def process_count_stat(stat: CountStat, fill_to_time: datetime,
realm: Optional[Realm]=None) -> None:
# TODO: The realm argument is not yet supported, in that we don't
# have a solution for how to update FillState if it is passed. It
# exists solely as partial plumbing for when we do fully implement
# doing single-realm analytics runs for use cases like data import.
#
# Also, note that for the realm argument to be properly supported,
# the CountStat object passed in needs to have come from
# E.g. get_count_stats(realm), i.e. have the realm_id already
# entered into the SQL query defined by the CountState object.
if stat.frequency == CountStat.HOUR:
time_increment = timedelta(hours=1)
elif stat.frequency == CountStat.DAY:
time_increment = timedelta(days=1)
else:
raise AssertionError("Unknown frequency: %s" % (stat.frequency,))
verify_UTC(fill_to_time)
if floor_to_hour(fill_to_time) != fill_to_time:
raise ValueError("fill_to_time must be on an hour boundary: %s" % (fill_to_time,))
fill_state = FillState.objects.filter(property=stat.property).first()
if fill_state is None:
currently_filled = installation_epoch()
fill_state = FillState.objects.create(property=stat.property,
end_time=currently_filled,
state=FillState.DONE)
logger.info("INITIALIZED %s %s", stat.property, currently_filled)
elif fill_state.state == FillState.STARTED:
logger.info("UNDO START %s %s", stat.property, fill_state.end_time)
do_delete_counts_at_hour(stat, fill_state.end_time)
currently_filled = fill_state.end_time - time_increment
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
logger.info("UNDO DONE %s", stat.property)
elif fill_state.state == FillState.DONE:
currently_filled = fill_state.end_time
else:
raise AssertionError("Unknown value for FillState.state: %s." % (fill_state.state,))
if isinstance(stat, DependentCountStat):
for dependency in stat.dependencies:
dependency_fill_time = last_successful_fill(dependency)
if dependency_fill_time is None:
logger.warning("DependentCountStat %s run before dependency %s.",
stat.property, dependency)
return
fill_to_time = min(fill_to_time, dependency_fill_time)
currently_filled = currently_filled + time_increment
while currently_filled <= fill_to_time:
logger.info("START %s %s", stat.property, currently_filled)
start = time.time()
do_update_fill_state(fill_state, currently_filled, FillState.STARTED)
do_fill_count_stat_at_hour(stat, currently_filled, realm)
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
end = time.time()
currently_filled = currently_filled + time_increment
logger.info("DONE %s (%dms)", stat.property, (end-start)*1000)
def process_count_stat(stat, fill_to_time):
# type: (CountStat, datetime) -> None
if stat.frequency == CountStat.HOUR:
time_increment = timedelta(hours=1)
elif stat.frequency == CountStat.DAY:
time_increment = timedelta(days=1)
else:
raise AssertionError("Unknown frequency: %s" % (stat.frequency, ))
verify_UTC(fill_to_time)
if floor_to_hour(fill_to_time) != fill_to_time:
raise ValueError("fill_to_time must be on an hour boundary: %s" %
(fill_to_time, ))
fill_state = FillState.objects.filter(property=stat.property).first()
if fill_state is None:
currently_filled = installation_epoch()
fill_state = FillState.objects.create(property=stat.property,
end_time=currently_filled,
state=FillState.DONE)
logger.info("INITIALIZED %s %s" % (stat.property, currently_filled))
elif fill_state.state == FillState.STARTED:
logger.info("UNDO START %s %s" % (stat.property, fill_state.end_time))
do_delete_counts_at_hour(stat, fill_state.end_time)
currently_filled = fill_state.end_time - time_increment
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
logger.info("UNDO DONE %s" % (stat.property, ))
elif fill_state.state == FillState.DONE:
currently_filled = fill_state.end_time
else:
raise AssertionError("Unknown value for FillState.state: %s." %
(fill_state.state, ))
if isinstance(stat, DependentCountStat):
for dependency in stat.dependencies:
dependency_fill_time = last_successful_fill(dependency)
if dependency_fill_time is None:
logger.warning(
"DependentCountStat %s run before dependency %s." %
(stat.property, dependency))
return
fill_to_time = min(fill_to_time, dependency_fill_time)
currently_filled = currently_filled + time_increment
while currently_filled <= fill_to_time:
logger.info("START %s %s" % (stat.property, currently_filled))
start = time.time()
do_update_fill_state(fill_state, currently_filled, FillState.STARTED)
do_fill_count_stat_at_hour(stat, currently_filled)
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
end = time.time()
currently_filled = currently_filled + time_increment
logger.info("DONE %s (%dms)" % (stat.property, (end - start) * 1000))
def get_fill_state(self) -> Dict[str, Any]:
if not Realm.objects.exists():
return {"status": 0, "message": "No realms exist, so not checking FillState."}
warning_unfilled_properties = []
critical_unfilled_properties = []
for property, stat in COUNT_STATS.items():
last_fill = stat.last_successful_fill()
if last_fill is None:
last_fill = installation_epoch()
try:
verify_UTC(last_fill)
except TimeZoneNotUTCException:
return {"status": 2, "message": f"FillState not in UTC for {property}"}
if stat.frequency == CountStat.DAY:
floor_function = floor_to_day
warning_threshold = timedelta(hours=26)
critical_threshold = timedelta(hours=50)
else: # CountStat.HOUR
floor_function = floor_to_hour
warning_threshold = timedelta(minutes=90)
critical_threshold = timedelta(minutes=150)
if floor_function(last_fill) != last_fill:
return {
"status": 2,
"message": f"FillState not on {stat.frequency} boundary for {property}",
}
time_to_last_fill = timezone_now() - last_fill
if time_to_last_fill > critical_threshold:
critical_unfilled_properties.append(property)
elif time_to_last_fill > warning_threshold:
warning_unfilled_properties.append(property)
if len(critical_unfilled_properties) == 0 and len(warning_unfilled_properties) == 0:
return {"status": 0, "message": "FillState looks fine."}
if len(critical_unfilled_properties) == 0:
return {
"status": 1,
"message": "Missed filling {} once.".format(
", ".join(warning_unfilled_properties),
),
}
return {
"status": 2,
"message": "Missed filling {} once. Missed filling {} at least twice.".format(
", ".join(warning_unfilled_properties),
", ".join(critical_unfilled_properties),
),
}
def process_count_stat(stat, fill_to_time):
# type: (CountStat, datetime) -> None
if stat.frequency == CountStat.HOUR:
time_increment = timedelta(hours=1)
elif stat.frequency == CountStat.DAY:
time_increment = timedelta(days=1)
else:
raise AssertionError("Unknown frequency: %s" % (stat.frequency,))
verify_UTC(fill_to_time)
if floor_to_hour(fill_to_time) != fill_to_time:
raise ValueError("fill_to_time must be on an hour boundary: %s" % (fill_to_time,))
fill_state = FillState.objects.filter(property=stat.property).first()
if fill_state is None:
currently_filled = installation_epoch()
fill_state = FillState.objects.create(property=stat.property,
end_time=currently_filled,
state=FillState.DONE)
logger.info("INITIALIZED %s %s" % (stat.property, currently_filled))
elif fill_state.state == FillState.STARTED:
logger.info("UNDO START %s %s" % (stat.property, fill_state.end_time))
do_delete_counts_at_hour(stat, fill_state.end_time)
currently_filled = fill_state.end_time - time_increment
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
logger.info("UNDO DONE %s" % (stat.property,))
elif fill_state.state == FillState.DONE:
currently_filled = fill_state.end_time
else:
raise AssertionError("Unknown value for FillState.state: %s." % (fill_state.state,))
if isinstance(stat, DependentCountStat):
for dependency in stat.dependencies:
dependency_fill_time = last_successful_fill(dependency)
if dependency_fill_time is None:
logger.warning("DependentCountStat %s run before dependency %s." %
(stat.property, dependency))
return
fill_to_time = min(fill_to_time, dependency_fill_time)
currently_filled = currently_filled + time_increment
while currently_filled <= fill_to_time:
logger.info("START %s %s" % (stat.property, currently_filled))
start = time.time()
do_update_fill_state(fill_state, currently_filled, FillState.STARTED)
do_fill_count_stat_at_hour(stat, currently_filled)
do_update_fill_state(fill_state, currently_filled, FillState.DONE)
end = time.time()
currently_filled = currently_filled + time_increment
logger.info("DONE %s (%dms)" % (stat.property, (end-start)*1000))
def get_fill_state(self):
# type: () -> Dict[str, Any]
if not Realm.objects.exists():
return {'status': 0, 'message': 'No realms exist, so not checking FillState.'}
warning_unfilled_properties = []
critical_unfilled_properties = []
for property, stat in COUNT_STATS.items():
last_fill = last_successful_fill(property)
if last_fill is None:
last_fill = installation_epoch()
try:
verify_UTC(last_fill)
except TimezoneNotUTCException:
return {'status': 2, 'message': 'FillState not in UTC for %s' % (property,)}
if stat.frequency == CountStat.DAY:
floor_function = floor_to_day
warning_threshold = timedelta(hours=26)
critical_threshold = timedelta(hours=50)
else: # CountStat.HOUR
floor_function = floor_to_hour
warning_threshold = timedelta(minutes=90)
critical_threshold = timedelta(minutes=150)
if floor_function(last_fill) != last_fill:
return {'status': 2, 'message': 'FillState not on %s boundary for %s' %
(stat.frequency, property)}
time_to_last_fill = timezone_now() - last_fill
if time_to_last_fill > critical_threshold:
critical_unfilled_properties.append(property)
elif time_to_last_fill > warning_threshold:
warning_unfilled_properties.append(property)
if len(critical_unfilled_properties) == 0 and len(warning_unfilled_properties) == 0:
return {'status': 0, 'message': 'FillState looks fine.'}
if len(critical_unfilled_properties) == 0:
return {'status': 1, 'message': 'Missed filling %s once.' %
(', '.join(warning_unfilled_properties),)}
return {'status': 2, 'message': 'Missed filling %s once. Missed filling %s at least twice.' %
(', '.join(warning_unfilled_properties), ', '.join(critical_unfilled_properties))}
def get_fill_state(self) -> Dict[str, Any]:
if not Realm.objects.exists():
return {'status': 0, 'message': 'No realms exist, so not checking FillState.'}
warning_unfilled_properties = []
critical_unfilled_properties = []
for property, stat in COUNT_STATS.items():
last_fill = last_successful_fill(property)
if last_fill is None:
last_fill = installation_epoch()
try:
verify_UTC(last_fill)
except TimezoneNotUTCException:
return {'status': 2, 'message': 'FillState not in UTC for %s' % (property,)}
if stat.frequency == CountStat.DAY:
floor_function = floor_to_day
warning_threshold = timedelta(hours=26)
critical_threshold = timedelta(hours=50)
else: # CountStat.HOUR
floor_function = floor_to_hour
warning_threshold = timedelta(minutes=90)
critical_threshold = timedelta(minutes=150)
if floor_function(last_fill) != last_fill:
return {'status': 2, 'message': 'FillState not on %s boundary for %s' %
(stat.frequency, property)}
time_to_last_fill = timezone_now() - last_fill
if time_to_last_fill > critical_threshold:
critical_unfilled_properties.append(property)
elif time_to_last_fill > warning_threshold:
warning_unfilled_properties.append(property)
if len(critical_unfilled_properties) == 0 and len(warning_unfilled_properties) == 0:
return {'status': 0, 'message': 'FillState looks fine.'}
if len(critical_unfilled_properties) == 0:
return {'status': 1, 'message': 'Missed filling %s once.' %
(', '.join(warning_unfilled_properties),)}
return {'status': 2, 'message': 'Missed filling %s once. Missed filling %s at least twice.' %
(', '.join(warning_unfilled_properties), ', '.join(critical_unfilled_properties))}
def time_range(start, end, frequency, min_length):
# type: (datetime, datetime, str, Optional[int]) -> List[datetime]
verify_UTC(start)
verify_UTC(end)
if frequency == CountStat.HOUR:
end = floor_to_hour(end)
step = timedelta(hours=1)
elif frequency == CountStat.DAY:
end = floor_to_day(end)
step = timedelta(days=1)
else:
raise AssertionError("Unknown frequency: %s" % (frequency, ))
times = []
if min_length is not None:
start = min(start, end - (min_length - 1) * step)
current = end
while current >= start:
times.append(current)
current -= step
return list(reversed(times))
def time_range(start, end, frequency, min_length):
# type: (datetime, datetime, str, Optional[int]) -> List[datetime]
verify_UTC(start)
verify_UTC(end)
if frequency == CountStat.HOUR:
end = floor_to_hour(end)
step = timedelta(hours=1)
elif frequency == CountStat.DAY:
end = floor_to_day(end)
step = timedelta(days=1)
else:
raise AssertionError("Unknown frequency: %s" % (frequency,))
times = []
if min_length is not None:
start = min(start, end - (min_length-1)*step)
current = end
while current >= start:
times.append(current)
current -= step
return list(reversed(times))
