_id)
        uow.end_id = str(end_id)
        uow.start_timeperiod = start_timeperiod
        uow.end_timeperiod = end_timeperiod
        uow.created_at = datetime.utcnow()
        uow.source = context.process_context[process_name].source
        uow.sink = context.process_context[process_name].sink
        uow.state = unit_of_work.STATE_REQUESTED
        uow.unit_of_work_type = TYPE_MANAGED
        uow.number_of_retries = 0
        uow.arguments = context.process_context[process_name].arguments
        uow.db_id = self.uow_dao.insert(uow)

        msg = 'Created: UOW %s for %s in timeperiod [%s:%s).' \
              % (uow.db_id, process_name, start_timeperiod, end_timeperiod)
        self._log_message(INFO, process_name, start_timeperiod, msg)
        return uow

    def _publish_uow(self, uow):
        mq_request = SynergyMqTransmission(process_name=uow.process_name, unit_of_work_id=uow.db_id)

        publisher = self.publishers.get(uow.process_name)
        publisher.publish(mq_request.document)
        publisher.release()

        msg = 'Published: UOW %r for %r in timeperiod %r.' % (uow.db_id, uow.process_name, uow.start_timeperiod)
        self._log_message(INFO, uow.process_name, uow.start_timeperiod, msg)

    def insert_and_publish_uow(self, process_name, start_timeperiod, end_timeperiod, start_id, end_id):
        """ method creates and publishes a unit_of_work. it also handles DuplicateKeyError and attempts recovery
        :return: tuple (uow, is_duplicate)
        :raise UserWarning: if the recovery from DuplicateKeyError was unsuccessful
        """
        is_duplicate = False
        try:
            uow = self._insert_uow(process_name, start_timeperiod, end_timeperiod, start_id, end_id)
        except DuplicateKeyError as e:
            is_duplicate = True
            msg = 'Catching up with latest unit_of_work %s in timeperiod %s, because of: %r' \
                  % (process_name, start_timeperiod, e)
            self._log_message(WARNING, process_name, start_timeperiod, msg)
            uow = self.uow_dao.recover_from_duplicatekeyerror(e)

        if uow is None:
            msg = 'MANUAL INTERVENTION REQUIRED! Unable to locate unit_of_work for %s in %s' \
                  % (process_name, start_timeperiod)
            self._log_message(WARNING, process_name, start_timeperiod, msg)
            raise UserWarning(msg)

        # publish the created/caught up unit_of_work
        self._publish_uow(uow)
        return uow, is_duplicate

    def shallow_state_update(self, uow):
        """ method does not trigger any new actions
        if applicable, it will update job_record state and Timetable tree node state
        :assumptions: uow is either in STATE_CANCELED or STATE_PROCESSED """
        pass

    def _process_state_embryo(self, job_record):
        """ method that takes care of processing job records in STATE_EMBRYO state"""
        pass

    def _process_state_in_progress(self, job_record):
        """ method that takes care of processing job records in STATE_IN_PROGRESS state"""
        pass

    def _process_state_final_run(self, job_record):
        """method takes care of processing job records in STATE_FINAL_RUN state"""
        pass
#.........这里部分代码省略.........

# 需要导入模块: from synergy.db.dao.unit_of_work_dao import UnitOfWorkDao [as 别名]
# 或者: from synergy.db.dao.unit_of_work_dao.UnitOfWorkDao import recover_from_duplicatekeyerror [as 别名]
class StateMachineFreerun(object):
    """ State Machine to handle freerun jobs/triggers """

    def __init__(self, logger, name=STATE_MACHINE_FREERUN):
        self.name = name
        self.logger = logger
        self.mq_transmitter = MqTransmitter(self.logger)
        self.uow_dao = UnitOfWorkDao(self.logger)
        self.sfe_dao = FreerunProcessDao(self.logger)

    @with_reconnect
    def _log_message(self, level, freerun_entry, msg):
        """ method performs logging into log file and the freerun_entry """
        self.logger.log(level, msg)

        assert isinstance(freerun_entry, FreerunProcessEntry)
        event_log = freerun_entry.event_log
        if len(event_log) > MAX_NUMBER_OF_EVENTS:
            del event_log[-1]
        event_log.insert(0, msg)
        self.sfe_dao.update(freerun_entry)

    @with_reconnect
    def _insert_uow(self, freerun_entry):
        """ creates unit_of_work and inserts it into the DB
            :raise DuplicateKeyError: if unit_of_work with given parameters already exists """
        current_timeperiod = time_helper.actual_timeperiod(QUALIFIER_REAL_TIME)
        process_entry = context.process_context[freerun_entry.process_name]

        uow = UnitOfWork()
        uow.process_name = freerun_entry.schedulable_name
        uow.timeperiod = current_timeperiod
        uow.start_id = 0
        uow.end_id = 0
        uow.start_timeperiod = current_timeperiod
        uow.end_timeperiod = current_timeperiod
        uow.created_at = datetime.utcnow()
        uow.submitted_at = datetime.utcnow()
        uow.source = process_entry.source if hasattr(process_entry, 'source') else None
        uow.sink = process_entry.sink if hasattr(process_entry, 'sink') else None
        uow.state = unit_of_work.STATE_REQUESTED
        uow.unit_of_work_type = unit_of_work.TYPE_FREERUN
        uow.number_of_retries = 0
        uow.arguments = freerun_entry.arguments
        uow.db_id = self.uow_dao.insert(uow)

        msg = 'Created: UOW {0} for {1}@{2}.' \
              .format(uow.db_id, freerun_entry.schedulable_name, current_timeperiod)
        self._log_message(INFO, freerun_entry, msg)
        return uow

    def _publish_uow(self, freerun_entry, uow):
        self.mq_transmitter.publish_freerun_uow(freerun_entry, uow)
        msg = 'Published: UOW {0} for {1}.'.format(uow.db_id, freerun_entry.schedulable_name)
        self._log_message(INFO, freerun_entry, msg)

    def insert_and_publish_uow(self, freerun_entry):
        try:
            uow = self._insert_uow(freerun_entry)
        except DuplicateKeyError as e:
            msg = 'Duplication of UOW found for {0}. Error msg: {1}'.format(freerun_entry.schedulable_name, e)
            self._log_message(WARNING, freerun_entry, msg)
            uow = self.uow_dao.recover_from_duplicatekeyerror(e)

        if uow is not None:
            # publish the created/caught up unit_of_work
            self._publish_uow(freerun_entry, uow)
            freerun_entry.related_unit_of_work = uow.db_id
            self.sfe_dao.update(freerun_entry)
        else:
            msg = 'PERSISTENT TIER ERROR! Unable to locate UOW for {0}' \
                  .format(freerun_entry.schedulable_name)
            self._log_message(WARNING, freerun_entry, msg)

    def _process_state_embryo(self, freerun_entry):
        """ method creates unit_of_work and associates it with the FreerunProcessEntry """
        self.insert_and_publish_uow(freerun_entry)

    def _process_state_in_progress(self, freerun_entry, uow):
        """ method that takes care of processing unit_of_work records in STATE_REQUESTED or STATE_IN_PROGRESS states"""
        self._publish_uow(freerun_entry, uow)

    def _process_terminal_state(self, freerun_entry, uow):
        """ method that takes care of processing unit_of_work records in
            STATE_PROCESSED, STATE_NOOP, STATE_INVALID, STATE_CANCELED states"""
        msg = 'UOW for {0} found in state {1}.'.format(freerun_entry.schedulable_name, uow.state)
        self._log_message(INFO, freerun_entry, msg)
        self.insert_and_publish_uow(freerun_entry)

    def manage_schedulable(self, freerun_entry):
        """ method main duty - is to _avoid_ publishing another unit_of_work, if previous was not yet processed
        In case the Scheduler sees that the unit_of_work is pending it will fire another WorkerMqRequest """

        assert isinstance(freerun_entry, FreerunProcessEntry)
        if freerun_entry.related_unit_of_work is None:
            uow = None
        else:
            uow = self.uow_dao.get_one(freerun_entry.related_unit_of_work)

        try:
#.........这里部分代码省略.........

# 需要导入模块: from synergy.db.dao.unit_of_work_dao import UnitOfWorkDao [as 别名]
# 或者: from synergy.db.dao.unit_of_work_dao.UnitOfWorkDao import recover_from_duplicatekeyerror [as 别名]
class AbstractStateMachine(object):
    """ Abstract state machine used to host common logic for the rest of registered State Machines """

    def __init__(self, logger, timetable, name):
        self.name = name
        self.logger = logger
        self.mq_transmitter = MqTransmitter(self.logger)
        self.timetable = timetable
        self.uow_dao = UnitOfWorkDao(self.logger)
        self.job_dao = JobDao(self.logger)

    @property
    def run_on_active_timeperiod(self):
        """
        :return: True if given State Machine allows execution on *live* timeperiod, as opposed to the finished one
        """
        raise NotImplementedError('property run_on_active_timeperiod must be implemented by {0}'
                                  .format(self.__class__.__name__))

    def _log_message(self, level, process_name, timeperiod, msg):
        """ method performs logging into log file and Timetable's tree node"""
        self.timetable.add_log_entry(process_name, timeperiod, msg)
        self.logger.log(level, msg)

    @with_reconnect
    def _insert_uow(self, process_name, timeperiod, start_timeperiod, end_timeperiod, start_id, end_id):
        """creates unit_of_work and inserts it into the DB
            :raise DuplicateKeyError: if unit_of_work with given parameters already exists """
        uow = UnitOfWork()
        uow.process_name = process_name
        uow.timeperiod = timeperiod
        uow.start_id = str(start_id)
        uow.end_id = str(end_id)
        uow.start_timeperiod = start_timeperiod
        uow.end_timeperiod = end_timeperiod
        uow.created_at = datetime.utcnow()
        uow.submitted_at = datetime.utcnow()
        uow.source = context.process_context[process_name].source
        uow.sink = context.process_context[process_name].sink
        uow.state = unit_of_work.STATE_REQUESTED
        uow.unit_of_work_type = unit_of_work.TYPE_MANAGED
        uow.number_of_retries = 0
        uow.arguments = context.process_context[process_name].arguments
        uow.db_id = self.uow_dao.insert(uow)

        msg = 'Created: UOW {0} for {1}@{2}.'.format(uow.db_id, process_name, start_timeperiod)
        self._log_message(INFO, process_name, start_timeperiod, msg)
        return uow

    def _publish_uow(self, uow):
        self.mq_transmitter.publish_managed_uow(uow)
        msg = 'Published: UOW {0} for {1}@{2}.'.format(uow.db_id, uow.process_name, uow.start_timeperiod)
        self._log_message(INFO, uow.process_name, uow.start_timeperiod, msg)

    def insert_and_publish_uow(self, job_record, start_id, end_id):
        """ method creates and publishes a unit_of_work. it also handles DuplicateKeyError and attempts recovery
        :return: tuple (uow, is_duplicate)
        :raise UserWarning: if the recovery from DuplicateKeyError was unsuccessful
        """
        process_name = job_record.process_name
        timeperiod = job_record.timeperiod
        start_timeperiod = self.compute_start_timeperiod(job_record.process_name, job_record.timeperiod)
        end_timeperiod = self.compute_end_timeperiod(job_record.process_name, job_record.timeperiod)

        try:
            is_duplicate = False
            uow = self._insert_uow(process_name, timeperiod, start_timeperiod, end_timeperiod, start_id, end_id)
        except DuplicateKeyError as e:
            is_duplicate = True
            msg = 'Catching up with latest UOW {0}@{1}, because of: {2}' \
                  .format(process_name, start_timeperiod, e)
            self._log_message(WARNING, process_name, start_timeperiod, msg)
            uow = self.uow_dao.recover_from_duplicatekeyerror(e)

        if not uow:
            msg = 'PERSISTENT TIER ERROR! Unable to locate UOW for {0}@{1}' \
                  .format(process_name, start_timeperiod)
            self._log_message(WARNING, process_name, start_timeperiod, msg)
            raise UserWarning(msg)

        if uow.is_canceled:
            # this UOW was marked for re-processing. recycle it
            uow.created_at = datetime.utcnow()      # reset created_at to bypass GC cancellation logic
            uow.submitted_at = datetime.utcnow()    # reset submitted_at to allow 1 hour free of GC resubmitting
            del uow.started_at
            del uow.finished_at
            del uow.number_of_aggregated_documents
            del uow.number_of_processed_documents
            uow.state = unit_of_work.STATE_REQUESTED
            self.uow_dao.update(uow)

        # publish the created/recovered/recycled unit_of_work
        self._publish_uow(uow)
        return uow, is_duplicate

    def notify(self, uow):
        """ method is used by StateMachine's users to provide notifications about change in UOW
        if applicable, method will update job_record state and Timetable tree node state
        :assumptions: uow is in [STATE_NOOP, STATE_CANCELED, STATE_PROCESSED] """
        pass
#.........这里部分代码省略.........

# 需要导入模块: from synergy.db.dao.unit_of_work_dao import UnitOfWorkDao [as 别名]
# 或者: from synergy.db.dao.unit_of_work_dao.UnitOfWorkDao import recover_from_duplicatekeyerror [as 别名]
class StateMachineFreerun(object):
    """ State Machine to handle freerun jobs/triggers """

    def __init__(self, logger, name=STATE_MACHINE_FREERUN):
        self.name = name
        self.logger = logger
        self.publishers = PublishersPool(self.logger)
        self.uow_dao = UnitOfWorkDao(self.logger)
        self.sfe_dao = FreerunProcessDao(self.logger)

    def __del__(self):
        try:
            self.logger.info('Closing Flopsy Publishers Pool...')
            self.publishers.close()
        except Exception as e:
            self.logger.error('Exception caught while closing Flopsy Publishers Pool: %s' % str(e))

    @with_reconnect
    def _log_message(self, level, freerun_entry, msg):
        """ method performs logging into log file and the freerun_entry """
        self.logger.log(level, msg)

        assert isinstance(freerun_entry, FreerunProcessEntry)
        log = freerun_entry.log
        if len(log) > MAX_NUMBER_OF_LOG_ENTRIES:
            del log[-1]
        log.insert(0, msg)
        self.sfe_dao.update(freerun_entry)

    @with_reconnect
    def _insert_uow(self, freerun_entry):
        """ creates unit_of_work and inserts it into the DB
            :raise DuplicateKeyError: if unit_of_work with given parameters already exists """
        current_timeperiod = time_helper.actual_timeperiod(QUALIFIER_REAL_TIME)

        uow = UnitOfWork()
        uow.process_name = freerun_entry.schedulable_name
        uow.timeperiod = current_timeperiod
        uow.start_id = 0
        uow.end_id = 0
        uow.start_timeperiod = current_timeperiod
        uow.end_timeperiod = current_timeperiod
        uow.created_at = datetime.utcnow()
        uow.source = context.process_context[freerun_entry.process_name].source
        uow.sink = context.process_context[freerun_entry.process_name].sink
        uow.state = unit_of_work.STATE_REQUESTED
        uow.unit_of_work_type = TYPE_FREERUN
        uow.number_of_retries = 0
        uow.arguments = freerun_entry.arguments
        uow.db_id = self.uow_dao.insert(uow)

        msg = 'Created: UOW %s for %s in timeperiod %s.' \
              % (uow.db_id, freerun_entry.schedulable_name, current_timeperiod)
        self._log_message(INFO, freerun_entry, msg)
        return uow

    def _publish_uow(self, freerun_entry, uow):
        mq_request = SynergyMqTransmission(process_name=freerun_entry.process_name,
                                           entry_name=freerun_entry.entry_name,
                                           unit_of_work_id=uow.db_id)

        publisher = self.publishers.get(freerun_entry.process_name)
        publisher.publish(mq_request.document)
        publisher.release()

        msg = 'Published: UOW %s for %s.' % (uow.db_id, freerun_entry.schedulable_name)
        self._log_message(INFO, freerun_entry, msg)

    def insert_and_publish_uow(self, freerun_entry):
        try:
            uow = self._insert_uow(freerun_entry)
        except DuplicateKeyError as e:
            msg = 'Duplication of unit_of_work found for %s. Error msg: %r' % (freerun_entry.schedulable_name, e)
            self._log_message(WARNING, freerun_entry, msg)
            uow = self.uow_dao.recover_from_duplicatekeyerror(e)

        if uow is not None:
            # publish the created/caught up unit_of_work
            self._publish_uow(freerun_entry, uow)
            freerun_entry.related_unit_of_work = uow.db_id
            self.sfe_dao.update(freerun_entry)
        else:
            msg = 'SYSTEM IS LIKELY IN UNSTABLE STATE! Unable to locate unit_of_work for %s' \
                  % freerun_entry.schedulable_name
            self._log_message(WARNING, freerun_entry, msg)

    def manage_schedulable(self, freerun_entry):
        """ method main duty - is to _avoid_ publishing another unit_of_work, if previous was not yet processed
        In case the Scheduler sees that the unit_of_work is pending it will fire another WorkerMqRequest """

        assert isinstance(freerun_entry, FreerunProcessEntry)
        if freerun_entry.related_unit_of_work is None:
            uow = None
        else:
            uow = self.uow_dao.get_one(freerun_entry.related_unit_of_work)

        try:
            if uow is None:
                self._process_state_embryo(freerun_entry)

#.........这里部分代码省略.........