def get(self, viewLength): """Handles a new error report via POST.""" if viewLength != 'day': # TODO(robbyw): For viewLength == week or viewLength == month, aggregate the aggregates. viewLength = 'day' data = AggregatedStats.all().order('-date').get() data = json.loads(data.json)[:25] for _, row in data: logging.info(row) row['servers'] = sorted(row['servers'].items(), key=lambda x: x[1], reverse=True) row['environments'] = sorted(row['environments'].items(), key=lambda x: x[1], reverse=True) keys, values = zip(*data) errors = LoggedError.get([db.Key(key) for key in keys]) context = { 'title': 'Top 25 exceptions over the last %s' % viewLength, 'errors': zip(errors, values), 'total': len(data) } self.response.out.write( template.render(getTemplatePath('aggregation.html'), context))
def doAuthenticatedGet(self, _, *args): key, = args self.response.headers['Content-Type'] = 'text/plain' error = LoggedError.get(key) error.active = False error.put() self.response.out.write('ok')
def doAuthenticatedGet(self, user, *args): key, = args self.response.headers['Content-Type'] = 'text/html' error = LoggedError.get(key) filters = getFilters(self.request) context = { 'title': '%s - %s' % (error.lastMessage, NAME), 'extraScripts': ['view'], 'user': user, 'error': error, 'filters': filters.items(), 'instances': getInstances(filters, parent=error)[:100] } self.response.out.write(template.render(getTemplatePath('view.html'), context))
def doAuthenticatedGet(self, user, *args): key, = args self.response.headers['Content-Type'] = 'text/html' error = LoggedError.get(key) filters = getFilters(self.request) context = { 'title': '%s - %s' % (error.lastMessage, NAME), 'extraScripts': ['view'], 'user': user, 'error': error, 'filters': filters.items(), 'instances': getInstances(filters, parent=error)[:100] } self.response.out.write( template.render(getTemplatePath('view.html'), context))
def get(self, viewLength): """Handles a new error report via POST.""" if viewLength != 'day': # TODO(robbyw): For viewLength == week or viewLength == month, aggregate the aggregates. viewLength = 'day' data = AggregatedStats.all().order('-date').get() data = json.loads(data.json)[:25] for _, row in data: logging.info(row) row['servers'] = sorted(row['servers'].items(), key = lambda x: x[1], reverse=True) row['environments'] = sorted(row['environments'].items(), key = lambda x: x[1], reverse=True) keys, values = zip(*data) errors = LoggedError.get([db.Key(key) for key in keys]) context = { 'title': 'Top 25 exceptions over the last %s' % viewLength, 'errors': zip(errors, values), 'total': len(data) } self.response.out.write(template.render(getTemplatePath('aggregation.html'), context))
def post(self): # pylint: disable=R0914, R0915 """Handles a new error report via POST.""" taskId = self.request.get('id', '0') currentId = memcache.get(AGGREGATION_ID) if taskId == 'None' or not (taskId == currentId or int(taskId) % 50 == 0): # Skip this task unless it is the most recently added or if it is one of every fifty tasks. logging.debug('Skipping task %s, current is %s', taskId, currentId) return q = taskqueue.Queue('aggregation') tasks = _getTasks(q) logging.info('Leased %d tasks', len(tasks)) byError = collections.defaultdict(list) instanceKeys = [] tasksByError = collections.defaultdict(list) for task in tasks: data = json.loads(task.payload) errorKey = data['error'] if 'instance' in data and 'backtrace' in data: instanceKey = data['instance'] byError[errorKey].append((instanceKey, data['backtrace'])) instanceKeys.append(instanceKey) tasksByError[errorKey].append(task) elif 'aggregation' in data: byError[errorKey].append(data['aggregation']) tasksByError[errorKey].append(task) else: # Clean up any old tasks in the queue. logging.warn('Deleting an old task') q.delete_tasks([task]) retries = 0 instanceByKey = getInstanceMap(instanceKeys) for errorKey, instances in byError.items(): instances = [keyOrDict if isinstance(keyOrDict, dict) else aggregateSingleInstance(instanceByKey[keyOrDict[0]], keyOrDict[1]) for keyOrDict in instances] aggregation = aggregateInstances(instances) success = False if _lockError(errorKey): try: error = LoggedError.get(errorKey) aggregate( error, aggregation.count, aggregation.firstOccurrence, aggregation.lastOccurrence, aggregation.lastMessage, aggregation.backtrace, aggregation.environments, aggregation.servers) error.put() logging.info('Successfully aggregated %r items for key %s', aggregation.count, errorKey) success = True except: # pylint: disable=W0702 logging.exception('Error writing to data store for key %s.', errorKey) finally: _unlockError(errorKey) else: logging.info('Could not lock %s', errorKey) if not success: # Add a retry task. logging.info('Retrying aggregation for %d items for key %s', len(instances), errorKey) aggregation.firstOccurrence = str(aggregation.firstOccurrence) aggregation.lastOccurrence = str(aggregation.lastOccurrence) aggregation.environments = list(aggregation.environments) aggregation.servers = list(aggregation.servers) taskqueue.Queue('aggregation').add([ taskqueue.Task(payload = json.dumps({'error': errorKey, 'aggregation': aggregation}), method='PULL') ]) retries += 1 q.delete_tasks(tasksByError[errorKey]) if retries: logging.warn("Retrying %d tasks", retries) for _ in range(retries): queueAggregationWorker()
def post(self): # pylint: disable=R0914, R0915 """Handles a new error report via POST.""" taskId = self.request.get("id", "0") currentId = memcache.get(AGGREGATION_ID) if taskId == "None" or not (taskId == currentId or int(taskId) % 50 == 0): # Skip this task unless it is the most recently added or if it is one of every fifty tasks. logging.debug("Skipping task %s, current is %s", taskId, currentId) return q = taskqueue.Queue("aggregation") tasks = _getTasks(q) logging.info("Leased %d tasks", len(tasks)) byError = collections.defaultdict(list) instanceKeys = [] tasksByError = collections.defaultdict(list) for task in tasks: data = json.loads(task.payload) errorKey = data["error"] if "instance" in data and "backtrace" in data: instanceKey = data["instance"] byError[errorKey].append((instanceKey, data["backtrace"])) instanceKeys.append(instanceKey) tasksByError[errorKey].append(task) elif "aggregation" in data: byError[errorKey].append(data["aggregation"]) tasksByError[errorKey].append(task) else: # Clean up any old tasks in the queue. logging.warn("Deleting an old task") q.delete_tasks([task]) retries = 0 instanceByKey = getInstanceMap(instanceKeys) for errorKey, instances in byError.items(): instances = [ keyOrDict if isinstance(keyOrDict, dict) else aggregateSingleInstance(instanceByKey[keyOrDict[0]], keyOrDict[1]) for keyOrDict in instances ] aggregation = aggregateInstances(instances) success = False if _lockError(errorKey): try: error = LoggedError.get(errorKey) aggregate( error, aggregation.count, aggregation.firstOccurrence, aggregation.lastOccurrence, aggregation.lastMessage, aggregation.backtrace, aggregation.environments, aggregation.servers, ) error.put() logging.info("Successfully aggregated %r items for key %s", aggregation.count, errorKey) success = True except: # pylint: disable=W0702 logging.exception("Error writing to data store for key %s.", errorKey) finally: _unlockError(errorKey) else: logging.info("Could not lock %s", errorKey) if not success: # Add a retry task. logging.info("Retrying aggregation for %d items for key %s", len(instances), errorKey) aggregation.firstOccurrence = str(aggregation.firstOccurrence) aggregation.lastOccurrence = str(aggregation.lastOccurrence) aggregation.environments = list(aggregation.environments) aggregation.servers = list(aggregation.servers) taskqueue.Queue("aggregation").add( [taskqueue.Task(payload=json.dumps({"error": errorKey, "aggregation": aggregation}), method="PULL")] ) retries += 1 q.delete_tasks(tasksByError[errorKey]) if retries: logging.warn("Retrying %d tasks", retries) for _ in range(retries): queueAggregationWorker()