def RunOnce(client, job, callback):
"""Get list of files and call processing function."""
logs_paths = logs_util.ServerLogsPaths('viewfinder', 'full')
merged_store = ObjectStore.GetInstance(logs_paths.MERGED_LOGS_BUCKET)
start_date = options.options.start_date
if options.options.smart_scan:
# Search for successful full-scan run in the last week.
last_run = yield gen.Task(job.FindLastSuccess, with_payload_key='stats.last_day')
if last_run is None:
logging.info('No previous successful scan found, rerun with --start_date')
callback(None)
return
last_run_start = last_run['start_time']
if util.HoursSince(last_run_start) < options.options.hours_between_runs:
logging.info('Last successful run started at %s, less than %d hours ago; skipping.' %
(time.asctime(time.localtime(last_run_start)), options.options.hours_between_runs))
callback(None)
return
last_day = last_run['stats.last_day']
# Set scan_start to start of previous run - 1d. The extra 1d is in case some logs were pushed to S3 late.
# This really recomputes two days (the last day that was successfully processed and the one prior).
start_time = util.ISO8601ToUTCTimestamp(last_day, hour=12) - constants.SECONDS_PER_DAY
start_date = util.TimestampUTCToISO8601(start_time)
logging.info('Last successful analyze_logs run (%s) scanned up to %s, setting analysis start date to %s' %
(time.asctime(time.localtime(last_run_start)), last_day, start_date))
# Fetch list of merged logs.
files = yield gen.Task(GetMergedLogsFileList, merged_store, logs_paths, start_date)
last_day = yield gen.Task(ProcessFiles, merged_store, logs_paths, files)
callback(last_day)
return
def _GetMetric(self, day, h=12, m=0, s=0):
timestamp = util.ISO8601ToUTCTimestamp(day, hour=h, minute=m, second=s)
existing_metrics = self._RunAsync(metric.Metric.QueryTimespan,
self._client, self._group_key,
timestamp, timestamp)
if len(existing_metrics) == 0:
return None
return DotDict(json.loads(existing_metrics[0].payload))
def DayTimeStringsToUTCTimestamp(day, time):
"""Given day (YYYY-MM-DD) and time (HH:MM:SS:ms) strings, return the timestamp in UTC, or None if parsing failed."""
try:
hour, minute, second, _ = re.match(kTimeRe, time).groups()
return util.ISO8601ToUTCTimestamp(day, hour=int(hour), minute=int(minute), second=int(second))
except Exception as e:
logging.warning('Error parsing day and time strings: %s %s, error: %r' % (day, time, e))
return None
def RunOnce(client, job, apple_id, password, callback):
start_date = yield gen.Task(DetermineStartDate, client, job)
if start_date is None:
logging.info(
'Start date not specified, last run too recent, or smart_scan could not determine a date; exiting.'
)
callback(None)
return
query_dates = []
start_time = util.ISO8601ToUTCTimestamp(start_date)
today = util.NowUTCToISO8601()
while start_time < time.time():
date = util.TimestampUTCToISO8601(start_time)
if date != today:
query_dates.append(date)
start_time += constants.SECONDS_PER_DAY
logging.info('fetching data for dates: %r' % query_dates)
results = {}
itc = ITunesTrends(apple_id, password, options.options.vendor_id)
failed = False
for day in query_dates:
if failed:
break
try:
result = yield gen.Task(itc.FetchOneDay, day)
if not result:
# We don't get an exception when iTunesConnect has no data. We also don't want to
# fail as there's no way it will have this data later.
logging.warning('No data for day %s' % day)
else:
results[day] = result
except Exception:
msg = traceback.format_exc()
logging.warning(
'Error fetching iTunes Connect data for day %s: %s', day, msg)
failed = True
if len(results) == 0:
callback(None)
else:
# Replace the entire 'itunes' category of previous metrics. This is so we can fix any processing errors we
# may have had.
hms = logs_util.kDailyMetricsTimeByLogType['itunes_trends']
yield gen.Task(logs_util.UpdateMetrics,
client,
results,
prefix_to_erase='itunes',
dry_run=options.options.dry_run,
hms_tuple=hms)
callback(sorted(results.keys())[-1])
def RunOnce(client, job, callback):
"""Get list of files and call processing function."""
merged_store = ObjectStore.GetInstance(logs_util.UserAnalyticsLogsPaths.MERGED_LOGS_BUCKET)
start_date = options.options.start_date
if options.options.smart_scan:
# Search for successful full-scan run in the last week.
last_run = yield gen.Task(job.FindLastSuccess, with_payload_key='stats.last_day')
if last_run is None:
logging.info('No previous successful scan found, rerun with --start_date')
callback(None)
return
last_run_start = last_run['start_time']
if util.HoursSince(last_run_start) < options.options.hours_between_runs:
logging.info('Last successful run started at %s, less than %d hours ago; skipping.' %
(time.asctime(time.localtime(last_run_start)), options.options.hours_between_runs))
callback(None)
return
last_day = last_run['stats.last_day']
# Set scan_start to start of previous run - 30d (we need 30 days' worth of data to properly compute
# 30-day active users. Add an extra 3 days just in case we had some missing logs during the last run.
start_time = util.ISO8601ToUTCTimestamp(last_day, hour=12) - constants.SECONDS_PER_WEEK
start_date = util.TimestampUTCToISO8601(start_time)
logging.info('Last successful analyze_analytics run (%s) scanned up to %s, setting analysis start date to %s' %
(time.asctime(time.localtime(last_run_start)), last_day, start_date))
# Fetch list of merged logs.
files = yield gen.Task(GetMergedLogsFileList, merged_store, start_date)
day_stats = yield gen.Task(ProcessFiles, merged_store, files)
# Write per-day stats to dynamodb.
if len(day_stats) > 0:
hms = logs_util.kDailyMetricsTimeByLogType['analytics_logs']
yield gen.Task(logs_util.UpdateMetrics, client, day_stats, dry_run=options.options.dry_run, hms_tuple=hms)
last_day = sorted(day_stats.keys())[-1]
callback(last_day)
else:
callback(None)
def DetermineStartDate(client, job, callback):
"""If smart_scan is true, lookup the start date from previous job summaries, otherwise use --start_date.
--start_date and job summary days are of the form YYYY-MM-DD.
"""
start_date = options.options.start_date
# Lookup previous runs started in the last week.
if options.options.smart_scan:
# Search for successful full-scan run in the last week.
last_run = yield gen.Task(job.FindLastSuccess,
with_payload_key='stats.last_day')
if last_run is None:
logging.info(
'No previous successful scan found, rerun with --start_date')
callback(None)
return
last_run_start = last_run['start_time']
if (last_run_start +
options.options.hours_between_runs * constants.SECONDS_PER_HOUR
> time.time()):
logging.info(
'Last successful run started at %s, less than %d hours ago; skipping.'
% (time.asctime(time.localtime(last_run_start)),
options.options.hours_between_runs))
callback(None)
return
# Start start_date to the last processed day + 1.
last_day = last_run['stats.last_day']
start_time = util.ISO8601ToUTCTimestamp(
last_day) + constants.SECONDS_PER_DAY
start_date = util.TimestampUTCToISO8601(start_time)
logging.info(
'Last successful run (%s) scanned up to %s, setting start date to %s'
% (time.asctime(
time.localtime(last_run_start)), last_day, start_date))
callback(start_date)
def _WriteMetric(self, day, dotdict):
new_metric = metric.Metric.Create(
self._group_key, 'logs_daily',
util.ISO8601ToUTCTimestamp(day, hour=12), json.dumps(dotdict))
self._RunAsync(new_metric.Update, self._client)
def UpdateMetrics(db_client, day_stats, callback, dry_run=True, prefix_to_erase=None, hms_tuple=None):
"""Write 'day_stats' to the metrics table. First lookup any existing metrics and update them.
'day_stats' is a dictionary of {day_in_iso8601: DotDict}.
If 'dry_run' is True, don't commit the changes to the metrics table, but perform all the work and log to info.
If 'prefix_to_erase' is not None, we first replace the passed-in prefix with an empty dotdict.
If 'hms_tuple' is not None, the timestamp for the metric entry will be with the specified hour/minute/second,
otherwise, we use noon. To help with consistency, hms_tuple should come from kDailyMetricsTimeByLogType above.
For example, given the existing metric: { itunes: { downloads: { 'US': 5, 'UK': 3 }, update: { ... }}}
We can either:
- Replace the downloads numbers: (the entire tree under 'prefix_to_erase' gets replaced)
UpdateMetrics({'2013-02-01': {'itunes': {'downloads': { 'DE': 3, 'FR': 1 }}}}, prefix_to_erase='itunes.downloads')
resulting in: { itunes: { downloads: { 'DE': 3, 'FR': 1 }, update: { ... }}}
- Or we can update with partial stats:
UpdateMetrics({'2013-02-01': {'itunes': { 'downloads': { 'DE': 3, 'FR': 1 }}}}, replace=False)
resulting in: { itunes: { downloads: { 'US': 5, 'UK': 3, 'DE': 3, 'FR': 3 }, update: { ... }}}
"""
if len(day_stats) == 0:
callback()
return
cluster = metric.LOGS_STATS_NAME
group_key = metric.Metric.EncodeGroupKey(cluster, metric.Metric.FindIntervalForCluster(cluster, 'daily'))
# Convert YYYY-MM-DD into the timestamp for noon UTC.
h, m, s = hms_tuple if hms_tuple is not None else (12, 0, 0)
timestamps = [(util.ISO8601ToUTCTimestamp(day, hour=h, minute=m, second=s), day) for day in sorted(day_stats.keys())]
# Query Metrics table for all metrics between the timestamps we have data for.
existing_metrics = yield gen.Task(metric.Metric.QueryTimespan, db_client, group_key,
timestamps[0][0], timestamps[-1][0])
existing_dict = dict((m.timestamp, m) for m in existing_metrics)
tasks = []
for t, day in timestamps:
data = day_stats[day]
prev_metric = existing_dict.get(t, None)
payload = json.dumps(data)
if prev_metric is None:
logging.info('%s: new metric: %r' % (day, payload))
else:
prev_payload = prev_metric.payload
# We do this twice, it's simpler than making deepcopy work on DotDict.
prev_data = DotDict(json.loads(prev_payload))
new_data = DotDict(json.loads(prev_payload))
if prefix_to_erase is not None:
# We can't call 'del' on a DotDict's internals, so simply replace with an empty dotdict, we'll be repopulating.
new_data[prefix_to_erase] = DotDict()
# DotDict doesn't have an update() method.
flat = new_data.flatten()
flat.update(data.flatten())
new_data = DotDict(flat)
payload = json.dumps(new_data)
if new_data.flatten() == prev_data.flatten():
logging.info('%s: metric has not changed, skipping' % day)
continue
else:
logging.info('%s: changed metric: %s -> %s' % (day, prev_payload, payload))
if not dry_run:
new_metric = metric.Metric.Create(group_key, 'logs_daily', t, payload)
tasks.append(gen.Task(new_metric.Update, db_client))
yield tasks
callback()
def RunOnce(client, job, callback):
"""Get list of files and call processing function."""
logs_paths = logs_util.ServerLogsPaths('viewfinder', 'full')
merged_store = ObjectStore.GetInstance(logs_paths.MERGED_LOGS_BUCKET)
start_date = options.options.start_date
if options.options.smart_scan:
# Search for successful full-scan run in the last week.
last_run = yield gen.Task(job.FindLastSuccess,
with_payload_key='stats.last_day')
if last_run is None:
logging.info(
'No previous successful scan found, rerun with --start_date')
callback(None)
return
last_run_start = last_run['start_time']
if util.HoursSince(
last_run_start) < options.options.hours_between_runs:
logging.info(
'Last successful run started at %s, less than %d hours ago; skipping.'
% (time.asctime(time.localtime(last_run_start)),
options.options.hours_between_runs))
callback(None)
return
# Compute stats for all days since the last run (inclusive).
start_date = last_run['stats.last_day']
logging.info(
'Last successful server_log_metrics run (%s) scanned up to %s' %
(time.asctime(time.localtime(last_run_start)), start_date))
assert start_date is not None, 'No start date provided. Use --start_date=YYYY-MM-DD or --smart_scan=True'
last_day = None
if options.options.compute_user_requests:
# We look for files going back 32 days (30 because we need 30-day active users, and +2 for extra safety).
user_start_time = util.ISO8601ToUTCTimestamp(
start_date, hour=12) - 32 * constants.SECONDS_PER_DAY
user_start_date = util.TimestampUTCToISO8601(user_start_time)
# Fetch list of per-day user request stats.
files = yield gen.Task(GetFileList, merged_store, 'user_requests',
user_start_date)
user_request_stats = yield gen.Task(ComputeUserRequests, merged_store,
files)
# Write per-day stats to dynamodb.
if len(user_request_stats) > 0:
# We do not replace the 'active_users' category in previous metrics as we may not be recomputing all data
# eg: we need to analyze 30 days of logs to get the full metric info.
hms = logs_util.kDailyMetricsTimeByLogType['active_users']
yield gen.Task(logs_util.UpdateMetrics,
client,
user_request_stats,
dry_run=options.options.dry_run,
hms_tuple=hms)
last_day = sorted(user_request_stats.keys())[-1]
if options.options.compute_registration_delay:
# We compute stats for days within the registration window (plus an extra two for safety).
device_start_time = util.ISO8601ToUTCTimestamp(start_date, hour=12) - \
(kDeviceRegistrationWindowDays + 2) * constants.SECONDS_PER_DAY
device_start_date = util.TimestampUTCToISO8601(device_start_time)
# However, we search for files for an extra 15 days. This is so that unregistered devices that ping every day
# are not counted as starting on the first day of stats.
device_search_time = device_start_time - 15 * constants.SECONDS_PER_DAY
device_search_date = util.TimestampUTCToISO8601(device_search_time)
# Fetch list of merged logs.
files = yield gen.Task(GetFileList, merged_store, 'device_details',
device_search_date)
device_stats = yield gen.Task(ComputeRegistrationDelay, merged_store,
files, device_start_date)
# Write per-day stats to dynamodb.
if len(device_stats) > 0:
# Replace the entire 'device_installs.registration' category
hms = logs_util.kDailyMetricsTimeByLogType['device_installs']
yield gen.Task(logs_util.UpdateMetrics,
client,
device_stats,
dry_run=options.options.dry_run,
hms_tuple=hms,
prefix_to_erase='device_installs.registration')
last_day_device = sorted(device_stats.keys())[-1]
if last_day is None or last_day > last_day_device:
# We consider the last successful processing day to be the earlier of the two.
last_day = last_day_device
if options.options.compute_app_versions:
# Look at an extra two days for safety.
version_start_time = util.ISO8601ToUTCTimestamp(
start_date, hour=12) - 2 * constants.SECONDS_PER_DAY
version_start_date = util.TimestampUTCToISO8601(version_start_time)
# Fetch list of merged logs.
files = yield gen.Task(GetFileList, merged_store, 'device_details',
version_start_date)
version_stats = yield gen.Task(ComputeAppVersions, merged_store, files)
# Write per-day stats to dynamodb.
if len(version_stats) > 0:
# Replace the entire 'device_installs.registration' category
hms = logs_util.kDailyMetricsTimeByLogType['device_count']
yield gen.Task(logs_util.UpdateMetrics,
client,
version_stats,
dry_run=options.options.dry_run,
hms_tuple=hms,
prefix_to_erase='device_count')
last_day_device = sorted(version_stats.keys())[-1]
if last_day is None or last_day > last_day_device:
# We consider the last successful processing day to be the earlier of the two.
last_day = last_day_device
if options.options.compute_traces or options.options.compute_aborts:
# Fetch an extra 7 days to filter out previously-seen traces.
trace_start_time = (
util.ISO8601ToUTCTimestamp(start_date, hour=12) -
options.options.extra_trace_days * constants.SECONDS_PER_DAY)
trace_start_date = util.TimestampUTCToISO8601(trace_start_time)
# Fetch list of merged logs.
files = yield gen.Task(GetFileList, merged_store, 'traces',
trace_start_date)
# Run them separately, we'll be sending separate email reports.
if options.options.compute_traces:
yield gen.Task(ComputeTraces, merged_store, files, start_date)
if options.options.compute_aborts:
yield gen.Task(ComputeAborts, merged_store, files, start_date)
callback(last_day)
