thermalctld

#!/usr/bin/env python3

"""
    thermalctld
    Thermal control daemon for SONiC
"""

import signal
import sys
import threading
import time
from datetime import datetime

import sonic_platform
from sonic_py_common import daemon_base, logger
from sonic_py_common.task_base import ProcessTaskBase
from swsscommon import swsscommon


# TODO: Once we no longer support Python 2, we can eliminate this and get the
# name using the 'name' field (e.g., `signal.SIGINT.name`) starting with Python 3.5
SIGNALS_TO_NAMES_DICT = dict((getattr(signal, n), n)
                             for n in dir(signal) if n.startswith('SIG') and '_' not in n)

SYSLOG_IDENTIFIER = 'thermalctld'
NOT_AVAILABLE = 'N/A'
CHASSIS_INFO_KEY = 'chassis 1'
PHYSICAL_ENTITY_INFO_TABLE = 'PHYSICAL_ENTITY_INFO'
INVALID_SLOT = -1

ERR_UNKNOWN = 1

# Thermal control daemon is designed to never exit, it must always
# return non-zero exit code when exiting and so that supervisord will
# restart it automatically.
exit_code = ERR_UNKNOWN

# utility functions

def try_get(callback, default=NOT_AVAILABLE):
    """
    Handy function to invoke the callback and catch NotImplementedError
    :param callback: Callback to be invoked
    :param default: Default return value if exception occur
    :return: Default return value if exception occur else return value of the callback
    """
    try:
        ret = callback()
        if ret is None:
            ret = default
    except NotImplementedError:
        ret = default

    return ret


def update_entity_info(table, parent_name, key, device, device_index):
    fvs = swsscommon.FieldValuePairs(
        [('position_in_parent', str(try_get(device.get_position_in_parent, device_index))),
         ('parent_name', parent_name)])
    table.set(key, fvs)


class FanStatus(logger.Logger):
    absent_fan_count = 0
    faulty_fan_count = 0

    def __init__(self, fan=None, is_psu_fan=False):
        """
        Initializer of FanStatus
        """
        super(FanStatus, self).__init__(SYSLOG_IDENTIFIER)

        self.fan = fan
        self.is_psu_fan = is_psu_fan
        self.presence = True
        self.status = True
        self.under_speed = False
        self.over_speed = False
        self.invalid_direction = False
        self.led_initialized = False

    @classmethod
    def get_bad_fan_count(cls):
        return cls.absent_fan_count + cls.faulty_fan_count

    @classmethod
    def reset_fan_counter(cls):
        cls.absent_fan_count = 0
        cls.faulty_fan_count = 0

    def set_presence(self, presence):
        """
        Set and cache Fan presence status
        :param presence: Fan presence status
        :return: True if status changed else False
        """
        if not presence and not self.is_psu_fan:
            FanStatus.absent_fan_count += 1

        if presence == self.presence:
            return False

        self.presence = presence
        return True

    def set_fault_status(self, status):
        """
        Set and cache Fan fault status
        :param status: Fan fault status, False indicate Fault
        :return: True if status changed else False
        """
        if not status:
            FanStatus.faulty_fan_count += 1

        if status == self.status:
            return False

        self.status = status
        return True

    def _check_speed_value_available(self, speed, target_speed, tolerance, current_status):
        if speed == NOT_AVAILABLE or target_speed == NOT_AVAILABLE or tolerance == NOT_AVAILABLE:
            if isinstance(tolerance, int) and (tolerance > 100 or tolerance < 0):
                self.log_warning('Invalid tolerance value: {}'.format(tolerance))
                return False

            if current_status is True:
                self.log_warning('Fan speed or target_speed or tolerance became unavailable, '
                                 'speed={}, target_speed={}, tolerance={}'.format(speed, target_speed, tolerance))
            return False
        return True

    def set_under_speed(self, speed, target_speed, tolerance):
        """
        Set and cache Fan under speed status
        :param speed: Fan speed
        :param target_speed: Fan target speed
        :param tolerance: Threshold between Fan speed and target speed
        :return: True if status changed else False
        """
        if not self._check_speed_value_available(speed, target_speed, tolerance, self.under_speed):
            old_status = self.under_speed
            self.under_speed = False
            return old_status != self.under_speed

        status = speed < target_speed * (1 - float(tolerance) / 100)
        if status == self.under_speed:
            return False

        self.under_speed = status
        return True

    def set_over_speed(self, speed, target_speed, tolerance):
        """
        Set and cache Fan over speed status
        :param speed: Fan speed
        :param target_speed: Fan target speed
        :param tolerance: Threshold between Fan speed and target speed
        :return: True if status changed else False
        """
        if not self._check_speed_value_available(speed, target_speed, tolerance, self.over_speed):
            old_status = self.over_speed
            self.over_speed = False
            return old_status != self.over_speed

        status = speed > target_speed * (1 + float(tolerance) / 100)
        if status == self.over_speed:
            return False

        self.over_speed = status
        return True

    def is_ok(self):
        """
        Indicate the Fan works as expect
        :return: True if Fan works normal else False
        """
        return (self.presence and
                self.status and
                not self.under_speed and
                not self.over_speed and
                not self.invalid_direction)


#
# FanUpdater ===================================================================
#
class FanUpdater(logger.Logger):
    # Fan information table name in database
    FAN_INFO_TABLE_NAME = 'FAN_INFO'
    FAN_DRAWER_INFO_TABLE_NAME = 'FAN_DRAWER_INFO'

    def __init__(self, chassis, task_stopping_event):
        """
        Initializer for FanUpdater
        :param chassis: Object representing a platform chassis
        """
        super(FanUpdater, self).__init__(SYSLOG_IDENTIFIER)

        self.chassis = chassis
        self.task_stopping_event = task_stopping_event
        self.fan_status_dict = {}
        state_db = daemon_base.db_connect("STATE_DB")
        self.table = swsscommon.Table(state_db, FanUpdater.FAN_INFO_TABLE_NAME)
        self.drawer_table = swsscommon.Table(state_db, FanUpdater.FAN_DRAWER_INFO_TABLE_NAME)
        self.phy_entity_table = swsscommon.Table(state_db, PHYSICAL_ENTITY_INFO_TABLE)

    def deinit(self):
        """
        Deinitializer of FanUpdater
        :return:
        """
        for name in self.fan_status_dict.keys():
            self.table._del(name)

    def _log_on_status_changed(self, normal_status, normal_log, abnormal_log):
        """
        Log when any status changed
        :param normal_status: Expected status.
        :param normal_log: Log string for expected status.
        :param abnormal_log: Log string for unexpected status
        :return:
        """
        if normal_status:
            self.log_notice(normal_log)
        else:
            self.log_warning(abnormal_log)

    def update(self):
        """
        Update all Fan information to database
        :return:
        """
        self.log_debug("Start fan updating")
        old_bad_fan_count = FanStatus.get_bad_fan_count()
        FanStatus.reset_fan_counter()

        for drawer_index, drawer in enumerate(self.chassis.get_all_fan_drawers()):
            if self.task_stopping_event.is_set():
                return
            self._refresh_fan_drawer_status(drawer, drawer_index)
            for fan_index, fan in enumerate(drawer.get_all_fans()):
                if self.task_stopping_event.is_set():
                    return
                try:
                    self._refresh_fan_status(drawer, drawer_index, fan, fan_index)
                except Exception as e:
                    self.log_warning('Failed to update fan status - {}'.format(repr(e)))

        for psu_index, psu in enumerate(self.chassis.get_all_psus()):
            for fan_index, fan in enumerate(psu.get_all_fans()):
                if self.task_stopping_event.is_set():
                    return
                try:
                    self._refresh_fan_status(psu, psu_index, fan, fan_index, True)
                except Exception as e:
                    self.log_warning('Failed to update PSU fan status - {}'.format(repr(e)))

        self._update_led_color()

        bad_fan_count = FanStatus.get_bad_fan_count()
        if bad_fan_count > 0 and old_bad_fan_count != bad_fan_count:
            self.log_warning("Insufficient number of working fans warning: {} fan{} not working".format(
                bad_fan_count, " is" if bad_fan_count == 1 else "s are"
            ))
        elif old_bad_fan_count > 0 and bad_fan_count == 0:
            self.log_notice("Insufficient number of working fans warning cleared: all fans are back to normal")

        self.log_debug("End fan updating")

    def _refresh_fan_drawer_status(self, fan_drawer, drawer_index):
        drawer_name = try_get(fan_drawer.get_name)
        if drawer_name == NOT_AVAILABLE:
            return

        update_entity_info(self.phy_entity_table, CHASSIS_INFO_KEY, drawer_name, fan_drawer, drawer_index)

        fvs = swsscommon.FieldValuePairs(
            [('presence', str(try_get(fan_drawer.get_presence, False))),
             ('model', str(try_get(fan_drawer.get_model))),
             ('serial', str(try_get(fan_drawer.get_serial))),
             ('status', str(try_get(fan_drawer.get_status))),
             ('is_replaceable', str(try_get(fan_drawer.is_replaceable, False))),
             ])

        self.drawer_table.set(drawer_name, fvs)

    def _refresh_fan_status(self, parent, parent_index, fan, fan_index, is_psu_fan=False):
        """
        Get Fan status by platform API and write to database for a given Fan
        :param parent: Parent device of this fan
        :param parent_index: Parent device index
        :param fan: Object representing a platform Fan
        :param fan_index: Index of the Fan object in its parent device
        :param name_prefix: name prefix of Fan object if Fan.get_name not presented
        :return:
        """
        drawer_name = NOT_AVAILABLE if is_psu_fan else str(try_get(parent.get_name))
        if is_psu_fan:
            parent_name = 'PSU {}'.format(parent_index + 1)
        else:
            parent_name = drawer_name if drawer_name != NOT_AVAILABLE else CHASSIS_INFO_KEY
        fan_name = try_get(fan.get_name, '{} fan {}'.format(parent_name, fan_index + 1))
        update_entity_info(self.phy_entity_table, parent_name, fan_name, fan, fan_index + 1)
        if fan_name not in self.fan_status_dict:
            self.fan_status_dict[fan_name] = FanStatus(fan, is_psu_fan)

        fan_status = self.fan_status_dict[fan_name]

        speed = NOT_AVAILABLE
        speed_tolerance = NOT_AVAILABLE
        speed_target = NOT_AVAILABLE
        fan_fault_status = NOT_AVAILABLE
        fan_direction = NOT_AVAILABLE
        is_replaceable = try_get(fan.is_replaceable, False)
        presence = try_get(fan.get_presence, False)
        if presence:
            speed = try_get(fan.get_speed)
            speed_tolerance = try_get(fan.get_speed_tolerance)
            speed_target = try_get(fan.get_target_speed)
            fan_fault_status = try_get(fan.get_status, False)
            fan_direction = try_get(fan.get_direction)

        set_led = not fan_status.led_initialized
        if fan_status.set_presence(presence):
            set_led = True
            self._log_on_status_changed(fan_status.presence,
                                        'Fan removed warning cleared: {} was inserted'.format(fan_name),
                                        'Fan removed warning: {} was removed from '
                                        'the system, potential overheat hazard'.format(fan_name)
                                        )

        if presence and fan_status.set_fault_status(fan_fault_status):
            set_led = True
            self._log_on_status_changed(fan_status.status,
                                        'Fan fault warning cleared: {} is back to normal'.format(fan_name),
                                        'Fan fault warning: {} is broken'.format(fan_name)
                                        )

        if presence and fan_status.set_under_speed(speed, speed_target, speed_tolerance):
            set_led = True
            self._log_on_status_changed(not fan_status.under_speed,
                                        'Fan low speed warning cleared: {} speed is back to normal'.format(fan_name),
                                        'Fan low speed warning: {} current speed={}, target speed={}, tolerance={}'.
                                        format(fan_name, speed, speed_target, speed_tolerance)
                                        )

        if presence and fan_status.set_over_speed(speed, speed_target, speed_tolerance):
            set_led = True
            self._log_on_status_changed(not fan_status.over_speed,
                                        'Fan high speed warning cleared: {} speed is back to normal'.format(fan_name),
                                        'Fan high speed warning: {} target speed={}, current speed={}, tolerance={}'.
                                        format(fan_name, speed_target, speed, speed_tolerance)
                                        )

        # TODO: handle invalid fan direction

        # We don't set PSU led here, PSU led will be handled in psud
        if set_led:
            if not is_psu_fan:
                self._set_fan_led(parent, fan, fan_name, fan_status)

        if fan_fault_status != NOT_AVAILABLE:
            fan_fault_status = fan_status.is_ok()

        fvs = swsscommon.FieldValuePairs(
            [('presence', str(presence)),
             ('drawer_name', drawer_name),
             ('model', str(try_get(fan.get_model))),
             ('serial', str(try_get(fan.get_serial))),
             ('status', str(fan_fault_status)),
             ('direction', str(fan_direction)),
             ('speed', str(speed)),
             ('speed_tolerance', str(speed_tolerance)),
             ('speed_target', str(speed_target)),
             ('is_replaceable', str(is_replaceable)),
             ('timestamp', datetime.now().strftime('%Y%m%d %H:%M:%S'))
             ])

        self.table.set(fan_name, fvs)

    def _set_fan_led(self, fan_drawer, fan, fan_name, fan_status):
        """
        Set fan led according to current status
        :param fan_drawer: Object representing a platform Fan drawer or PSU
        :param fan: Object representing a platform Fan
        :param fan_name: Name of the Fan object in case any vendor not implement Fan.get_name
        :param fan_status: Object representing the FanStatus
        :return:
        """
        try:
            led_color = fan.STATUS_LED_COLOR_GREEN if fan_status.is_ok() else fan.STATUS_LED_COLOR_RED
            fan.set_status_led(led_color)
            fan_drawer.set_status_led(led_color)
        except NotImplementedError as e:
            self.log_warning('Failed to set status LED for fan {}, set_status_led not implemented'.format(fan_name))
        
        # Set led_initialized to True even if there is NotImplementedError as it is not neccessary to 
        # print the warning log again and again. But if there is other exception, we could not
        # reach this line, and it will retry setting led color in the next run.
        fan_status.led_initialized = True

    def _update_led_color(self):
        for fan_name, fan_status in self.fan_status_dict.items():
            if self.task_stopping_event.is_set():
                return
            try:
                fvs = swsscommon.FieldValuePairs([
                    ('led_status', str(try_get(fan_status.fan.get_status_led)))
                ])
            except Exception as e:
                self.log_warning('Failed to get status LED state for fan {} - {}'.format(fan_name, e))
                fvs = swsscommon.FieldValuePairs([
                    ('led_status', NOT_AVAILABLE)
                ])
            self.table.set(fan_name, fvs)

        for drawer in self.chassis.get_all_fan_drawers():
            if self.task_stopping_event.is_set():
                return
            drawer_name = try_get(drawer.get_name)
            if drawer_name == NOT_AVAILABLE:
                continue
            try:
                fvs = swsscommon.FieldValuePairs([
                    ('led_status', str(try_get(drawer.get_status_led)))
                ])
            except Exception as e:
                self.log_warning('Failed to get status LED state for fan drawer')
                fvs = swsscommon.FieldValuePairs([
                    ('led_status', NOT_AVAILABLE)
                ])
            self.drawer_table.set(drawer_name, fvs)


class TemperatureStatus(logger.Logger):
    TEMPERATURE_DIFF_THRESHOLD = 10

    def __init__(self):
        super(TemperatureStatus, self).__init__(SYSLOG_IDENTIFIER)

        self.temperature = None
        self.over_temperature = False
        self.under_temperature = False

    def set_temperature(self, name, temperature):
        """
        Record temperature changes, if it changed too fast, raise a warning.
        :param name: Name of the thermal.
        :param temperature: New temperature value.
        :return:
        """
        if temperature == NOT_AVAILABLE:
            if self.temperature is not None:
                self.log_warning('Temperature of {} became unavailable'.format(name))
                self.temperature = None
            return

        if self.temperature is None:
            self.temperature = temperature
        else:
            diff = abs(temperature - self.temperature)
            if diff > TemperatureStatus.TEMPERATURE_DIFF_THRESHOLD:
                self.log_warning(
                    'Temperature of {} changed too fast, from {} to {}, please check your hardware'.format(
                        name, self.temperature, temperature))
            self.temperature = temperature

    def _check_temperature_value_available(self, temperature, threshold, current_status):
        if temperature == NOT_AVAILABLE or threshold == NOT_AVAILABLE:
            if current_status is True:
                self.log_warning('Thermal temperature or threshold became unavailable, '
                                 'temperature={}, threshold={}'.format(temperature, threshold))
            return False
        return True

    def set_over_temperature(self, temperature, threshold):
        """
        Set over temperature status
        :param temperature: Temperature
        :param threshold: High threshold
        :return: True if over temperature status changed else False
        """
        if not self._check_temperature_value_available(temperature, threshold, self.over_temperature):
            old_status = self.over_temperature
            self.over_temperature = False
            return old_status != self.over_temperature

        status = temperature > threshold
        if status == self.over_temperature:
            return False

        self.over_temperature = status
        return True

    def set_under_temperature(self, temperature, threshold):
        """
        Set over temperature status
        :param temperature: Temperature
        :param threshold: Low threshold
        :return: True if under temperature status changed else False
        """
        if not self._check_temperature_value_available(temperature, threshold, self.under_temperature):
            old_status = self.under_temperature
            self.under_temperature = False
            return old_status != self.under_temperature

        status = temperature < threshold
        if status == self.under_temperature:
            return False

        self.under_temperature = status
        return True


#
# TemperatureUpdater  ======================================================================
#
class TemperatureUpdater(logger.Logger):
    # Temperature information table name in database
    TEMPER_INFO_TABLE_NAME = 'TEMPERATURE_INFO'

    def __init__(self, chassis, task_stopping_event):
        """
        Initializer of TemperatureUpdater
        :param chassis: Object representing a platform chassis
        """
        super(TemperatureUpdater, self).__init__(SYSLOG_IDENTIFIER)

        self.chassis = chassis
        self.task_stopping_event = task_stopping_event
        self.temperature_status_dict = {}
        state_db = daemon_base.db_connect("STATE_DB")
        self.table = swsscommon.Table(state_db, TemperatureUpdater.TEMPER_INFO_TABLE_NAME)
        self.phy_entity_table = swsscommon.Table(state_db, PHYSICAL_ENTITY_INFO_TABLE)
        self.chassis_table = None

        self.is_chassis_system = chassis.is_modular_chassis()
        if self.is_chassis_system:
            my_slot = try_get(chassis.get_my_slot, INVALID_SLOT)
            if my_slot != INVALID_SLOT:
                table_name = TemperatureUpdater.TEMPER_INFO_TABLE_NAME+'_'+str(my_slot)
                chassis_state_db = daemon_base.db_connect("CHASSIS_STATE_DB")
                self.chassis_table = swsscommon.Table(chassis_state_db, table_name)

    def deinit(self):
        """
        Deinitializer of TemperatureUpdater
        :return:
        """
        for name in self.temperature_status_dict.keys():
            self.table._del(name)
            if self.is_chassis_system and self.chassis_table is not None:
                self.chassis_table._del(name)

    def _log_on_status_changed(self, normal_status, normal_log, abnormal_log):
        """
        Log when any status changed
        :param normal_status: Expected status.
        :param normal_log: Log string for expected status.
        :param abnormal_log: Log string for unexpected status
        :return:
        """
        if normal_status:
            self.log_notice(normal_log)
        else:
            self.log_warning(abnormal_log)

    def update(self):
        """
        Update all temperature information to database
        :return:
        """
        self.log_debug("Start temperature updating")
        for index, thermal in enumerate(self.chassis.get_all_thermals()):
            if self.task_stopping_event.is_set():
                return
            try:
                self._refresh_temperature_status(CHASSIS_INFO_KEY, thermal, index)
            except Exception as e:
                self.log_warning('Failed to update thermal status - {}'.format(repr(e)))

        for psu_index, psu in enumerate(self.chassis.get_all_psus()):
            parent_name = 'PSU {}'.format(psu_index + 1)
            for thermal_index, thermal in enumerate(psu.get_all_thermals()):
                if self.task_stopping_event.is_set():
                    return
                try:
                    self._refresh_temperature_status(parent_name, thermal, thermal_index)
                except Exception as e:
                    self.log_warning('Failed to update thermal status - {}'.format(repr(e)))

        for sfp_index, sfp in enumerate(self.chassis.get_all_sfps()):
            parent_name = 'SFP {}'.format(sfp_index + 1)
            for thermal_index, thermal in enumerate(sfp.get_all_thermals()):
                if self.task_stopping_event.is_set():
                    return
                try:
                    self._refresh_temperature_status(parent_name, thermal, thermal_index)
                except Exception as e:
                    self.log_warning('Failed to update thermal status - {}'.format(repr(e)))

        self.log_debug("End temperature updating")

    def _refresh_temperature_status(self, parent_name, thermal, thermal_index):
        """
        Get temperature status by platform API and write to database
        :param parent_name: Name of parent device of the thermal object
        :param thermal: Object representing a platform thermal zone
        :param thermal_index: Index of the thermal object in platform chassis
        :return:
        """
        name = try_get(thermal.get_name, '{} Thermal {}'.format(parent_name, thermal_index + 1))

        # Only save entity info for thermals that belong to chassis and PSU
        # for SFP thermal, they don't need save entity info because snmp can deduce the relation from TRANSCEIVER_DOM_SENSOR
        # and as we save logical port in TRANSCEIVER_INFO table, for split cable, a SFP thermal might have multiple parent
        # logical port
        if 'SFP' not in parent_name:
            update_entity_info(self.phy_entity_table, parent_name, name, thermal, thermal_index + 1)

        if name not in self.temperature_status_dict:
            self.temperature_status_dict[name] = TemperatureStatus()

        temperature_status = self.temperature_status_dict[name]

        high_threshold = NOT_AVAILABLE
        low_threshold = NOT_AVAILABLE
        high_critical_threshold = NOT_AVAILABLE
        low_critical_threshold = NOT_AVAILABLE
        maximum_temperature = NOT_AVAILABLE
        minimum_temperature = NOT_AVAILABLE
        temperature = try_get(thermal.get_temperature)
        is_replaceable = try_get(thermal.is_replaceable, False)
        if temperature != NOT_AVAILABLE:
            temperature_status.set_temperature(name, temperature)
            minimum_temperature = try_get(thermal.get_minimum_recorded)
            maximum_temperature = try_get(thermal.get_maximum_recorded)
            high_threshold = try_get(thermal.get_high_threshold)
            low_threshold = try_get(thermal.get_low_threshold)
            high_critical_threshold = try_get(thermal.get_high_critical_threshold)
            low_critical_threshold = try_get(thermal.get_low_critical_threshold)

        warning = False
        if temperature != NOT_AVAILABLE and temperature_status.set_over_temperature(temperature, high_threshold):
            self._log_on_status_changed(not temperature_status.over_temperature,
                                        'High temperature warning cleared: {} temperature restored to {}C, high threshold {}C'.
                                        format(name, temperature, high_threshold),
                                        'High temperature warning: {} current temperature {}C, high threshold {}C'.
                                        format(name, temperature, high_threshold)
                                        )
        warning = warning | temperature_status.over_temperature

        if temperature != NOT_AVAILABLE and temperature_status.set_under_temperature(temperature, low_threshold):
            self._log_on_status_changed(not temperature_status.under_temperature,
                                        'Low temperature warning cleared: {} temperature restored to {}C, low threshold {}C'.
                                        format(name, temperature, low_threshold),
                                        'Low temperature warning: {} current temperature {}C, low threshold {}C'.
                                        format(name, temperature, low_threshold)
                                        )
        warning = warning | temperature_status.under_temperature

        fvs = swsscommon.FieldValuePairs(
            [('temperature', str(temperature)),
             ('minimum_temperature', str(minimum_temperature)),
             ('maximum_temperature', str(maximum_temperature)),
             ('high_threshold', str(high_threshold)),
             ('low_threshold', str(low_threshold)),
             ('warning_status', str(warning)),
             ('critical_high_threshold', str(high_critical_threshold)),
             ('critical_low_threshold', str(low_critical_threshold)),
             ('is_replaceable', str(is_replaceable)),
             ('timestamp', datetime.now().strftime('%Y%m%d %H:%M:%S'))
             ])

        self.table.set(name, fvs)
        if self.is_chassis_system and self.chassis_table is not None:
            self.chassis_table.set(name, fvs)


class ThermalMonitor(ProcessTaskBase):
    # Initial update interval
    INITIAL_INTERVAL = 5

    # Update interval value
    UPDATE_INTERVAL = 60

    # Update elapse threshold. If update used time is larger than the value, generate a warning log.
    UPDATE_ELAPSED_THRESHOLD = 30

    def __init__(self, chassis):
        """
        Initializer for ThermalMonitor
        :param chassis: Object representing a platform chassis
        """
        super(ThermalMonitor, self).__init__()

        self.wait_time = self.INITIAL_INTERVAL

        # TODO: Refactor to eliminate the need for this Logger instance
        self.logger = logger.Logger(SYSLOG_IDENTIFIER)

        # Set minimum logging level to INFO
        self.logger.set_min_log_priority_info()

        self.fan_updater = FanUpdater(chassis, self.task_stopping_event)
        self.temperature_updater = TemperatureUpdater(chassis, self.task_stopping_event)

    def main(self):
        begin = time.time()
        self.fan_updater.update()
        self.temperature_updater.update()
        elapsed = time.time() - begin
        if elapsed < self.UPDATE_INTERVAL:
            self.wait_time = self.UPDATE_INTERVAL - elapsed
        else:
            self.wait_time = self.INITIAL_INTERVAL

        if elapsed > self.UPDATE_ELAPSED_THRESHOLD:
            self.logger.log_warning('Update fan and temperature status took {} seconds, '
                                    'there might be performance risk'.format(elapsed))

    def task_worker(self):
        """
        Thread function to handle Fan status update and temperature status update
        :return:
        """
        self.logger.log_info("Start thermal monitoring loop")

        # Start loop to update fan, temperature info in DB periodically
        while not self.task_stopping_event.wait(self.wait_time):
            self.main()

        self.fan_updater.deinit()
        self.temperature_updater.deinit()

        self.logger.log_info("Stop thermal monitoring loop")


#
# Daemon =======================================================================
#
class ThermalControlDaemon(daemon_base.DaemonBase):
    # Interval to run thermal control logic
    INTERVAL = 60
    RUN_POLICY_WARN_THRESHOLD_SECS = 30
    FAST_START_INTERVAL = 15

    POLICY_FILE = '/usr/share/sonic/platform/thermal_policy.json'

    def __init__(self):
        """
        Initializer of ThermalControlDaemon
        """
        super(ThermalControlDaemon, self).__init__(SYSLOG_IDENTIFIER)

        # Set minimum logging level to INFO
        self.set_min_log_priority_info()

        self.stop_event = threading.Event()

        self.wait_time = self.INTERVAL

        self.chassis = sonic_platform.platform.Platform().get_chassis()

        self.thermal_monitor = ThermalMonitor(self.chassis)
        self.thermal_monitor.task_run()

        self.thermal_manager = None
        try:
            self.thermal_manager = self.chassis.get_thermal_manager()
            if self.thermal_manager:
                self.thermal_manager.initialize()
                self.thermal_manager.load(ThermalControlDaemon.POLICY_FILE)
                self.thermal_manager.init_thermal_algorithm(self.chassis)
        except NotImplementedError:
            self.log_warning('Thermal manager is not supported on this platform')
        except Exception as e:
            self.log_error('Caught exception while initializing thermal manager - {}'.format(repr(e)))

    def deinit(self):
        """
        Deinitializer of ThermalControlDaemon
        """
        try:
            if self.thermal_manager:
                self.thermal_manager.deinitialize()
        except Exception as e:
            self.log_error('Caught exception while destroying thermal manager - {}'.format(repr(e)))

        self.thermal_monitor.task_stop()

    # Override signal handler from DaemonBase
    def signal_handler(self, sig, frame):
        """
        Signal handler
        :param sig: Signal number
        :param frame: not used
        :return:
        """
        FATAL_SIGNALS = [signal.SIGINT, signal.SIGTERM]
        NONFATAL_SIGNALS = [signal.SIGHUP]

        global exit_code

        if sig in FATAL_SIGNALS:
            self.log_info("Caught signal '{}' - exiting...".format(SIGNALS_TO_NAMES_DICT[sig]))
            exit_code = 128 + sig  # Make sure we exit with a non-zero code so that supervisor will try to restart us
            self.thermal_monitor.task_stop()
            self.thermal_manager.stop()
            self.stop_event.set()
        elif sig in NONFATAL_SIGNALS:
            self.log_info("Caught signal '{}' - ignoring...".format(SIGNALS_TO_NAMES_DICT[sig]))
        else:
            self.log_warning("Caught unhandled signal '{}' - ignoring...".format(SIGNALS_TO_NAMES_DICT[sig]))

    # Main daemon logic
    def run(self):
        """
        Run main logical of this daemon
        :return:
        """
        if self.stop_event.wait(self.wait_time):
            # We received a fatal signal
            return False

        begin = time.time()
        try:
            if self.thermal_manager:
                self.thermal_manager.run_policy(self.chassis)
        except Exception as e:
            self.log_error('Caught exception while running thermal policy - {}'.format(repr(e)))

        elapsed = time.time() - begin
        if elapsed < self.INTERVAL:
            self.wait_time = self.INTERVAL - elapsed
        else:
            self.wait_time = self.FAST_START_INTERVAL

        if elapsed > self.RUN_POLICY_WARN_THRESHOLD_SECS:
            self.log_warning('Thermal policy execution took {} seconds, '
                             'there might be performance risk'.format(elapsed))

        return True


#
# Main =========================================================================
#
def main():
    thermal_control = ThermalControlDaemon()

    thermal_control.log_info("Starting up...")

    while thermal_control.run():
        pass

    thermal_control.log_info("Shutting down with exit code {}...".format(exit_code))

    thermal_control.deinit()

    return exit_code


if __name__ == '__main__':
    sys.exit(main())