def setup_meters_config():
"""load the meters definitions from yaml config file."""
config_file = get_config_file()
LOG.debug("Hardware snmp meter definition file: %s" % config_file)
with open(config_file) as cf:
config = cf.read()
try:
meters_config = yaml.safe_load(config)
except yaml.YAMLError as err:
if hasattr(err, 'problem_mark'):
mark = err.problem_mark
errmsg = (_LE("Invalid YAML syntax in Meter Definitions file "
"%(file)s at line: %(line)s, column: %(column)s.")
% dict(file=config_file,
line=mark.line + 1,
column=mark.column + 1))
else:
errmsg = (_LE("YAML error reading Meter Definitions file "
"%(file)s")
% dict(file=config_file))
LOG.error(errmsg)
raise
LOG.info(_LI("Meter Definitions: %s") % meters_config)
return meters_config
def record_metering_data(self, data):
if self.target == '':
# if the target was not set, do not do anything
LOG.error(_LE('Dispatcher target was not set, no meter will '
'be posted. Set the target in the ceilometer.conf '
'file.'))
return
# We may have receive only one counter on the wire
if not isinstance(data, list):
data = [data]
for meter in data:
LOG.debug(
'metering data %(counter_name)s '
'for %(resource_id)s @ %(timestamp)s: %(counter_volume)s',
{'counter_name': meter['counter_name'],
'resource_id': meter['resource_id'],
'timestamp': meter.get('timestamp', 'NO TIMESTAMP'),
'counter_volume': meter['counter_volume']})
try:
# Every meter should be posted to the target
res = requests.post(self.target,
data=json.dumps(meter),
headers=self.headers,
timeout=self.timeout)
LOG.debug('Message posting finished with status code '
'%d.', res.status_code)
except Exception as err:
LOG.exception(_LE('Failed to record metering data: %s.'), err)
def inspect_perf_events(self, instance, duration=None):
domain = self._get_domain_not_shut_off_or_raise(instance)
try:
stats = self.connection.domainListGetStats(
[domain], libvirt.VIR_DOMAIN_STATS_PERF)
perf = stats[0][1]
return virt_inspector.PerfEventsStats(
cpu_cycles=perf["perf.cpu_cycles"],
instructions=perf["perf.instructions"],
cache_references=perf["perf.cache_references"],
cache_misses=perf["perf.cache_misses"])
# NOTE(sileht): KeyError if for libvirt >=2.0.0,<2.3.0, the perf
# subsystem ws existing but not these attributes
# https://github.com/libvirt/libvirt/commit/bae660869de0612bee2a740083fb494c27e3f80c
except (AttributeError, KeyError) as e:
msg = _LE('Perf is not supported by current version of libvirt, '
'and failed to inspect perf events of '
'%(instance_uuid)s, can not get info from libvirt: '
'%(error)s') % {
'instance_uuid': instance.id, 'error': e}
raise virt_inspector.NoDataException(msg)
# domainListGetStats might launch an exception if the method or
# mbmt/mbml perf event is not supported by the underlying hypervisor
# being used by libvirt.
except libvirt.libvirtError as e:
msg = _LE('Failed to inspect perf events of %(instance_uuid)s, '
'can not get info from libvirt: %(error)s') % {
'instance_uuid': instance.id, 'error': e}
raise virt_inspector.NoDataException(msg)
def setup_meters_config():
"""Setup the meters definitions from yaml config file."""
config_file = get_config_file()
if config_file is not None:
LOG.debug(_LE("Meter Definitions configuration file: %s"), config_file)
with open(config_file) as cf:
config = cf.read()
try:
meters_config = yaml.safe_load(config)
except yaml.YAMLError as err:
if hasattr(err, 'problem_mark'):
mark = err.problem_mark
errmsg = (_LE("Invalid YAML syntax in Meter Definitions file "
"%(file)s at line: %(line)s, column: %(column)s.")
% dict(file=config_file,
line=mark.line + 1,
column=mark.column + 1))
else:
errmsg = (_LE("YAML error reading Meter Definitions file "
"%(file)s")
% dict(file=config_file))
LOG.error(errmsg)
raise
else:
LOG.debug(_LE("No Meter Definitions configuration file found!"
" Using default config."))
meters_config = {}
LOG.info(_LE("Meter Definitions: %s"), meters_config)
return meters_config
def __init__(self, definition_cfg):
self.cfg = definition_cfg
missing = [field for field in self.REQUIRED_FIELDS if not self.cfg.get(field)]
if missing:
raise MeterDefinitionException(_LE("Required fields %s not specified") % missing, self.cfg)
self._event_type = self.cfg.get("event_type")
if isinstance(self._event_type, six.string_types):
self._event_type = [self._event_type]
if "type" not in self.cfg.get("lookup", []) and self.cfg["type"] not in sample.TYPES:
raise MeterDefinitionException(_LE("Invalid type %s specified") % self.cfg["type"], self.cfg)
self._field_getter = {}
for name, field in self.cfg.items():
if name in ["event_type", "lookup"] or not field:
continue
elif isinstance(field, six.integer_types):
self._field_getter[name] = field
elif isinstance(field, dict) and name == "metadata":
meta = {}
for key, val in field.items():
parts = self.parse_jsonpath(val)
meta[key] = functools.partial(self._parse_jsonpath_field, parts)
self._field_getter["metadata"] = meta
else:
parts = self.parse_jsonpath(field)
self._field_getter[name] = functools.partial(self._parse_jsonpath_field, parts)
def __init__(self, definition_cfg):
self.cfg = definition_cfg
self._event_type = self.cfg.get('event_type')
if not self._event_type:
raise MeterDefinitionException(
_LE("Required field event_type not specified"), self.cfg)
if isinstance(self._event_type, six.string_types):
self._event_type = [self._event_type]
if ('type' not in self.cfg.get('lookup', []) and
self.cfg['type'] not in sample.TYPES):
raise MeterDefinitionException(
_LE("Invalid type %s specified") % self.cfg['type'], self.cfg)
self._field_getter = {}
for name, field in self.cfg.items():
if name in ["event_type", "lookup"] or not field:
continue
elif isinstance(field, six.integer_types):
self._field_getter[name] = field
elif isinstance(field, dict) and name == 'metadata':
meta = {}
for key, val in field.items():
parts = self.parse_jsonpath(val)
meta[key] = functools.partial(self._parse_jsonpath_field,
parts)
self._field_getter['metadata'] = meta
else:
parts = self.parse_jsonpath(field)
self._field_getter[name] = functools.partial(
self._parse_jsonpath_field, parts)
def __init__(self, definition_cfg, default_archive_policy, plugin_manager):
self._default_archive_policy = default_archive_policy
self.cfg = definition_cfg
for field, field_type in self.MANDATORY_FIELDS.items():
if field not in self.cfg:
raise declarative.DefinitionException(
_LE("Required field %s not specified") % field, self.cfg)
if not isinstance(self.cfg[field], field_type):
raise declarative.DefinitionException(
_LE("Required field %(field)s should be a %(type)s") %
{'field': field, 'type': field_type}, self.cfg)
self._attributes = {}
for name, attr_cfg in self.cfg.get('attributes', {}).items():
self._attributes[name] = declarative.Definition(name, attr_cfg,
plugin_manager)
self.metrics = {}
for t in self.cfg['metrics']:
archive_policy = self.cfg.get('archive_policy',
self._default_archive_policy)
if archive_policy is None:
self.metrics[t] = {}
else:
self.metrics[t] = dict(archive_policy_name=archive_policy)
def __init__(self, conf, parsed_url):
super(FilePublisher, self).__init__(conf, parsed_url)
self.publisher_logger = None
path = parsed_url.path
if not path:
LOG.error(_LE('The path for the file publisher is required'))
return
rfh = None
max_bytes = 0
backup_count = 0
# Handling other configuration options in the query string
if parsed_url.query:
params = urlparse.parse_qs(parsed_url.query)
if params.get('max_bytes') and params.get('backup_count'):
try:
max_bytes = int(params.get('max_bytes')[0])
backup_count = int(params.get('backup_count')[0])
except ValueError:
LOG.error(_LE('max_bytes and backup_count should be '
'numbers.'))
return
# create rotating file handler
rfh = logging.handlers.RotatingFileHandler(
path, encoding='utf8', maxBytes=max_bytes,
backupCount=backup_count)
self.publisher_logger = logging.Logger('publisher.file')
self.publisher_logger.propagate = False
self.publisher_logger.setLevel(logging.INFO)
rfh.setLevel(logging.INFO)
self.publisher_logger.addHandler(rfh)
def __init__(self, definition_cfg, plugin_manager):
self.cfg = definition_cfg
missing = [field for field in self.REQUIRED_FIELDS if not self.cfg.get(field)]
if missing:
raise declarative.MeterDefinitionException(_LE("Required fields %s not specified") % missing, self.cfg)
self._event_type = self.cfg.get("event_type")
if isinstance(self._event_type, six.string_types):
self._event_type = [self._event_type]
if "type" not in self.cfg.get("lookup", []) and self.cfg["type"] not in sample.TYPES:
raise declarative.MeterDefinitionException(_LE("Invalid type %s specified") % self.cfg["type"], self.cfg)
self._fallback_user_id = declarative.Definition("user_id", "_context_user_id|_context_user", plugin_manager)
self._fallback_project_id = declarative.Definition(
"project_id", "_context_tenant_id|_context_tenant", plugin_manager
)
self._attributes = {}
self._metadata_attributes = {}
for name in self.SAMPLE_ATTRIBUTES:
attr_cfg = self.cfg.get(name)
if attr_cfg:
self._attributes[name] = declarative.Definition(name, attr_cfg, plugin_manager)
metadata = self.cfg.get("metadata", {})
for name in metadata:
self._metadata_attributes[name] = declarative.Definition(name, metadata[name], plugin_manager)
# List of fields we expected when multiple meter are in the payload
self.lookup = self.cfg.get("lookup")
if isinstance(self.lookup, six.string_types):
self.lookup = [self.lookup]
def __init__(self, definition_cfg, default_archive_policy,
legacy_archive_policy_definition):
self._default_archive_policy = default_archive_policy
self._legacy_archive_policy_definition =\
legacy_archive_policy_definition
self.cfg = definition_cfg
for field, field_type in self.MANDATORY_FIELDS.items():
if field not in self.cfg:
raise ResourcesDefinitionException(
_LE("Required field %s not specified") % field, self.cfg)
if not isinstance(self.cfg[field], field_type):
raise ResourcesDefinitionException(
_LE("Required field %(field)s should be a %(type)s") %
{'field': field, 'type': field_type}, self.cfg)
self._field_getter = {}
for name, fval in self.cfg.get('attributes', {}).items():
if isinstance(fval, six.integer_types):
self._field_getter[name] = fval
else:
try:
parts = self.JSONPATH_RW_PARSER.parse(fval)
except Exception as e:
raise ResourcesDefinitionException(
_LE("Parse error in JSONPath specification "
"'%(jsonpath)s': %(err)s")
% dict(jsonpath=fval, err=e), self.cfg)
self._field_getter[name] = functools.partial(
self._parse_jsonpath_field, parts)
def _validate(self):
for field, field_type in self.MANDATORY_FIELDS.items():
if field not in self.cfg:
raise ResourcesDefinitionException(
_LE("Required field %s not specified") % field, self.cfg)
if not isinstance(self.cfg[field], field_type):
raise ResourcesDefinitionException(
_LE("Required field %(field)s should be a %(type)s") %
{'field': field, 'type': field_type}, self.cfg)
def _check_required_and_types(expected, definition):
for field, field_type in expected.items():
if field not in definition:
raise declarative.ResourceDefinitionException(
_LE("Required field %s not specified") % field, definition)
if not isinstance(definition[field], field_type):
raise declarative.ResourceDefinitionException(
_LE("Required field %(field)s should be a %(type)s") %
{'field': field, 'type': field_type}, definition)
def _catch_extension_load_error(mgr, ep, exc):
# Extension raising ExtensionLoadError can be ignored,
# and ignore anything we can't import as a safety measure.
if isinstance(exc, plugin_base.ExtensionLoadError):
LOG.exception(_LE("Skip loading extension for %s"), ep.name)
return
if isinstance(exc, ImportError):
LOG.error(_LE("Failed to import extension for %(name)s: "
"%(error)s"),
{'name': ep.name, 'error': exc})
return
raise exc
def _ensure_connection(self):
if self._producer:
return
try:
self._producer = kafka.KafkaProducer(bootstrap_servers=["%s:%s" % (self._host, self._port)])
except kafka.errors.KafkaError as e:
LOG.exception(_LE("Failed to connect to Kafka service: %s"), e)
raise messaging.DeliveryFailure("Kafka Client is not available, " "please restart Kafka client")
except Exception as e:
LOG.exception(_LE("Failed to connect to Kafka service: %s"), e)
raise messaging.DeliveryFailure("Kafka Client is not available, " "please restart Kafka client")
def get_samples(self, manager, cache, resources=None):
"""Return an iterable of Sample instances from polling the resources.
:param manager: The service manager invoking the plugin
:param cache: A dictionary for passing data between plugins
:param resources: end point to poll data from
"""
resources = resources or []
h_cache = cache.setdefault(self.CACHE_KEY, {})
sample_iters = []
# Get the meter identifiers to poll
identifier = self.meter_definition.name
for resource in resources:
parsed_url, res, extra_metadata = self._parse_resource(resource)
if parsed_url is None:
LOG.error(_LE("Skip invalid resource %s"), resource)
continue
ins = self._get_inspector(parsed_url)
try:
# Call hardware inspector to poll for the data
i_cache = h_cache.setdefault(res, {})
# Prepare inspector parameters and cache it for performance
param_key = parsed_url.scheme + '.' + identifier
inspector_param = self.cached_inspector_params.get(param_key)
if not inspector_param:
param = getattr(self.meter_definition,
parsed_url.scheme + '_inspector', {})
inspector_param = ins.prepare_params(param)
self.cached_inspector_params[param_key] = inspector_param
if identifier not in i_cache:
i_cache[identifier] = list(ins.inspect_generic(
host=parsed_url,
cache=i_cache,
extra_metadata=extra_metadata,
param=inspector_param))
# Generate samples
if i_cache[identifier]:
sample_iters.append(self.generate_samples(
parsed_url,
i_cache[identifier]))
except Exception as err:
LOG.exception(_LE('inspector call failed for %(ident)s '
'host %(host)s: %(err)s'),
dict(ident=identifier,
host=parsed_url.hostname,
err=err))
return itertools.chain(*sample_iters)
def refresh_pipeline(self):
"""Refreshes appropriate pipeline, then delegates to agent."""
if self.conf.refresh_pipeline_cfg:
manager = None
if hasattr(self, 'pipeline_manager'):
manager = self.pipeline_manager
elif hasattr(self, 'polling_manager'):
manager = self.polling_manager
pipeline_hash = manager.cfg_changed() if manager else None
if pipeline_hash:
try:
LOG.debug("Pipeline has been refreshed. "
"old hash: %(old)s, new hash: %(new)s",
{'old': manager.cfg_hash,
'new': pipeline_hash})
# Pipeline in the notification agent.
if hasattr(self, 'pipeline_manager'):
self.pipeline_manager = pipeline.setup_pipeline(
self.conf)
# Polling in the polling agent.
elif hasattr(self, 'polling_manager'):
self.polling_manager = pipeline.setup_polling(
self.conf)
self.pipeline_validated = True
except Exception as err:
LOG.exception(_LE('Unable to load changed pipeline: %s')
% err)
if self.conf.refresh_event_pipeline_cfg:
# Pipeline in the notification agent.
manager = (self.event_pipeline_manager
if hasattr(self, 'event_pipeline_manager') else None)
ev_pipeline_hash = manager.cfg_changed()
if ev_pipeline_hash:
try:
LOG.debug("Event Pipeline has been refreshed. "
"old hash: %(old)s, new hash: %(new)s",
{'old': manager.cfg_hash,
'new': ev_pipeline_hash})
self.event_pipeline_manager = (
pipeline. setup_event_pipeline(self.conf))
self.event_pipeline_validated = True
except Exception as err:
LOG.exception(_LE('Unable to load changed event pipeline:'
' %s') % err)
if self.pipeline_validated or self.event_pipeline_validated:
self.reload_pipeline()
self.clear_pipeline_validation_status()
def refresh_pipeline(self):
"""Refreshes appropriate pipeline, then delegates to agent."""
if cfg.CONF.refresh_pipeline_cfg:
pipeline_hash = self.pipeline_changed()
if pipeline_hash:
try:
# Pipeline in the notification agent.
if hasattr(self, 'pipeline_manager'):
self.pipeline_manager = pipeline.setup_pipeline()
# Polling in the polling agent.
elif hasattr(self, 'polling_manager'):
self.polling_manager = pipeline.setup_polling()
LOG.debug("Pipeline has been refreshed. "
"old hash: %(old)s, new hash: %(new)s",
{'old': self.pipeline_hash,
'new': pipeline_hash})
self.set_pipeline_hash(pipeline_hash)
self.pipeline_validated = True
except Exception as err:
LOG.debug("Active pipeline config's hash is %s",
self.pipeline_hash)
LOG.exception(_LE('Unable to load changed pipeline: %s')
% err)
if cfg.CONF.refresh_event_pipeline_cfg:
ev_pipeline_hash = self.pipeline_changed(pipeline.EVENT_TYPE)
if ev_pipeline_hash:
try:
# Pipeline in the notification agent.
if hasattr(self, 'event_pipeline_manager'):
self.event_pipeline_manager = (pipeline.
setup_event_pipeline())
LOG.debug("Event Pipeline has been refreshed. "
"old hash: %(old)s, new hash: %(new)s",
{'old': self.event_pipeline_hash,
'new': ev_pipeline_hash})
self.set_pipeline_hash(ev_pipeline_hash,
pipeline.EVENT_TYPE)
self.event_pipeline_validated = True
except Exception as err:
LOG.debug("Active event pipeline config's hash is %s",
self.event_pipeline_hash)
LOG.exception(_LE('Unable to load changed event pipeline:'
' %s') % err)
if self.pipeline_validated or self.event_pipeline_validated:
self.reload_pipeline()
self.clear_pipeline_validation_status()
def record_metering_data(self, data):
# We may have receive only one counter on the wire
if not data:
return
if not isinstance(data, list):
data = [data]
for meter in data:
LOG.debug(
'metering data %(counter_name)s '
'for %(resource_id)s @ %(timestamp)s: %(counter_volume)s',
{'counter_name': meter['counter_name'],
'resource_id': meter['resource_id'],
'timestamp': meter.get('timestamp', 'NO TIMESTAMP'),
'counter_volume': meter['counter_volume']})
# Convert the timestamp to a datetime instance.
# Storage engines are responsible for converting
# that value to something they can store.
if meter.get('timestamp'):
ts = timeutils.parse_isotime(meter['timestamp'])
meter['timestamp'] = timeutils.normalize_time(ts)
try:
self.conn.record_metering_data_batch(data)
except Exception as err:
LOG.error(_LE('Failed to record %(len)s: %(err)s.'),
{'len': len(data), 'err': err})
raise
def _load_definitions():
plugin_manager = extension.ExtensionManager(
namespace='ceilometer.event.trait_plugin')
meters_cfg = declarative.load_definitions(
{}, cfg.CONF.meter.meter_definitions_cfg_file,
pkg_resources.resource_filename(__name__, "data/meters.yaml"))
definitions = {}
for meter_cfg in reversed(meters_cfg['metric']):
if meter_cfg.get('name') in definitions:
# skip duplicate meters
LOG.warning(_LW("Skipping duplicate meter definition %s")
% meter_cfg)
continue
if (meter_cfg.get('volume') != 1
or not cfg.CONF.notification.disable_non_metric_meters):
try:
md = MeterDefinition(meter_cfg, plugin_manager)
except declarative.DefinitionException as me:
errmsg = (_LE("Error loading meter definition : %(err)s")
% dict(err=six.text_type(me)))
LOG.error(errmsg)
else:
definitions[meter_cfg['name']] = md
return definitions.values()
def get_samples(self, manager, cache, resources):
for instance in resources:
try:
disk_iops_info = self._populate_cache(
self.inspector,
cache,
instance,
)
for disk_iops in self._get_samples(instance,
disk_iops_info):
yield disk_iops
except virt_inspector.InstanceNotFoundException as err:
# Instance was deleted while getting samples. Ignore it.
LOG.debug('Exception while getting samples %s', err)
except ceilometer.NotImplementedError:
# Selected inspector does not implement this pollster.
LOG.debug('%(inspector)s does not provide data for '
'%(pollster)s',
{'inspector': self.inspector.__class__.__name__,
'pollster': self.__class__.__name__})
raise plugin_base.PollsterPermanentError(resources)
except Exception as err:
instance_name = util.instance_name(instance)
LOG.exception(_LE('Ignoring instance %(name)s: %(error)s'),
{'name': instance_name, 'error': err})
def record_events(self, event_models):
"""Write the events to database.
:param event_models: a list of models.Event objects.
"""
error = None
for event_model in event_models:
traits = []
if event_model.traits:
for trait in event_model.traits:
traits.append({'trait_name': trait.name,
'trait_type': trait.dtype,
'trait_value': trait.value})
try:
self.db.event.insert_one(
{'_id': event_model.message_id,
'event_type': event_model.event_type,
'timestamp': event_model.generated,
'traits': traits, 'raw': event_model.raw})
except pymongo.errors.DuplicateKeyError as ex:
LOG.info(_LI("Duplicate event detected, skipping it: %s") % ex)
except Exception as ex:
LOG.exception(_LE("Failed to record event: %s") % ex)
error = ex
if error:
raise error
def __init__(self, conf, cfg, transformer_manager, publisher_manager):
self.conf = conf
self.cfg = cfg
try:
self.name = cfg['name']
# It's legal to have no transformer specified
self.transformer_cfg = cfg.get('transformers') or []
except KeyError as err:
raise PipelineException(
"Required field %s not specified" % err.args[0], cfg)
if not cfg.get('publishers'):
raise PipelineException("No publisher specified", cfg)
self.publishers = []
for p in cfg['publishers']:
if '://' not in p:
# Support old format without URL
p = p + "://"
try:
self.publishers.append(publisher_manager.get(p))
except Exception:
LOG.error(_LE("Unable to load publisher %s"), p,
exc_info=True)
self.multi_publish = True if len(self.publishers) > 1 else False
self.transformers = self._setup_transformers(cfg, transformer_manager)
def record_events(self, events):
def _build_bulk_index(event_list):
for ev in event_list:
traits = {t.name: t.value for t in ev.traits}
yield {'_op_type': 'create',
'_index': '%s_%s' % (self.index_name,
ev.generated.date().isoformat()),
'_type': ev.event_type,
'_id': ev.message_id,
'_source': {'timestamp': ev.generated.isoformat(),
'traits': traits,
'raw': ev.raw}}
error = None
for ok, result in helpers.streaming_bulk(
self.conn, _build_bulk_index(events)):
if not ok:
__, result = result.popitem()
if result['status'] == 409:
LOG.info(_LI('Duplicate event detected, skipping it: %s'),
result)
else:
LOG.exception(_LE('Failed to record event: %s'), result)
error = storage.StorageUnknownWriteError(result)
if self._refresh_on_write:
self.conn.indices.refresh(index='%s_*' % self.index_name)
while self.conn.cluster.pending_tasks(local=True)['tasks']:
pass
if error:
raise error
def record_events(self, events):
for event in events:
rd = self._get_resource_definition_from_event(event['event_type'])
if not rd:
LOG.debug("No gnocchi definition for event type: %s",
event['event_type'])
continue
rd, operation = rd
resource_type = rd.cfg['resource_type']
resource = rd.event_attributes(event)
if operation == EVENT_DELETE:
ended_at = timeutils.utcnow().isoformat()
resources_to_end = [resource]
extra_resources = cfg['event_associated_resources'].items()
for resource_type, filters in extra_resources:
resources_to_end.extend(self._gnocchi.search_resource(
resource_type, filters['query'] % resource['id']))
for resource in resources_to_end:
try:
self._gnocchi.update_resource(resource_type,
resource['id'],
{'ended_at': ended_at})
except gnocchi_exc.NoSuchResource:
LOG.debug(_("Delete event received on unexiting "
"resource (%s), ignore it.") %
resource['id'])
except Exception:
LOG.error(_LE("Fail to update the resource %s") %
resource, exc_info=True)
def _publish_samples(self, start, samples):
"""Push samples into pipeline for publishing.
:param start: The first transformer that the sample will be injected.
This is mainly for flush() invocation that transformer
may emit samples.
:param samples: Sample list.
"""
transformed_samples = []
if not self.transformers:
transformed_samples = samples
else:
for sample in samples:
LOG.debug(
"Pipeline %(pipeline)s: Transform sample "
"%(smp)s from %(trans)s transformer", {'pipeline': self,
'smp': sample,
'trans': start})
sample = self._transform_sample(start, sample)
if sample:
transformed_samples.append(sample)
if transformed_samples:
for p in self.publishers:
try:
p.publish_samples(transformed_samples)
except Exception:
LOG.error(_LE("Pipeline %(pipeline)s: Continue after "
"error from publisher %(pub)s")
% {'pipeline': self, 'pub': p},
exc_info=True)
def get_samples(self, manager, cache, resources):
self._inspection_duration = self._record_poll_time()
for instance in resources:
try:
c_data = self._populate_cache(
self.inspector,
cache,
instance,
)
for s in self._get_samples(instance, c_data):
yield s
except virt_inspector.InstanceNotFoundException as err:
# Instance was deleted while getting samples. Ignore it.
LOG.debug('Exception while getting samples %s', err)
except virt_inspector.InstanceShutOffException as e:
LOG.debug('Instance %(instance_id)s was shut off while '
'getting samples of %(pollster)s: %(exc)s',
{'instance_id': instance.id,
'pollster': self.__class__.__name__, 'exc': e})
except virt_inspector.NoDataException as e:
LOG.warning(_LW('Cannot inspect data of %(pollster)s for '
'%(instance_id)s, non-fatal reason: %(exc)s'),
{'pollster': self.__class__.__name__,
'instance_id': instance.id, 'exc': e})
raise plugin_base.PollsterPermanentError(resources)
except ceilometer.NotImplementedError:
# Selected inspector does not implement this pollster.
LOG.debug('Obtaining memory bandwidth is not implemented'
' for %s', self.inspector.__class__.__name__)
except Exception as err:
LOG.exception(_LE('Could not get memory bandwidth for '
'%(id)s: %(e)s'), {'id': instance.id,
'e': err})
def record_metering_data(self, data):
# We may have receive only one counter on the wire
if not isinstance(data, list):
data = [data]
for meter in data:
LOG.debug(
"metering data %(counter_name)s " "for %(resource_id)s @ %(timestamp)s: %(counter_volume)s",
{
"counter_name": meter["counter_name"],
"resource_id": meter["resource_id"],
"timestamp": meter.get("timestamp", "NO TIMESTAMP"),
"counter_volume": meter["counter_volume"],
},
)
if publisher_utils.verify_signature(meter, self.conf.publisher.telemetry_secret):
try:
# Convert the timestamp to a datetime instance.
# Storage engines are responsible for converting
# that value to something they can store.
if meter.get("timestamp"):
ts = timeutils.parse_isotime(meter["timestamp"])
meter["timestamp"] = timeutils.normalize_time(ts)
self.meter_conn.record_metering_data(meter)
except Exception as err:
LOG.exception(_LE("Failed to record metering data: %s"), err)
# raise the exception to propagate it up in the chain.
raise
else:
LOG.warning(_LW("message signature invalid, discarding message: %r"), meter)
def get_connection(purpose):
try:
return storage.get_connection_from_config(cfg.CONF, purpose)
except Exception as err:
params = {"purpose": purpose, "err": err}
LOG.exception(_LE("Failed to connect to db, purpose %(purpose)s "
"retry later: %(err)s") % params)
def record_events(self, event_models):
"""Write the events to Hbase.
:param event_models: a list of models.Event objects.
"""
error = None
with self.conn_pool.connection() as conn:
events_table = conn.table(self.EVENT_TABLE)
for event_model in event_models:
# Row key consists of timestamp and message_id from
# models.Event or purposes of storage event sorted by
# timestamp in the database.
ts = event_model.generated
row = hbase_utils.prepare_key(hbase_utils.timestamp(ts, reverse=False), event_model.message_id)
event_type = event_model.event_type
traits = {}
if event_model.traits:
for trait in event_model.traits:
key = hbase_utils.prepare_key(trait.name, trait.dtype)
traits[key] = trait.value
record = hbase_utils.serialize_entry(traits, event_type=event_type, timestamp=ts, raw=event_model.raw)
try:
events_table.put(row, record)
except Exception as ex:
LOG.exception(_LE("Failed to record event: %s") % ex)
error = ex
if error:
raise error
def extract_my_subset(self, group_id, iterable, attempt=0):
"""Filters an iterable, returning only objects assigned to this agent.
We have a list of objects and get a list of active group members from
`tooz`. We then hash all the objects into buckets and return only
the ones that hashed into *our* bucket.
"""
if not group_id:
return iterable
if group_id not in self._groups:
self.join_group(group_id)
try:
members = self._get_members(group_id)
LOG.debug('Members of group: %s, Me: %s', members, self._my_id)
if self._my_id not in members:
raise tooz.coordination.MemberNotJoined(group_id, self._my_id)
hr = utils.HashRing(members)
filtered = [v for v in iterable
if hr.get_node(str(v)) == self._my_id]
LOG.debug('My subset: %s', [str(f) for f in filtered])
return filtered
except tooz.coordination.MemberNotJoined:
if attempt >= 5:
raise
LOG.warning(_LW('Cannot extract tasks because agent failed to '
'join group properly. Rejoining group.'))
self.join_group(group_id)
return self.extract_my_subset(group_id, iterable, attempt + 1)
except tooz.coordination.ToozError:
LOG.exception(_LE('Error getting group membership info from '
'coordination backend.'))
return []
def discover(self, manager, param=None):
"""Discover resources to monitor.
instance_get_all will return all instances if last_run is None,
and will return only the instances changed since the last_run time.
"""
try:
instances = self.nova_cli.instance_get_all(self.last_run)
except Exception:
# NOTE(zqfan): instance_get_all is wrapped and will log exception
# when there is any error. It is no need to raise it again and
# print one more time.
return []
for instance in instances:
if getattr(instance, 'OS-EXT-STS:vm_state',
None) in ['deleted', 'error']:
self.instances.pop(instance.id, None)
else:
self.instances[instance.id] = instance
self.last_run = timeutils.utcnow(True).isoformat()
resources = []
for instance in self.instances.values():
try:
ip_address = self._address(instance, 'addr')
final_address = self._make_resource_url(ip_address)
resource = {
'resource_id': instance.id,
'resource_url': final_address,
'mac_addr': self._address(instance,
'OS-EXT-IPS-MAC:mac_addr'),
'image_id': instance.image['id'],
'flavor_id': instance.flavor['id']
}
resources.append(resource)
except KeyError:
LOG.error(
_LE("Couldn't obtain IP address of "
"instance %s") % instance.id)
return resources
def discover(self, discovery=None, discovery_cache=None):
resources = []
discovery = discovery or []
for url in discovery:
if discovery_cache is not None and url in discovery_cache:
resources.extend(discovery_cache[url])
continue
name, param = self._parse_discoverer(url)
discoverer = self._discoverer(name)
if discoverer:
try:
if discoverer.KEYSTONE_REQUIRED_FOR_SERVICE:
service_type = getattr(
cfg.CONF.service_types,
discoverer.KEYSTONE_REQUIRED_FOR_SERVICE)
if not self.keystone.service_catalog.get_endpoints(
service_type=service_type):
LOG.warning(
_LW('Skipping %(name)s, %(service_type)s service '
'is not registered in keystone'), {
'name': name,
'service_type': service_type
})
continue
discovered = discoverer.discover(self, param)
partitioned = self.partition_coordinator.extract_my_subset(
self.construct_group_id(discoverer.group_id),
discovered)
resources.extend(partitioned)
if discovery_cache is not None:
discovery_cache[url] = partitioned
except ks_exceptions.ClientException as e:
LOG.error(
_LE('Skipping %(name)s, keystone issue: '
'%(exc)s'), {
'name': name,
'exc': e
})
except Exception as err:
LOG.exception(_('Unable to discover resources: %s') % err)
else:
LOG.warning(_('Unknown discovery extension: %s') % name)
return resources
def get_samples(self, manager, cache, resources):
self._inspection_duration = self._record_poll_time()
for instance in resources:
try:
c_data = self._populate_cache(
self.inspector,
cache,
instance,
)
for s in self._get_samples(instance, c_data):
yield s
except virt_inspector.InstanceNotFoundException as err:
# Instance was deleted while getting samples. Ignore it.
LOG.debug('Exception while getting samples %s', err)
except virt_inspector.InstanceShutOffException as e:
LOG.debug(
'Instance %(instance_id)s was shut off while '
'getting samples of %(pollster)s: %(exc)s', {
'instance_id': instance.id,
'pollster': self.__class__.__name__,
'exc': e
})
except virt_inspector.NoDataException as e:
LOG.warning(
_LW('Cannot inspect data of %(pollster)s for '
'%(instance_id)s, non-fatal reason: %(exc)s'), {
'pollster': self.__class__.__name__,
'instance_id': instance.id,
'exc': e
})
raise plugin_base.PollsterPermanentError(resources)
except ceilometer.NotImplementedError:
# Selected inspector does not implement this pollster.
LOG.debug(
'Obtaining memory bandwidth is not implemented'
' for %s', self.inspector.__class__.__name__)
except Exception as err:
LOG.exception(
_LE('Could not get memory bandwidth for '
'%(id)s: %(e)s'), {
'id': instance.id,
'e': err
})
def parse_fields(self, field, message):
fval = self.cfg.get(field)
if not fval:
return
if isinstance(fval, six.integer_types):
return fval
try:
parts = jsonpath_rw.parse(fval)
except Exception as e:
raise MeterDefinitionException(
_LE("Parse error in JSONPath specification "
"'%(jsonpath)s': %(err)s") % dict(jsonpath=parts, err=e),
self.cfg)
values = [
match.value for match in parts.find(message)
if match.value is not None
]
if values:
return values[0]
def _inner():
try:
join_req = self._coordinator.join_group(group_id)
join_req.get()
LOG.info(_LI('Joined partitioning group %s'), group_id)
except tooz.coordination.MemberAlreadyExist:
return
except tooz.coordination.GroupNotCreated:
create_grp_req = self._coordinator.create_group(group_id)
try:
create_grp_req.get()
except tooz.coordination.GroupAlreadyExist:
pass
raise ErrorJoiningPartitioningGroup()
except tooz.coordination.ToozError:
LOG.exception(_LE('Error joining partitioning group %s,'
' re-trying'), group_id)
raise ErrorJoiningPartitioningGroup()
self._groups.add(group_id)
def _parse_field(self, field, sample):
# TODO(sileht): share this with
# https://review.openstack.org/#/c/197633/
if not field:
return
if isinstance(field, six.integer_types):
return field
try:
parts = jsonpath_rw.parse(field)
except Exception as e:
raise ResourcesDefinitionException(
_LE("Parse error in JSONPath specification "
"'%(jsonpath)s': %(err)s") % dict(jsonpath=field, err=e),
self.cfg)
values = [
match.value for match in parts.find(sample)
if match.value is not None
]
if values:
return values[0]
def _load_definitions():
plugin_manager = extension.ExtensionManager(
namespace='ceilometer.event.trait_plugin')
meters_cfg = declarative.load_definitions(
{}, cfg.CONF.meter.meter_definitions_cfg_file,
pkg_resources.resource_filename(__name__, "data/meters.yaml"))
definitions = []
for meter_cfg in reversed(meters_cfg['metric']):
if (meter_cfg['volume'] != 1
or not cfg.CONF.notification.disable_non_metric_meters):
try:
md = MeterDefinition(meter_cfg, plugin_manager)
except declarative.DefinitionException as me:
errmsg = (_LE("Error loading meter definition : %(err)s")
% dict(err=six.text_type(me)))
LOG.error(errmsg)
else:
definitions.append(md)
return definitions
def record_events(self, events):
if not isinstance(events, list):
events = [events]
for event in events:
res = None
try:
res = requests.post(self.event_target,
data=event,
headers=self.headers,
timeout=self.timeout)
res.raise_for_status()
except Exception:
error_code = res.status_code if res else 'unknown'
LOG.exception(
_LE('Status Code: %{code}s. Failed to dispatch '
'event: %{event}s'), {
'code': error_code,
'event': event
})
def post_event(self, event):
res = None
try:
event_json = json.dumps(event)
LOG.trace('Event Message: %s', event_json)
res = requests.post(self.event_target,
data=event_json,
headers=self.headers,
verify=self.verify_ssl,
timeout=self.timeout)
LOG.debug('Event Message posting to %s: status code %d.',
self.event_target, res.status_code)
res.raise_for_status()
except requests.exceptions.HTTPError:
LOG.exception(
_LE('Status Code: %(code)s. '
'Failed to dispatch event: %(event)s') % {
'code': res.status_code,
'event': event_json
})
def sample(self, messages):
"""RPC endpoint for notification messages
When another service sends a notification over the message
bus, this method receives it.
"""
goods = []
for sample in chain.from_iterable(m["payload"] for m in messages):
if publisher_utils.verify_signature(sample, self.secret):
goods.append(sample)
else:
LOG.warning(
_LW('notification signature invalid, '
'discarding: %s'), sample)
try:
self.dispatcher_manager.map_method(self.method, goods)
except Exception:
LOG.exception(
_LE("Dispatcher failed to handle the notification, "
"re-queuing it."))
return oslo_messaging.NotificationResult.REQUEUE
def gnocchi_project_id(self):
if self._gnocchi_project_id is not None:
return self._gnocchi_project_id
with self._gnocchi_project_id_lock:
if self._gnocchi_project_id is None:
try:
project = self._ks_client.projects.find(
name=self.conf.dispatcher_gnocchi.filter_project)
except ka_exceptions.NotFound:
LOG.warning(_LW('gnocchi project not found in keystone,'
' ignoring the filter_service_activity '
'option'))
self.filter_service_activity = False
return None
except Exception:
LOG.exception(_LE('fail to retrieve user of Gnocchi '
'service'))
raise
self._gnocchi_project_id = project.id
LOG.debug("gnocchi project found: %s", self.gnocchi_project_id)
return self._gnocchi_project_id
def post_meter(self, meter):
meter_json = json.dumps(meter)
res = None
try:
LOG.trace('Meter Message: %s', meter_json)
res = requests.post(self.target,
data=meter_json,
headers=self.headers,
verify=self.verify_ssl,
timeout=self.timeout)
LOG.debug('Meter message posting finished with status code '
'%d.', res.status_code)
res.raise_for_status()
except requests.exceptions.HTTPError:
LOG.exception(
_LE('Status Code: %(code)s. '
'Failed to dispatch meter: %(meter)s') % {
'code': res.status_code,
'meter': meter_json
})
def sample(self, ctxt, publisher_id, event_type, payload, metadata):
events = []
for ev in payload:
try:
events.append(
models.Event(
message_id=ev['message_id'],
event_type=ev['event_type'],
generated=timeutils.normalize_time(
timeutils.parse_isotime(ev['generated'])),
traits=[models.Trait(
name, dtype,
models.Trait.convert_value(dtype, value))
for name, dtype, value in ev['traits']],
raw=ev.get('raw', {}))
)
except Exception:
LOG.exception(_LE("Error processing event and it will be "
"dropped: %s"), ev)
return super(EventEndpoint, self).sample(
ctxt, publisher_id, event_type, events, metadata)
def record_events(self, events):
if not isinstance(events, list):
events = [events]
for event in events:
if publisher_utils.verify_signature(
event, self.conf.publisher.telemetry_secret):
res = None
try:
res = requests.post(self.event_target, data=event,
headers=self.headers,
timeout=self.timeout)
res.raise_for_status()
except Exception:
error_code = res.status_code if res else 'unknown'
LOG.exception(_LE('Status Code: %{code}s. Failed to'
'dispatch event: %{event}s'),
{'code': error_code, 'event': event})
else:
LOG.warning(_LW(
'event signature invalid, discarding event: %s'), event)
def load_definitions(config_def):
if not config_def:
return []
meter_defs = {}
for event_def in reversed(config_def['metric']):
if event_def.get('name') in meter_defs:
# skip duplicate meters
LOG.warning(
_LW("Skipping duplicate meter definition %s") % event_def)
continue
try:
if (event_def['volume'] != 1
or not cfg.CONF.notification.disable_non_metric_meters):
md = MeterDefinition(event_def)
meter_defs[event_def['name']] = md
except MeterDefinitionException as me:
errmsg = (_LE("Error loading meter definition : %(err)s") %
dict(err=me.message))
LOG.error(errmsg)
return meter_defs.values()
def _do_post(self, data):
if not data:
LOG.debug('Data set is empty!')
return
session = requests.Session()
session.mount(self.target,
adapters.HTTPAdapter(max_retries=self.max_retries))
content = ','.join([jsonutils.dumps(item) for item in data])
content = '[' + content + ']'
LOG.debug('Data to be posted by HttpPublisher: %s' % content)
res = session.post(self.target,
data=content,
headers=self.headers,
timeout=self.timeout)
if res.status_code >= 300:
LOG.error(
_LE('Data post failed with status code %s') % res.status_code)
def get_samples(self, manager, cache, resources):
self._inspection_duration = self._record_poll_time()
for instance in resources:
instance_name = util.instance_name(instance)
LOG.debug('checking net info for instance %s', instance.id)
try:
vnics = self._get_vnics_for_instance(
cache,
self.inspector,
instance,
)
for vnic, info in vnics:
LOG.debug(self.NET_USAGE_MESSAGE, instance_name, vnic.name,
self._get_rx_info(info), self._get_tx_info(info))
yield self._get_sample(instance, vnic, info)
except virt_inspector.InstanceNotFoundException as err:
# Instance was deleted while getting samples. Ignore it.
LOG.debug('Exception while getting samples %s', err)
except virt_inspector.InstanceShutOffException as e:
LOG.debug(
'Instance %(instance_id)s was shut off while '
'getting samples of %(pollster)s: %(exc)s', {
'instance_id': instance.id,
'pollster': self.__class__.__name__,
'exc': e
})
except ceilometer.NotImplementedError:
# Selected inspector does not implement this pollster.
LOG.debug(
'%(inspector)s does not provide data for '
'%(pollster)s', {
'inspector': self.inspector.__class__.__name__,
'pollster': self.__class__.__name__
})
raise plugin_base.PollsterPermanentError(resources)
except Exception as err:
LOG.exception(_LE('Ignoring instance %(name)s: %(error)s'), {
'name': instance_name,
'error': err
})
def extract_my_subset(self, group_id, iterable):
"""Filters an iterable, returning only objects assigned to this agent.
We have a list of objects and get a list of active group members from
`tooz`. We then hash all the objects into buckets and return only
the ones that hashed into *our* bucket.
"""
if not group_id:
return iterable
if group_id not in self._groups:
self.join_group(group_id)
try:
members = self._get_members(group_id)
LOG.debug('Members of group %s are: %s, Me: %s', group_id, members,
self._my_id)
if self._my_id not in members:
LOG.warning(
_LW('Cannot extract tasks because agent failed to '
'join group properly. Rejoining group.'))
self.join_group(group_id)
members = self._get_members(group_id)
if self._my_id not in members:
raise MemberNotInGroupError(group_id, members, self._my_id)
LOG.debug('Members of group %s are: %s, Me: %s', group_id,
members, self._my_id)
hr = utils.HashRing(members)
iterable = list(iterable)
filtered = [
v for v in iterable
if hr.get_node(six.text_type(v)) == self._my_id
]
LOG.debug('The universal set: %s, my subset: %s',
[six.text_type(f) for f in iterable],
[six.text_type(f) for f in filtered])
return filtered
except tooz.coordination.ToozError:
LOG.exception(
_LE('Error getting group membership info from '
'coordination backend.'))
return []
def record_events(self, events):
def _build_bulk_index(event_list):
for ev in event_list:
traits = {t.name: t.value for t in ev.traits}
yield {
'_op_type':
'create',
'_index':
'%s_%s' %
(self.index_name, ev.generated.date().isoformat()),
'_type':
ev.event_type,
'_id':
ev.message_id,
'_source': {
'timestamp': ev.generated.isoformat(),
'traits': traits,
'raw': ev.raw
}
}
error = None
for ok, result in helpers.streaming_bulk(self.conn,
_build_bulk_index(events)):
if not ok:
__, result = result.popitem()
if result['status'] == 409:
LOG.info(
_LI('Duplicate event detected, skipping it: %s') %
result)
else:
LOG.exception(_LE('Failed to record event: %s') % result)
error = storage.StorageUnknownWriteError(result)
if self._refresh_on_write:
self.conn.indices.refresh(index='%s_*' % self.index_name)
while self.conn.cluster.pending_tasks(local=True)['tasks']:
pass
if error:
raise error
def closure(self, *args, **kwargs):
# NOTE(idegtiarov) options max_retries and retry_interval have been
# registered in storage.__init__ in oslo_db.options.set_defaults
# default values for both options are 10.
max_retries = self.conf.database.max_retries
retry_interval = self.conf.database.retry_interval
attempts = 0
while True:
try:
return call(self, *args, **kwargs)
except pymongo.errors.AutoReconnect as err:
if 0 <= max_retries <= attempts:
LOG.error(_LE('Unable to reconnect to the primary mongodb '
'after %(retries)d retries. Giving up.') %
{'retries': max_retries})
raise
LOG.warning(_('Unable to reconnect to the primary '
'mongodb: %(errmsg)s. Trying again in '
'%(retry_interval)d seconds.') %
{'errmsg': err, 'retry_interval': retry_interval})
attempts += 1
time.sleep(retry_interval)
def get_samples(self, manager, cache, resources):
self._inspection_duration = self._record_poll_time()
for instance in resources:
LOG.debug('Checking resident memory for instance %s',
instance.id)
try:
memory_info = self.inspector.inspect_memory_resident(
instance, self._inspection_duration)
LOG.debug("RESIDENT MEMORY: %(instance)s %(resident)f",
{'instance': instance,
'resident': memory_info.resident})
yield util.make_sample_from_instance(
instance,
name='memory.resident',
type=sample.TYPE_GAUGE,
unit='MB',
volume=memory_info.resident,
)
except virt_inspector.InstanceNotFoundException as err:
# Instance was deleted while getting samples. Ignore it.
LOG.debug('Exception while getting samples %s', err)
except virt_inspector.InstanceShutOffException as e:
LOG.warn(_LW('Instance %(instance_id)s was shut off while '
'getting samples of %(pollster)s: %(exc)s'),
{'instance_id': instance.id,
'pollster': self.__class__.__name__, 'exc': e})
except virt_inspector.NoDataException as e:
LOG.warn(_LW('Cannot inspect data of %(pollster)s for '
'%(instance_id)s, non-fatal reason: %(exc)s'),
{'pollster': self.__class__.__name__,
'instance_id': instance.id, 'exc': e})
except ceilometer.NotImplementedError:
# Selected inspector does not implement this pollster.
LOG.debug('Obtaining Resident Memory is not implemented'
' for %s', self.inspector.__class__.__name__)
except Exception as err:
LOG.exception(_LE('Could not get Resident Memory Usage for '
'%(id)s: %(e)s'), {'id': instance.id,
'e': err})
def _do_post(self, data):
if not data:
LOG.debug('Data set is empty!')
return
data = jsonutils.dumps(data)
LOG.trace('Message: %s', data)
try:
res = self.session.post(self.target,
data=data,
headers=self.headers,
timeout=self.timeout,
verify=self.verify_ssl)
res.raise_for_status()
LOG.debug('Message posting to %s: status code %d.', self.target,
res.status_code)
except requests.exceptions.HTTPError:
LOG.exception(
_LE('Status Code: %(code)s. '
'Failed to dispatch message: %(data)s') % {
'code': res.status_code,
'data': data
})
def _transform_sample(self, start, sample):
try:
for transformer in self.transformers[start:]:
sample = transformer.handle_sample(sample)
if not sample:
LOG.debug(
"Pipeline %(pipeline)s: Sample dropped by "
"transformer %(trans)s", {
'pipeline': self,
'trans': transformer
})
return
return sample
except Exception:
LOG.error(_LE("Pipeline %(pipeline)s: Exit after error "
"from transformer %(trans)s "
"for %(smp)s") % {
'pipeline': self,
'trans': transformer,
'smp': sample
},
exc_info=True)
def _load_resources_definitions(cls, conf):
res_def_file = cls._get_config_file(
conf, conf.dispatcher_gnocchi.resources_definition_file)
data = {}
if res_def_file is not None:
with open(res_def_file) as data_file:
try:
data = yaml.safe_load(data_file)
except ValueError:
data = {}
legacy_archive_policies = cls._load_archive_policy(conf)
resource_defs = []
for resource in data.get('resources', []):
try:
resource_defs.append(
ResourcesDefinition(resource,
conf.dispatcher_gnocchi.archive_policy,
legacy_archive_policies))
except Exception as exc:
LOG.error(_LE("Failed to load resource due to error %s") % exc)
return resource_defs
def _process_queue(self, queue):
current_retry = 0
while queue:
data = queue[0]
try:
self._send(data)
except Exception:
LOG.warn(_LW("Failed to publish %d datum"),
sum([len(d) for d in queue]))
if self.policy == 'queue':
return queue
elif self.policy == 'drop':
return []
current_retry += 1
if self.current_retry >= self.max_retry:
self.local_queue = []
LOG.exception(_LE("Failed to retry to send sample data "
"with max_retry times"))
raise
else:
queue.pop(0)
return []
def join_group(self, group_id):
if (not self._coordinator or not self._coordinator.is_started
or not group_id):
return
while True:
try:
join_req = self._coordinator.join_group(group_id)
join_req.get()
LOG.info(_LI('Joined partitioning group %s'), group_id)
break
except tooz.coordination.MemberAlreadyExist:
return
except tooz.coordination.GroupNotCreated:
create_grp_req = self._coordinator.create_group(group_id)
try:
create_grp_req.get()
except tooz.coordination.GroupAlreadyExist:
pass
except tooz.coordination.ToozError:
LOG.exception(_LE('Error joining partitioning group %s,'
' re-trying'), group_id)
self._groups.add(group_id)
def _load_definitions(self):
plugin_manager = extension.ExtensionManager(
namespace='ceilometer.event.trait_plugin')
meters_cfg = declarative.load_definitions(
self.manager.conf, {},
self.manager.conf.meter.meter_definitions_cfg_file,
pkg_resources.resource_filename(__name__, "data/meters.yaml"))
definitions = {}
for meter_cfg in reversed(meters_cfg['metric']):
if meter_cfg.get('name') in definitions:
# skip duplicate meters
LOG.warning(
_LW("Skipping duplicate meter definition %s") % meter_cfg)
continue
try:
md = MeterDefinition(meter_cfg, plugin_manager)
except declarative.DefinitionException as e:
errmsg = _LE("Error loading meter definition: %s")
LOG.error(errmsg, six.text_type(e))
else:
definitions[meter_cfg['name']] = md
return definitions.values()
def extract_my_subset(self, group_id, iterable):
"""Filters an iterable, returning only objects assigned to this agent.
We have a list of objects and get a list of active group members from
`tooz`. We then hash all the objects into buckets and return only
the ones that hashed into *our* bucket.
"""
if not group_id:
return iterable
if group_id not in self._groups:
self.join_group(group_id)
try:
members = self._get_members(group_id)
LOG.debug('Members of group: %s', members)
hr = utils.HashRing(members)
filtered = [v for v in iterable
if hr.get_node(str(v)) == self._my_id]
LOG.debug('My subset: %s', [str(f) for f in filtered])
return filtered
except tooz.coordination.ToozError:
LOG.exception(_LE('Error getting group membership info from '
'coordination backend.'))
return []
def _publish_samples(self, start, samples):
"""Push samples into pipeline for publishing.
:param start: The first transformer that the sample will be injected.
This is mainly for flush() invocation that transformer
may emit samples.
:param samples: Sample list.
"""
transformed_samples = []
if not self.transformers:
transformed_samples = samples
else:
for sample in samples:
LOG.debug(
"Pipeline %(pipeline)s: Transform sample "
"%(smp)s from %(trans)s transformer", {
'pipeline': self,
'smp': sample,
'trans': start
})
sample = self._transform_sample(start, sample)
if sample:
transformed_samples.append(sample)
if transformed_samples:
for p in self.publishers:
try:
p.publish_samples(transformed_samples)
except Exception:
LOG.error(_LE("Pipeline %(pipeline)s: Continue after "
"error from publisher %(pub)s") % {
'pipeline': self,
'pub': p
},
exc_info=True)
def _process_queue(self, queue, policy):
current_retry = 0
while queue:
topic, data = queue[0]
try:
self._send(topic, data)
except DeliveryFailure:
data = sum([len(m) for __, m in queue])
if policy == 'queue':
LOG.warning(_("Failed to publish %d datapoints, queue "
"them"), data)
return queue
elif policy == 'drop':
LOG.warning(_("Failed to publish %d datapoints, "
"dropping them"), data)
return []
current_retry += 1
if current_retry >= self.max_retry:
LOG.exception(_LE("Failed to retry to send sample data "
"with max_retry times"))
raise
else:
queue.pop(0)
return []