def check_storage_consistency(context, storage_id, storage_new):
"""Check storage response returned by driver whether it matches the
storage stored in database.
:param context: The context of delfin.
:type context: delfin.context.RequestContext
:param storage_id: The uuid of storage in database.
:type storage_id: string
:param storage_new: The storage response returned by driver.
:type storage_new: dict
"""
if not storage_new:
raise exception.StorageBackendNotFound()
if not storage_new.get('serial_number'):
msg = _("Serial number should be provided by storage.")
raise exception.InvalidResults(msg)
storage_present = db.storage_get(context, storage_id)
if storage_new['serial_number'] != storage_present['serial_number']:
msg = (
_("Serial number %s does not match "
"the existing storage serial number %s.") %
(storage_new['serial_number'], storage_present['serial_number']))
raise exception.StorageSerialNumberMismatch(msg)
def get_storage(self, context):
# Do something here
sn = six.text_type(uuidutils.generate_uuid())
try:
# use existing sn if already registered storage
storage = db.storage_get(context, self.storage_id)
if storage:
sn = storage['serial_number']
except exception.StorageNotFound:
LOG.debug('Registering new storage')
except Exception:
LOG.info('Error while retrieving storage from DB')
total, used, free = self._get_random_capacity()
raw = random.randint(2000, 3000)
subscribed = random.randint(3000, 4000)
return {
'name': 'fake_driver',
'description': 'fake driver.',
'vendor': 'fake_vendor',
'model': 'fake_model',
'status': 'normal',
'serial_number': sn,
'firmware_version': '1.0.0',
'location': 'HK',
'total_capacity': total,
'used_capacity': used,
'free_capacity': free,
'raw_capacity': raw,
'subscribed_capacity': subscribed
}
def _set_synced_after(func, *args, **kwargs):
call_args = inspect.getcallargs(func, *args, **kwargs)
self = call_args['self']
sync_result = constants.ResourceSync.SUCCEED
ret = None
try:
ret = func(*args, **kwargs)
except Exception:
sync_result = constants.ResourceSync.FAILED
lock = coordination.Lock(self.storage_id)
with lock:
try:
storage = db.storage_get(self.context, self.storage_id)
except exception.StorageNotFound:
LOG.warn('Storage %s not found when set synced' %
self.storage_id)
else:
# One sync task done, sync status minus 1
# When sync status get to 0
# means all the sync tasks are completed
if storage['sync_status'] != constants.SyncStatus.SYNCED:
storage['sync_status'] -= sync_result
db.storage_update(self.context, self.storage_id,
{'sync_status': storage['sync_status']})
return ret
def _check_deleted(func, *args, **kwargs):
call_args = inspect.getcallargs(func, *args, **kwargs)
self = call_args['self']
ret = func(*args, **kwargs)
# When context.read_deleted is 'yes', db.storage_get would
# only get the storage whose 'deleted' tag is not default value
self.context.read_deleted = 'yes'
try:
db.storage_get(self.context, self.storage_id)
except exception.StorageNotFound:
LOG.debug('Storage %s not found when checking deleted' %
self.storage_id)
else:
self.remove()
self.context.read_deleted = 'no'
return ret
def collect(self, ctx, storage_id, args, start_time, end_time):
try:
LOG.debug("Performance collection for storage [%s] with start time"
" [%s] and end time [%s]" %
(storage_id, start_time, end_time))
perf_metrics = self.driver_api \
.collect_perf_metrics(ctx, storage_id,
args,
start_time, end_time)
# Fill extra labels to metric by fetching metadata from resource DB
try:
storage_details = db.storage_get(ctx, storage_id)
for m in perf_metrics:
m.labels["name"] = storage_details.name
m.labels["serial_number"] = storage_details.serial_number
except Exception as e:
msg = _('Failed to add extra labels to performance '
'metrics: {0}'.format(e))
LOG.error(msg)
return TelemetryTaskStatus.TASK_EXEC_STATUS_FAILURE
self.perf_exporter.dispatch(context, perf_metrics)
return TelemetryTaskStatus.TASK_EXEC_STATUS_SUCCESS
except Exception as e:
LOG.error("Failed to collect performance metrics for "
"storage id :{0}, reason:{1}".format(
storage_id, six.text_type(e)))
return TelemetryTaskStatus.TASK_EXEC_STATUS_FAILURE
def show(self, req, id):
ctx = req.environ['delfin.context']
query_para = {}
query_para.update(req.GET)
try:
begin_time = None
end_time = None
if query_para.get('begin_time'):
begin_time = int(query_para.get('begin_time'))
if query_para.get('end_time'):
end_time = int(query_para.get('end_time'))
except Exception:
msg = "begin_time and end_time should be integer values in " \
"milliseconds."
raise exception.InvalidInput(msg)
# When both begin_time and end_time are provided, end_time should
# be greater than begin_time
if begin_time and end_time and end_time <= begin_time:
msg = "end_time should be greater than begin_time."
raise exception.InvalidInput(msg)
storage = db.storage_get(ctx, id)
alert_list = self.driver_manager.list_alerts(ctx, id, query_para)
# Update storage attributes in each alert model
for alert in alert_list:
alert_util.fill_storage_attributes(alert, storage)
return alerts_view.build_alerts(alert_list)
def sync_alerts(self, ctx, storage_id, query_para):
""" Syncs all alerts from storage side to exporter """
LOG.info('Syncing alerts for storage id:{0}'.format(storage_id))
try:
storage = db.storage_get(ctx, storage_id)
current_alert_list = self.driver_manager.list_alerts(
ctx, storage_id, query_para)
if not len(current_alert_list):
# No alerts to sync
LOG.info('No alerts to sync from storage device for '
'storage id:{0}'.format(storage_id))
return
for alert in current_alert_list:
alert_util.fill_storage_attributes(alert, storage)
self.alert_export_manager.dispatch(ctx, current_alert_list)
LOG.info(
'Syncing storage alerts successful for storage id:{0}'.format(
storage_id))
except Exception as e:
msg = _('Failed to sync alerts from storage device: {0}'.format(
six.text_type(e)))
LOG.error(msg)
def process_alert_info(self, alert):
"""Fills alert model using driver manager interface."""
ctxt = context.get_admin_context()
storage = db.storage_get(ctxt, alert['storage_id'])
alert_model = {}
try:
alert_model = self.driver_manager.parse_alert(
ctxt, alert['storage_id'], alert)
# Fill storage specific info
if alert_model:
storage = self.get_storage_from_parsed_alert(
ctxt, storage, alert_model)
alert_util.fill_storage_attributes(alert_model, storage)
except exception.IncompleteTrapInformation as e:
LOG.warning(e)
threading.Thread(target=self.sync_storage_alert,
args=(ctxt, alert['storage_id'])).start()
except exception.AlertSourceNotFound:
LOG.info("Could not identify alert source from parsed alert. "
"Skipping the dispatch of alert")
return
except Exception as e:
LOG.error(e)
raise exception.InvalidResults(
"Failed to fill the alert model from driver.")
# Export to base exporter which handles dispatch for all exporters
if alert_model:
LOG.info("Dispatching one SNMP Trap to {} with sn {}".format(
alert_model['storage_id'], alert_model['serial_number']))
self.exporter_manager.dispatch(ctxt, [alert_model])
def _storage_exist(self, context, access_info):
access_info_dict = copy.deepcopy(access_info)
# Remove unrelated query fields
unrelated_fields = ['username', 'password']
for access in constants.ACCESS_TYPE:
if access_info_dict.get(access):
for key in unrelated_fields:
access_info_dict[access].pop(key)
# Check if storage is registered
access_info_list = db.access_info_get_all(context,
filters=access_info_dict)
for _access_info in access_info_list:
try:
storage = db.storage_get(context, _access_info['storage_id'])
if storage:
LOG.error("Storage %s has same access "
"information." % storage['id'])
return True
except exception.StorageNotFound:
# Suppose storage was not saved successfully after access
# information was saved in database when registering storage.
# Therefore, removing access info if storage doesn't exist to
# ensure the database has no residual data.
LOG.debug("Remove residual access information.")
db.access_info_delete(context, _access_info['storage_id'])
return False
def collect(self):
"""
:return:
"""
LOG.info(
'Collecting array performance metrics for storage id:{0}'.format(
self.storage_id))
try:
# collect the performance metrics from driver and push to
# prometheus exporter api
array_metrics = self.driver_api.collect_array_metrics(
self.context, self.storage_id, self.interval, self.is_historic)
# fill extra labels to metric by fetching metadata from resource DB
try:
array_details = db.storage_get(self.context,
storage_id=self.storage_id)
for m in array_metrics:
m.labels["name"] = array_details.name
m.labels["serial_number"] = array_details.serial_number
except Exception as e:
msg = _('Failed to add extra labels to array performance '
'metrics: {0}'.format(e))
LOG.error(msg)
self.perf_exporter.dispatch(self.context, array_metrics)
except Exception as e:
msg = _('Failed to collect array performance metrics from '
'driver: {0}'.format(e))
LOG.error(msg)
else:
LOG.info("Array performance metrics collection done!!!")
def delete(self, req, id):
ctxt = req.environ['delfin.context']
storage = db.storage_get(ctxt, id)
for subclass in task.StorageResourceTask.__subclasses__():
self.task_rpcapi.remove_storage_resource(
ctxt, storage['id'],
subclass.__module__ + '.' + subclass.__name__)
self.task_rpcapi.remove_storage_in_cache(ctxt, storage['id'])
def put(self, req, id, body):
"""Create a new alert source or update an exist one."""
ctx = req.environ['delfin.context']
alert_source = body
alert_source["storage_id"] = id
db.storage_get(ctx, id)
alert_source = self._input_check(alert_source)
snmp_config_to_del = self._get_snmp_config_brief(ctx, id)
if snmp_config_to_del is not None:
alert_source = db.alert_source_update(ctx, id, alert_source)
else:
alert_source = db.alert_source_create(ctx, alert_source)
snmp_config_to_add = alert_source
self.alert_rpcapi.sync_snmp_config(ctx, snmp_config_to_del,
snmp_config_to_add)
return alert_view.build_alert_source(alert_source.to_dict())
def sync(self, req, id):
"""
:param req:
:param id:
:return:
"""
ctxt = req.environ['delfin.context']
storage = db.storage_get(ctxt, id)
resource_count = len(resources.StorageResourceTask.__subclasses__())
_set_synced_if_ok(ctxt, storage['id'], resource_count)
for subclass in resources.StorageResourceTask.__subclasses__():
self.task_rpcapi.sync_storage_resource(
ctxt, storage['id'],
subclass.__module__ + '.' + subclass.__name__)
def _get_alert_source_by_host(source_ip):
"""Gets alert source for given source ip address."""
filters = {'host~': source_ip}
ctxt = context.RequestContext()
# Using the known filter and db exceptions are handled by api
alert_sources = db.alert_source_get_all(ctxt, filters=filters)
if not alert_sources:
raise exception.AlertSourceNotFoundWithHost(source_ip)
# This is to make sure unique host is configured each alert source
unique_alert_source = None
if len(alert_sources) > 1:
# Clear invalid alert_source
for alert_source in alert_sources:
try:
db.storage_get(ctxt, alert_source['storage_id'])
except exception.StorageNotFound:
LOG.warning('Found redundancy alert source for storage %s'
% alert_source['storage_id'])
try:
db.alert_source_delete(
ctxt, alert_source['storage_id'])
except Exception as e:
LOG.warning('Delete the invalid alert source failed, '
'reason is %s' % six.text_type(e))
else:
unique_alert_source = alert_source
else:
unique_alert_source = alert_sources[0]
if unique_alert_source is None:
msg = (_("Failed to get unique alert source with host %s.")
% source_ip)
raise exception.InvalidResults(msg)
return unique_alert_source
def _handle_validation_result(self, ctxt, storage_id,
category=constants.Category.FAULT):
try:
storage = db.storage_get(ctxt, storage_id)
serial_number = storage.get('serial_number')
if category == constants.Category.FAULT:
self.snmp_error_flag[serial_number] = True
self._dispatch_snmp_validation_alert(ctxt, storage, category)
elif self.snmp_error_flag.get(serial_number, True):
self.snmp_error_flag[serial_number] = False
self._dispatch_snmp_validation_alert(ctxt, storage, category)
except Exception as e:
msg = six.text_type(e)
LOG.error("Exception occurred when handling validation "
"error: %s ." % msg)
def get_capabilities(self, req, id):
"""
The API fetches capabilities from driver
associated with the storage device.
"""
# Check and fetch storage with storage_id
ctx = req.environ['delfin.context']
storage_info = db.storage_get(ctx, id)
# Fetch supported driver's capability
capabilities = self.driver_api. \
get_capabilities(ctx, storage_info['id'])
# validate capabilities
validation.validate_capabilities(capabilities)
return storage_view.build_capabilities(storage_info, capabilities)
def process_alert_info(self, alert):
"""Fills alert model using driver manager interface."""
ctxt = context.get_admin_context()
storage = db.storage_get(ctxt, alert['storage_id'])
try:
alert_model = self.driver_manager.parse_alert(
ctxt, alert['storage_id'], alert)
# Fill storage specific info
alert_util.fill_storage_attributes(alert_model, storage)
except Exception as e:
LOG.error(e)
raise exception.InvalidResults(
"Failed to fill the alert model from driver.")
# Export to base exporter which handles dispatch for all exporters
self.exporter_manager.dispatch(ctxt, alert_model)
def _set_synced_if_ok(context, storage_id, resource_count):
try:
storage = db.storage_get(context, storage_id)
except exception.StorageNotFound:
msg = 'Storage %s not found when try to set sync_status' \
% storage_id
raise exception.InvalidInput(message=msg)
else:
last_update = storage['updated_at'] or storage['created_at']
current_time = timeutils.utcnow()
interval = (current_time - last_update).seconds
# If last synchronization was within
# CONF.sync_task_expiration(in seconds), and the sync status
# is bigger than 0, it means some sync task is still running,
# the new sync task should not launch
if interval < CONF.sync_task_expiration and \
storage['sync_status'] > 0:
raise exception.StorageIsSyncing(storage['id'])
storage['sync_status'] = resource_count * constants.ResourceSync.START
db.storage_update(context, storage['id'], storage)
def _storage_exist(self, context, access_info):
access_info_dict = copy.deepcopy(access_info)
access_info_list = access_info_filter(context, access_info_dict)
for _access_info in access_info_list:
try:
storage = db.storage_get(context, _access_info['storage_id'])
if storage:
LOG.error("Storage %s has same access "
"information." % storage['id'])
return True
except exception.StorageNotFound:
# Suppose storage was not saved successfully after access
# information was saved in database when registering storage.
# Therefore, removing access info if storage doesn't exist to
# ensure the database has no residual data.
LOG.debug("Remove residual access information.")
db.access_info_delete(context, _access_info['storage_id'])
return False
def sync(self, req, id, body):
ctx = req.environ['delfin.context']
# begin_time and end_time are optional parameters
begin_time = body.get('begin_time')
end_time = body.get('end_time')
# When both begin_time and end_time are provided, end_time should
# be greater than begin_time
if begin_time and end_time and end_time <= begin_time:
msg = "end_time should be greater than begin_time."
raise exception.InvalidInput(msg)
# Check for the storage existence
_ = db.storage_get(ctx, id)
query_para = {
'begin_time': body.get('begin_time'),
'end_time': body.get('end_time')
}
# Trigger asynchronous alert syncing from storage backend
self.task_rpcapi.sync_storage_alerts(ctx, id, query_para)
def process_alert_info(self, alert):
"""Fills alert model using driver manager interface."""
ctxt = context.get_admin_context()
storage = db.storage_get(ctxt, alert['storage_id'])
alert_model = {}
try:
alert_model = self.driver_manager.parse_alert(
ctxt, alert['storage_id'], alert)
# Fill storage specific info
if alert_model:
alert_util.fill_storage_attributes(alert_model, storage)
except exception.IncompleteTrapInformation as e:
LOG.warn(e)
threading.Thread(target=self.sync_storage_alert,
args=(ctxt, alert['storage_id'])).start()
except Exception as e:
LOG.error(e)
raise exception.InvalidResults(
"Failed to fill the alert model from driver.")
# Export to base exporter which handles dispatch for all exporters
if alert_model:
self.exporter_manager.dispatch(ctxt, alert_model)
def show(self, req, id):
ctxt = req.environ['delfin.context']
storage = db.storage_get(ctxt, id)
return storage_view.build_storage(storage)
def metrics_config(self, req, body, id):
"""
:param req:
:param body:
:param id:
:return:
"""
ctxt = req.environ['delfin.context']
# check storage is registered
db.storage_get(ctxt, id)
metrics_config_dict = body
metrics_config_dict.update(body)
# get scheduler object
schedule = config.Scheduler.getInstance()
# The path of scheduler config file
config_file = CONF.scheduler.config_path
try:
# Load the scheduler configuration file
data = config.load_json_file(config_file)
storage_found = False
for storage in data.get("storages"):
config_storage_id = storage.get('id')
if config_storage_id == id:
for resource in metrics_config_dict.keys():
storage_dict = storage.get(resource)
metric_dict = metrics_config_dict.get(resource)
storage_dict.update(metric_dict)
interval = storage_dict.get('interval')
is_historic = storage_dict.get('is_historic')
job_id = id + resource
if schedule.get_job(job_id):
schedule.reschedule_job(job_id=job_id,
trigger='interval',
seconds=interval)
else:
schedule.add_job(
self.perf_collect,
'interval',
args=[id, interval, is_historic, resource],
seconds=interval,
next_run_time=datetime.now(),
id=job_id)
storage_found = True
if not storage_found:
temp_dict = {'id': id}
temp_dict.update(metrics_config_dict)
data.get("storages").append(temp_dict)
for resource in metrics_config_dict.keys():
resource_dict = metrics_config_dict.get(resource)
interval = resource_dict.get('interval')
is_historic = resource_dict.get('is_historic')
job_id = id + resource
schedule.add_job(
self.perf_collect,
'interval',
args=[id, interval, is_historic, resource],
seconds=interval,
next_run_time=datetime.now(),
id=job_id)
with open(config_file, "w") as jsonFile:
json.dump(data, jsonFile)
jsonFile.close()
except TypeError as e:
LOG.error("Error occurred during parsing of config file")
raise exception.InvalidContentType(e)
except json.decoder.JSONDecodeError as e:
msg = ("Not able to open the config file: {0}".format(config_file))
LOG.error(msg)
raise exception.InvalidInput(e.msg)
else:
return metrics_config_dict
finally:
try:
schedule.start()
except Exception as e:
LOG.debug("Scheduler is already running.{0}".format(e))
def delete(self, req, id, sequence_number):
ctx = req.environ['delfin.context']
_ = db.storage_get(ctx, id)
self.driver_manager.clear_alert(ctx, id, sequence_number)