def __init__(self,
host: str,
client: ClientAPI,
kademlia: KademliaAPI,
influx_server: str,
influx_user: str,
influx_password: str,
influx_database: str,
influx_port: int = 443,
influx_protocol: str = 'https',
reporting_frequency: int = 10,
process_collection_frequency: int = 3):
self._influx_server = influx_server
self._reporting_frequency = reporting_frequency
self._process_collection_frequency = process_collection_frequency
self._registry = HostMetricsRegistry(host)
self._reporter = InfluxReporter(
registry=self._registry,
protocol=influx_protocol,
port=influx_port,
database=influx_database,
username=influx_user,
password=influx_password,
server=influx_server
)
self.client = client
self.kademlia = kademlia
def __init__(self, influx_server: str, influx_user: str,
influx_password: str, influx_database: str, host: str,
port: int, protocol: str, reporting_frequency: int):
self._unreported_error: Exception = None
self._last_time_reported: float = 0.0
self._influx_server = influx_server
self._reporting_frequency = reporting_frequency
self._registry = HostMetricsRegistry(host)
self._reporter = InfluxReporter(registry=self._registry,
database=influx_database,
username=influx_user,
password=influx_password,
protocol=protocol,
port=port,
server=influx_server)
def __init__(self,
influx_server: str,
influx_user: str,
influx_password: str,
influx_database: str,
host: str,
reporting_frequency: int = 10):
self._influx_server = influx_server
self._reporting_frequency = reporting_frequency
self._registry = HostMetricsRegistry(host)
self._reporter = InfluxReporter(registry=self._registry,
protocol='https',
port=443,
database=influx_database,
username=influx_user,
password=influx_password,
server=influx_server)
class MetricsService(Service):
"""
A service to provide a registry where metrics instruments can be registered and retrieved from.
It continuously reports metrics to the specified InfluxDB instance.
"""
def __init__(self,
influx_server: str,
influx_user: str,
influx_password: str,
influx_database: str,
host: str,
reporting_frequency: int = 10):
self._influx_server = influx_server
self._reporting_frequency = reporting_frequency
self._registry = HostMetricsRegistry(host)
self._reporter = InfluxReporter(registry=self._registry,
protocol='https',
port=443,
database=influx_database,
username=influx_user,
password=influx_password,
server=influx_server)
logger = get_extended_debug_logger(
'trinity.components.builtin.metrics.MetricsService')
@property
def registry(self) -> HostMetricsRegistry:
"""
Return the :class:`trinity.components.builtin.metrics.registry.HostMetricsRegistry` at which
metrics instruments can be registered and retrieved.
"""
return self._registry
async def run(self) -> None:
self.logger.info("Reporting metrics to %s", self._influx_server)
self.manager.run_daemon_task(self._continuously_report)
await self.manager.wait_finished()
async def _continuously_report(self) -> None:
async for _ in trio_utils.every(self._reporting_frequency):
self._reporter.report_now()
def main(args=None):
if args is None:
args = sys.argv[1:]
opts = parse_args(args)
verbose_level = opts.verbose
# Determine if any args which support delimited lists should be
# modified
if opts.peers:
opts.peers = _split_comma_append_args(opts.peers)
if opts.seeds:
opts.seeds = _split_comma_append_args(opts.seeds)
init_console_logging(verbose_level=verbose_level)
if opts.network_auth:
opts.network_auth = {"network": opts.network_auth}
try:
path_config = load_path_config(config_dir=opts.config_dir)
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
try:
opts_config = create_validator_config(opts)
validator_config = \
load_validator_config(opts_config, path_config.config_dir)
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
# Process initial initialization errors, delaying the sys.exit(1) until
# all errors have been reported to the user (via LOGGER.error()). This
# is intended to provide enough information to the user so they can correct
# multiple errors before restarting the validator.
init_errors = False
try:
identity_signer = load_identity_signer(key_dir=path_config.key_dir,
key_name='validator')
except LocalConfigurationError as e:
log_configuration(log_dir=path_config.log_dir, name="validator")
LOGGER.error(str(e))
init_errors = True
log_config = get_log_config()
if not init_errors:
if log_config is not None:
log_configuration(log_config=log_config)
if log_config.get('root') is not None:
init_console_logging(verbose_level=verbose_level)
else:
log_configuration(log_dir=path_config.log_dir, name="validator")
try:
version = pkg_resources.get_distribution(DISTRIBUTION_NAME).version
except pkg_resources.DistributionNotFound:
version = 'UNKNOWN'
LOGGER.info('%s (Hyperledger Sawtooth) version %s', DISTRIBUTION_NAME,
version)
if LOGGER.isEnabledFor(logging.INFO):
LOGGER.info('; '.join([
'config [path]: {}'.format(line)
for line in path_config.to_toml_string()
]))
if not check_directory(path=path_config.data_dir,
human_readable_name='Data'):
init_errors = True
if not check_directory(path=path_config.log_dir,
human_readable_name='Log'):
init_errors = True
endpoint = validator_config.endpoint
if endpoint is None:
# Need to use join here to get the string "0.0.0.0". Otherwise,
# bandit thinks we are binding to all interfaces and returns a
# Medium security risk.
interfaces = ["*", ".".join(["0", "0", "0", "0"])]
interfaces += netifaces.interfaces()
endpoint = validator_config.bind_network
for interface in interfaces:
if interface in validator_config.bind_network:
LOGGER.error("Endpoint must be set when using %s", interface)
init_errors = True
break
if init_errors:
LOGGER.error("Initialization errors occurred (see previous log "
"ERROR messages), shutting down.")
sys.exit(1)
bind_network = validator_config.bind_network
bind_component = validator_config.bind_component
if "tcp://" not in bind_network:
bind_network = "tcp://" + bind_network
if "tcp://" not in bind_component:
bind_component = "tcp://" + bind_component
if validator_config.network_public_key is None or \
validator_config.network_private_key is None:
LOGGER.warning("Network key pair is not configured, Network "
"communications between validators will not be "
"authenticated or encrypted.")
wrapped_registry = None
metrics_reporter = None
if validator_config.opentsdb_url:
LOGGER.info("Adding metrics reporter: url=%s, db=%s",
validator_config.opentsdb_url,
validator_config.opentsdb_db)
url = urlparse(validator_config.opentsdb_url)
proto, db_server, db_port, = url.scheme, url.hostname, url.port
registry = MetricsRegistry()
wrapped_registry = MetricsRegistryWrapper(registry)
metrics_reporter = InfluxReporter(
registry=registry,
reporting_interval=10,
database=validator_config.opentsdb_db,
prefix="sawtooth_validator",
port=db_port,
protocol=proto,
server=db_server,
username=validator_config.opentsdb_username,
password=validator_config.opentsdb_password)
metrics_reporter.start()
# Verify state integrity before startup
verify_state(bind_network, bind_component, validator_config.scheduler,
path_config.data_dir)
LOGGER.info('Starting validator with %s scheduler',
validator_config.scheduler)
validator = Validator(bind_network,
bind_component,
endpoint,
validator_config.peering,
validator_config.seeds,
validator_config.peers,
path_config.data_dir,
path_config.config_dir,
identity_signer,
validator_config.scheduler,
validator_config.permissions,
validator_config.minimum_peer_connectivity,
validator_config.maximum_peer_connectivity,
validator_config.network_public_key,
validator_config.network_private_key,
roles=validator_config.roles,
metrics_registry=wrapped_registry)
# pylint: disable=broad-except
try:
validator.start()
except KeyboardInterrupt:
LOGGER.info("Initiating graceful "
"shutdown (press Ctrl+C again to force)")
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
except GenesisError as genesis_err:
LOGGER.error(str(genesis_err))
sys.exit(1)
except Exception as e:
LOGGER.exception(e)
sys.exit(1)
finally:
if metrics_reporter:
metrics_reporter.stop()
validator.stop()
def main(args=None):
if args is None:
args = sys.argv[1:]
opts = parse_args(args)
verbose_level = opts.verbose
# Determine if any args which support delimited lists should be
# modified
if opts.peers:
opts.peers = _split_comma_append_args(opts.peers)
if opts.seeds:
opts.seeds = _split_comma_append_args(opts.seeds)
init_console_logging(verbose_level=verbose_level)
if opts.network_auth:
opts.network_auth = {"network": opts.network_auth}
try:
path_config = load_path_config(config_dir=opts.config_dir)
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
try:
opts_config = create_validator_config(opts)
validator_config = \
load_validator_config(opts_config, path_config.config_dir)
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
# Process initial initialization errors, delaying the sys.exit(1) until
# all errors have been reported to the user (via LOGGER.error()). This
# is intended to provide enough information to the user so they can correct
# multiple errors before restarting the validator.
init_errors = False
try:
identity_signer = load_identity_signer(
key_dir=path_config.key_dir,
key_name='validator')
except LocalConfigurationError as e:
log_configuration(log_dir=path_config.log_dir,
name="validator")
LOGGER.error(str(e))
init_errors = True
log_config = get_log_config()
if not init_errors:
if log_config is not None:
log_configuration(log_config=log_config)
if log_config.get('root') is not None:
init_console_logging(verbose_level=verbose_level)
else:
log_configuration(log_dir=path_config.log_dir,
name="validator")
try:
version = pkg_resources.get_distribution(DISTRIBUTION_NAME).version
except pkg_resources.DistributionNotFound:
version = 'UNKNOWN'
LOGGER.info(
'%s (Hyperledger Sawtooth) version %s', DISTRIBUTION_NAME, version)
if LOGGER.isEnabledFor(logging.INFO):
LOGGER.info(
'; '.join([
'config [path]: {}'.format(line)
for line in path_config.to_toml_string()
])
)
if not check_directory(path=path_config.data_dir,
human_readable_name='Data'):
init_errors = True
if not check_directory(path=path_config.log_dir,
human_readable_name='Log'):
init_errors = True
endpoint = validator_config.endpoint
if endpoint is None:
# Need to use join here to get the string "0.0.0.0". Otherwise,
# bandit thinks we are binding to all interfaces and returns a
# Medium security risk.
interfaces = ["*", ".".join(["0", "0", "0", "0"])]
interfaces += netifaces.interfaces()
endpoint = validator_config.bind_network
for interface in interfaces:
if interface in validator_config.bind_network:
LOGGER.error("Endpoint must be set when using %s", interface)
init_errors = True
break
if init_errors:
LOGGER.error("Initialization errors occurred (see previous log "
"ERROR messages), shutting down.")
sys.exit(1)
bind_network = validator_config.bind_network
bind_component = validator_config.bind_component
if "tcp://" not in bind_network:
bind_network = "tcp://" + bind_network
if "tcp://" not in bind_component:
bind_component = "tcp://" + bind_component
if validator_config.network_public_key is None or \
validator_config.network_private_key is None:
LOGGER.warning("Network key pair is not configured, Network "
"communications between validators will not be "
"authenticated or encrypted.")
wrapped_registry = None
metrics_reporter = None
if validator_config.opentsdb_url:
LOGGER.info("Adding metrics reporter: url=%s, db=%s",
validator_config.opentsdb_url,
validator_config.opentsdb_db)
url = urlparse(validator_config.opentsdb_url)
proto, db_server, db_port, = url.scheme, url.hostname, url.port
registry = MetricsRegistry()
wrapped_registry = MetricsRegistryWrapper(registry)
metrics_reporter = InfluxReporter(
registry=registry,
reporting_interval=10,
database=validator_config.opentsdb_db,
prefix="sawtooth_validator",
port=db_port,
protocol=proto,
server=db_server,
username=validator_config.opentsdb_username,
password=validator_config.opentsdb_password)
metrics_reporter.start()
LOGGER.info(
'Starting validator with %s scheduler',
validator_config.scheduler)
validator = Validator(
bind_network,
bind_component,
endpoint,
validator_config.peering,
validator_config.seeds,
validator_config.peers,
path_config.data_dir,
path_config.config_dir,
identity_signer,
validator_config.scheduler,
validator_config.permissions,
validator_config.minimum_peer_connectivity,
validator_config.maximum_peer_connectivity,
validator_config.network_public_key,
validator_config.network_private_key,
roles=validator_config.roles,
metrics_registry=wrapped_registry)
# pylint: disable=broad-except
try:
validator.start()
except KeyboardInterrupt:
LOGGER.info("Initiating graceful "
"shutdown (press Ctrl+C again to force)")
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
except GenesisError as genesis_err:
LOGGER.error(str(genesis_err))
sys.exit(1)
except Exception as e:
LOGGER.exception(e)
sys.exit(1)
finally:
if metrics_reporter:
metrics_reporter.stop()
validator.stop()
def main():
loop = ZMQEventLoop()
asyncio.set_event_loop(loop)
connection = None
try:
opts = parse_args(sys.argv[1:])
opts_config = RestApiConfig(bind=opts.bind,
connect=opts.connect,
timeout=opts.timeout,
opentsdb_url=opts.opentsdb_url,
opentsdb_db=opts.opentsdb_db)
rest_api_config = load_rest_api_config(opts_config)
url = None
if "tcp://" not in rest_api_config.connect:
url = "tcp://" + rest_api_config.connect
else:
url = rest_api_config.connect
connection = Connection(url)
log_config = get_log_config(filename="rest_api_log_config.toml")
# If no toml, try loading yaml
if log_config is None:
log_config = get_log_config(filename="rest_api_log_config.yaml")
if log_config is not None:
log_configuration(log_config=log_config)
else:
log_dir = get_log_dir()
log_configuration(log_dir=log_dir, name="rest_api")
init_console_logging(verbose_level=opts.verbose)
try:
host, port = rest_api_config.bind[0].split(":")
port = int(port)
except ValueError as e:
print("Unable to parse binding {}: Must be in the format"
" host:port".format(rest_api_config.bind[0]))
sys.exit(1)
wrapped_registry = None
if rest_api_config.opentsdb_url:
LOGGER.info("Adding metrics reporter: url=%s, db=%s",
rest_api_config.opentsdb_url,
rest_api_config.opentsdb_db)
url = urlparse(rest_api_config.opentsdb_url)
proto, db_server, db_port, = url.scheme, url.hostname, url.port
registry = MetricsRegistry()
wrapped_registry = MetricsRegistryWrapper(registry)
reporter = InfluxReporter(
registry=registry,
reporting_interval=10,
database=rest_api_config.opentsdb_db,
prefix="sawtooth_rest_api",
port=db_port,
protocol=proto,
server=db_server,
username=rest_api_config.opentsdb_username,
password=rest_api_config.opentsdb_password)
reporter.start()
start_rest_api(host, port, connection, int(rest_api_config.timeout),
wrapped_registry)
# pylint: disable=broad-except
except Exception as e:
LOGGER.exception(e)
sys.exit(1)
finally:
if connection is not None:
connection.close()
def main(args):
try:
path_config = load_path_config(config_dir=args['config_dir'])
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
try:
opts_config = ValidatorConfig(
bind_component=args['bind_component'],
bind_network=args['bind_network'],
bind_consensus=args['bind_consensus'],
endpoint=args['endpoint'],
maximum_peer_connectivity=args['maximum_peer_connectivity'],
minimum_peer_connectivity=args['minimum_peer_connectivity'],
roles=args['roles'],
opentsdb_db=args['opentsdb_db'],
opentsdb_url=args['opentsdb_url'],
peering=args['peering'],
peers=args['peers'],
scheduler=args['scheduler'],
seeds=args['seeds'],
state_pruning_block_depth=args['state_pruning_block_depth'],
fork_cache_keep_time=args['fork_cache_keep_time'],
)
validator_config = \
load_validator_config(opts_config, path_config.config_dir)
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
try:
log_configuration(log_dir=path_config.log_dir, name="validator")
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
# Process initial initialization errors, delaying the sys.exit(1) until
# all errors have been reported to the user (via LOGGER.error()). This
# is intended to provide enough information to the user so they can correct
# multiple errors before restarting the validator.
init_errors = False
try:
identity_signer = load_identity_signer(key_dir=path_config.key_dir,
key_name='validator')
except LocalConfigurationError as e:
LOGGER.error(str(e))
init_errors = True
log_config = get_log_config()
if not init_errors:
if log_config is not None:
log_configuration(log_config=log_config)
if log_config.get('root') is not None:
init_console_logging(verbose_level=args['verbose'])
else:
log_configuration(log_dir=path_config.log_dir, name="validator")
try:
version = pkg_resources.get_distribution(DISTRIBUTION_NAME).version
except pkg_resources.DistributionNotFound:
version = 'UNKNOWN'
LOGGER.info('%s (Hyperledger Sawtooth) version %s', DISTRIBUTION_NAME,
version)
if LOGGER.isEnabledFor(logging.INFO):
LOGGER.info('; '.join([
'config [path]: {}'.format(line)
for line in path_config.to_toml_string()
]))
if not check_directory(path=path_config.data_dir,
human_readable_name='Data'):
init_errors = True
if not check_directory(path=path_config.log_dir,
human_readable_name='Log'):
init_errors = True
endpoint = validator_config.endpoint
if endpoint is None:
# Need to use join here to get the string "0.0.0.0". Otherwise,
# bandit thinks we are binding to all interfaces and returns a
# Medium security risk.
interfaces = ["*", ".".join(["0", "0", "0", "0"])]
interfaces += netifaces.interfaces()
endpoint = validator_config.bind_network
parsed_endpoint = urlparse(validator_config.bind_network)
for interface in interfaces:
if interface == parsed_endpoint.hostname:
LOGGER.error("Endpoint must be set when using %s", interface)
init_errors = True
if init_errors:
LOGGER.error("Initialization errors occurred (see previous log "
"ERROR messages), shutting down.")
sys.exit(1)
bind_network = validator_config.bind_network
bind_component = validator_config.bind_component
bind_consensus = validator_config.bind_consensus
if "tcp://" not in bind_network:
bind_network = "tcp://" + bind_network
if "tcp://" not in bind_component:
bind_component = "tcp://" + bind_component
if bind_consensus and "tcp://" not in bind_consensus:
bind_consensus = "tcp://" + bind_consensus
if validator_config.network_public_key is None or \
validator_config.network_private_key is None:
LOGGER.warning("Network key pair is not configured, Network "
"communications between validators will not be "
"authenticated or encrypted.")
metrics_reporter = None
if validator_config.opentsdb_url:
LOGGER.info("Adding metrics reporter: url=%s, db=%s",
validator_config.opentsdb_url,
validator_config.opentsdb_db)
url = urlparse(validator_config.opentsdb_url)
proto, db_server, db_port, = url.scheme, url.hostname, url.port
registry = MetricsRegistry()
metrics.init_metrics(registry=registry)
metrics_reporter = InfluxReporter(
registry=registry,
reporting_interval=10,
database=validator_config.opentsdb_db,
prefix="sawtooth_validator",
port=db_port,
protocol=proto,
server=db_server,
username=validator_config.opentsdb_username,
password=validator_config.opentsdb_password)
metrics_reporter.start()
else:
metrics.init_metrics()
# Verify state integrity before startup
global_state_db, blockstore = state_verifier.get_databases(
bind_network, path_config.data_dir)
state_verifier.verify_state(global_state_db, blockstore, bind_component,
validator_config.scheduler)
# Explicitly drop this, so there are not two db instances
global_state_db.drop()
global_state_db = None
LOGGER.info('Starting validator with %s scheduler',
validator_config.scheduler)
component_workers = validator_config.component_thread_pool_workers
network_workers = validator_config.network_thread_pool_workers
sig_workers = validator_config.signature_thread_pool_workers
validator = Validator(bind_network,
bind_component,
bind_consensus,
endpoint,
validator_config.peering,
validator_config.seeds,
validator_config.peers,
path_config.data_dir,
path_config.config_dir,
identity_signer,
path_config.key_dir,
validator_config.scheduler,
validator_config.permissions,
validator_config.minimum_peer_connectivity,
validator_config.maximum_peer_connectivity,
validator_config.state_pruning_block_depth,
validator_config.fork_cache_keep_time,
validator_config.network_public_key,
validator_config.network_private_key,
roles=validator_config.roles,
component_thread_pool_workers=component_workers,
network_thread_pool_workers=network_workers,
signature_thread_pool_workers=sig_workers)
# pylint: disable=broad-except
try:
validator.start()
except KeyboardInterrupt:
LOGGER.info("Initiating graceful "
"shutdown (press Ctrl+C again to force)")
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
except GenesisError as genesis_err:
LOGGER.error(str(genesis_err))
sys.exit(1)
except Exception as e:
LOGGER.exception(e)
sys.exit(1)
finally:
if metrics_reporter:
metrics_reporter.stop()
validator.stop()
def main(args):
try:
path_config = load_path_config(config_dir=args['config_dir'])
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
try:
opts_config = ValidatorConfig(
bind_component=args['bind_component'],
bind_network=args['bind_network'],
bind_consensus=args['bind_consensus'],
endpoint=args['endpoint'],
maximum_peer_connectivity=args['maximum_peer_connectivity'],
minimum_peer_connectivity=args['minimum_peer_connectivity'],
roles=args['roles'],
opentsdb_db=args['opentsdb_db'],
opentsdb_url=args['opentsdb_url'],
peering=args['peering'],
peers=args['peers'],
scheduler=args['scheduler'],
seeds=args['seeds'],
state_pruning_block_depth=args['state_pruning_block_depth'],
fork_cache_keep_time=args['fork_cache_keep_time'],
)
validator_config = \
load_validator_config(opts_config, path_config.config_dir)
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
try:
log_configuration(log_dir=path_config.log_dir,
name="validator")
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
# Process initial initialization errors, delaying the sys.exit(1) until
# all errors have been reported to the user (via LOGGER.error()). This
# is intended to provide enough information to the user so they can correct
# multiple errors before restarting the validator.
init_errors = False
try:
identity_signer = load_identity_signer(
key_dir=path_config.key_dir,
key_name='validator')
except LocalConfigurationError as e:
LOGGER.error(str(e))
init_errors = True
log_config = get_log_config()
if not init_errors:
if log_config is not None:
log_configuration(log_config=log_config)
if log_config.get('root') is not None:
init_console_logging(verbose_level=args['verbose'])
else:
log_configuration(log_dir=path_config.log_dir,
name="validator")
try:
version = pkg_resources.get_distribution(DISTRIBUTION_NAME).version
except pkg_resources.DistributionNotFound:
version = 'UNKNOWN'
LOGGER.info(
'%s (Hyperledger Sawtooth) version %s', DISTRIBUTION_NAME, version)
if LOGGER.isEnabledFor(logging.INFO):
LOGGER.info(
'; '.join([
'config [path]: {}'.format(line)
for line in path_config.to_toml_string()
])
)
if not check_directory(path=path_config.data_dir,
human_readable_name='Data'):
init_errors = True
if not check_directory(path=path_config.log_dir,
human_readable_name='Log'):
init_errors = True
endpoint = validator_config.endpoint
if endpoint is None:
# Need to use join here to get the string "0.0.0.0". Otherwise,
# bandit thinks we are binding to all interfaces and returns a
# Medium security risk.
interfaces = ["*", ".".join(["0", "0", "0", "0"])]
interfaces += netifaces.interfaces()
endpoint = validator_config.bind_network
parsed_endpoint = urlparse(validator_config.bind_network)
for interface in interfaces:
if interface == parsed_endpoint.hostname:
LOGGER.error("Endpoint must be set when using %s", interface)
init_errors = True
if init_errors:
LOGGER.error("Initialization errors occurred (see previous log "
"ERROR messages), shutting down.")
sys.exit(1)
bind_network = validator_config.bind_network
bind_component = validator_config.bind_component
bind_consensus = validator_config.bind_consensus
if "tcp://" not in bind_network:
bind_network = "tcp://" + bind_network
if "tcp://" not in bind_component:
bind_component = "tcp://" + bind_component
if bind_consensus and "tcp://" not in bind_consensus:
bind_consensus = "tcp://" + bind_consensus
if validator_config.network_public_key is None or \
validator_config.network_private_key is None:
LOGGER.warning("Network key pair is not configured, Network "
"communications between validators will not be "
"authenticated or encrypted.")
metrics_reporter = None
if validator_config.opentsdb_url:
LOGGER.info("Adding metrics reporter: url=%s, db=%s",
validator_config.opentsdb_url,
validator_config.opentsdb_db)
url = urlparse(validator_config.opentsdb_url)
proto, db_server, db_port, = url.scheme, url.hostname, url.port
registry = MetricsRegistry()
metrics.init_metrics(registry=registry)
metrics_reporter = InfluxReporter(
registry=registry,
reporting_interval=10,
database=validator_config.opentsdb_db,
prefix="sawtooth_validator",
port=db_port,
protocol=proto,
server=db_server,
username=validator_config.opentsdb_username,
password=validator_config.opentsdb_password)
metrics_reporter.start()
else:
metrics.init_metrics()
# Verify state integrity before startup
global_state_db, blockstore = state_verifier.get_databases(
bind_network,
path_config.data_dir)
state_verifier.verify_state(
global_state_db,
blockstore,
bind_component,
validator_config.scheduler)
# Explicitly drop this, so there are not two db instances
global_state_db.drop()
global_state_db = None
LOGGER.info(
'Starting validator with %s scheduler',
validator_config.scheduler)
component_workers = validator_config.component_thread_pool_workers
network_workers = validator_config.network_thread_pool_workers
sig_workers = validator_config.signature_thread_pool_workers
validator = Validator(
bind_network,
bind_component,
bind_consensus,
endpoint,
validator_config.peering,
validator_config.seeds,
validator_config.peers,
path_config.data_dir,
path_config.config_dir,
identity_signer,
validator_config.scheduler,
validator_config.permissions,
validator_config.minimum_peer_connectivity,
validator_config.maximum_peer_connectivity,
validator_config.state_pruning_block_depth,
validator_config.fork_cache_keep_time,
validator_config.network_public_key,
validator_config.network_private_key,
roles=validator_config.roles,
component_thread_pool_workers=component_workers,
network_thread_pool_workers=network_workers,
signature_thread_pool_workers=sig_workers)
# pylint: disable=broad-except
try:
validator.start()
except KeyboardInterrupt:
LOGGER.info("Initiating graceful "
"shutdown (press Ctrl+C again to force)")
except LocalConfigurationError as local_config_err:
LOGGER.error(str(local_config_err))
sys.exit(1)
except GenesisError as genesis_err:
LOGGER.error(str(genesis_err))
sys.exit(1)
except Exception as e:
LOGGER.exception(e)
sys.exit(1)
finally:
if metrics_reporter:
metrics_reporter.stop()
validator.stop()
def main():
loop = ZMQEventLoop()
asyncio.set_event_loop(loop)
connection = None
try:
opts = parse_args(sys.argv[1:])
opts_config = RestApiConfig(
bind=opts.bind,
connect=opts.connect,
timeout=opts.timeout,
opentsdb_url=opts.opentsdb_url,
opentsdb_db=opts.opentsdb_db)
rest_api_config = load_rest_api_config(opts_config)
url = None
if "tcp://" not in rest_api_config.connect:
url = "tcp://" + rest_api_config.connect
else:
url = rest_api_config.connect
connection = Connection(url)
log_config = get_log_config(filename="rest_api_log_config.toml")
# If no toml, try loading yaml
if log_config is None:
log_config = get_log_config(filename="rest_api_log_config.yaml")
if log_config is not None:
log_configuration(log_config=log_config)
else:
log_dir = get_log_dir()
log_configuration(log_dir=log_dir, name="rest_api")
init_console_logging(verbose_level=opts.verbose)
try:
host, port = rest_api_config.bind[0].split(":")
port = int(port)
except ValueError as e:
print("Unable to parse binding {}: Must be in the format"
" host:port".format(rest_api_config.bind[0]))
sys.exit(1)
wrapped_registry = None
if rest_api_config.opentsdb_url:
LOGGER.info("Adding metrics reporter: url=%s, db=%s",
rest_api_config.opentsdb_url,
rest_api_config.opentsdb_db)
url = urlparse(rest_api_config.opentsdb_url)
proto, db_server, db_port, = url.scheme, url.hostname, url.port
registry = MetricsRegistry()
wrapped_registry = MetricsRegistryWrapper(registry)
reporter = InfluxReporter(
registry=registry,
reporting_interval=10,
database=rest_api_config.opentsdb_db,
prefix="sawtooth_rest_api",
port=db_port,
protocol=proto,
server=db_server,
username=rest_api_config.opentsdb_username,
password=rest_api_config.opentsdb_password)
reporter.start()
start_rest_api(
host,
port,
connection,
int(rest_api_config.timeout),
wrapped_registry)
# pylint: disable=broad-except
except Exception as e:
LOGGER.exception(e)
sys.exit(1)
finally:
if connection is not None:
connection.close()
class BaseMetricsService(Service, MetricsServiceAPI):
"""
A service to provide a registry where metrics instruments can be registered and retrieved from.
It continuously reports metrics to the specified InfluxDB instance.
"""
MIN_SECONDS_BETWEEN_ERROR_LOGS = 60
def __init__(self, influx_server: str, influx_user: str,
influx_password: str, influx_database: str, host: str,
port: int, protocol: str, reporting_frequency: int):
self._unreported_error: Exception = None
self._last_time_reported: float = 0.0
self._influx_server = influx_server
self._reporting_frequency = reporting_frequency
self._registry = HostMetricsRegistry(host)
self._reporter = InfluxReporter(registry=self._registry,
database=influx_database,
username=influx_user,
password=influx_password,
protocol=protocol,
port=port,
server=influx_server)
logger = get_logger('trinity.components.builtin.metrics.MetricsService')
@property
def registry(self) -> HostMetricsRegistry:
"""
Return the :class:`trinity.components.builtin.metrics.registry.HostMetricsRegistry` at which
metrics instruments can be registered and retrieved.
"""
return self._registry
async def run(self) -> None:
self.logger.info("Reporting metrics to %s", self._influx_server)
self.manager.run_daemon_task(self.continuously_report)
await self.manager.wait_finished()
def report_now(self) -> None:
try:
self._reporter.report_now()
except (HTTPException, ConnectionError) as exc:
# This method is usually called every few seconds. If there's an issue with the
# connection we do not want to flood the log and tame down warnings.
# 1. We log the first instance of an exception immediately
# 2. We log follow up exceptions only after a minimum time has elapsed
# This means that we also might overwrite exceptions for different errors
if self._is_justified_to_log_error():
self._log_and_clear(exc)
else:
self._unreported_error = exc
else:
# If errors disappear, we want to make sure we eventually report the last instance
if self._unreported_error is not None and self._is_justified_to_log_error(
):
self._log_and_clear(self._unreported_error)
def _log_and_clear(self, error: Exception) -> None:
self.logger.warning("Unable to report metrics: %s", error)
self._unreported_error = None
self._last_time_reported = time.monotonic()
def _is_justified_to_log_error(self) -> bool:
return (self._last_time_reported == 0.0
or time.monotonic() - self._last_time_reported >
self.MIN_SECONDS_BETWEEN_ERROR_LOGS)
@abstractmethod
async def continuously_report(self) -> None:
...
class Metrics(Service):
logger = logging.getLogger('alexandria.metrics.Metrics')
def __init__(self,
host: str,
client: ClientAPI,
kademlia: KademliaAPI,
influx_server: str,
influx_user: str,
influx_password: str,
influx_database: str,
influx_port: int = 443,
influx_protocol: str = 'https',
reporting_frequency: int = 10,
process_collection_frequency: int = 3):
self._influx_server = influx_server
self._reporting_frequency = reporting_frequency
self._process_collection_frequency = process_collection_frequency
self._registry = HostMetricsRegistry(host)
self._reporter = InfluxReporter(
registry=self._registry,
protocol=influx_protocol,
port=influx_port,
database=influx_database,
username=influx_user,
password=influx_password,
server=influx_server
)
self.client = client
self.kademlia = kademlia
@classmethod
def from_cli_args(cls,
args: Namespace,
client: ClientAPI,
kademlia: KademliaAPI,
) -> 'Metrics':
return cls(
host=args.metrics_host,
client=client,
kademlia=kademlia,
influx_server=args.metrics_influx_server,
influx_user=args.metrics_influx_user,
influx_password=args.metrics_influx_password,
influx_database=args.metrics_influx_database,
influx_port=args.metrics_influx_port,
influx_protocol=args.metrics_influx_protocol,
reporting_frequency=args.metrics_reporting_frequency,
)
async def run(self) -> None:
self.manager.run_daemon_task(
self._continuously_report,
self._reporting_frequency,
)
self.manager.run_daemon_task(
self._collect_system_metrics,
self._process_collection_frequency,
)
self.manager.run_daemon_task(
self._report_routing_table_stats,
10,
)
self.manager.run_daemon_task(
self._report_content_manager_stats,
10,
)
self.logger.info('Metrics started')
for payload_type in PAYLOAD_TYPES:
self.manager.run_daemon_task(self._report_inbound_message_stats, payload_type)
self.manager.run_daemon_task(self._report_event, self.client.events.session_created, 'events/session-created') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.session_idle, 'events/session-idle') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.handshake_complete, 'events/handshake-complete') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.handshake_timeout, 'events/handshake-timeout') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.datagram_received, 'datagram/inbound') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.datagram_sent, 'datagram/outbound') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.sent_ping, 'messages/outbound/Ping') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.sent_pong, 'messages/outbound/Pong') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.sent_find_nodes, 'messages/outbound/FindNodes') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.sent_found_nodes, 'messages/outbound/FoundNodes') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.sent_advertise, 'messages/outbound/Advertise') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.sent_ack, 'messages/outbound/Ack') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.sent_locate, 'messages/outbound/Locate') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.sent_locations, 'messages/outbound/Locations') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.sent_retrieve, 'messages/outbound/Retrieve') # noqa: E501
self.manager.run_daemon_task(self._report_event, self.client.events.sent_chunk, 'messages/outbound/Chunk') # noqa: E501
await self.manager.wait_finished()
async def _continuously_report(self, frequency: int) -> None:
async for _ in every(frequency):
self._reporter.report_now()
async def _report_event(self, event: EventAPI[Any], suffix: str) -> None:
counter = self._registry.counter(f'alexandria.{suffix}.counter')
meter = self._registry.meter(f'alexandria.{suffix}.meter')
async with event.subscribe() as subscription:
async for _ in subscription:
counter.inc()
meter.mark()
async def _report_routing_table_stats(self, frequency: int) -> None:
size_gauge = self._registry.gauge('alexandria.dht/routing-table/total-nodes.gauge')
async for _ in every(frequency):
stats = self.kademlia.routing_table.get_stats()
size_gauge.set_value(stats.total_nodes)
async def _report_inbound_message_stats(self, payload_type: Type[ssz.Serializable]) -> None:
name = payload_type.__name__
counter = self._registry.counter(f'alexandria.messages/inbound/{name}.counter')
meter = self._registry.meter(f'alexandria.messages/inbound/{name}.meter')
async with self.client.message_dispatcher.subscribe(payload_type) as subscription:
async for payload in subscription:
counter.inc()
meter.mark()
async def _report_content_manager_stats(self, frequency: int) -> None:
gauge = self._registry.gauge
durable_db_item_count_gauge = gauge('alexandria.content/durable-db/item-count.gauge')
ephemeral_db_item_count_gauge = gauge('alexandria.content/ephemeral-db/item-count.gauge')
ephemeral_db_capacity_gauge = gauge('alexandria.content/ephemeral-db/capacity.gauge')
ephemeral_db_size_gauge = gauge('alexandria.content/ephemeral-db/size.gauge')
ephemeral_index_capacity_gauge = gauge('alexandria.content/ephemeral-index/capacity.gauge')
ephemeral_index_size_gauge = gauge('alexandria.content/ephemeral-index/size.gauge')
cache_db_item_count_gauge = gauge('alexandria.content/cache-db/item-count.gauge')
cache_db_capacity_gauge = gauge('alexandria.content/cache-db/capacity.gauge')
cache_db_size_gauge = gauge('alexandria.content/cache-db/size.gauge')
cache_index_capacity_gauge = gauge('alexandria.content/cache-index/capacity.gauge')
cache_index_size_gauge = gauge('alexandria.content/cache-index/size.gauge')
async for _ in every(frequency):
stats = self.kademlia.content_manager.get_stats()
durable_db_item_count_gauge.set_value(stats.durable_item_count)
ephemeral_db_item_count_gauge.set_value(stats.ephemeral_db_count)
ephemeral_db_capacity_gauge.set_value(stats.ephemeral_db_capacity)
ephemeral_db_size_gauge.set_value(
stats.ephemeral_db_total_capacity - stats.ephemeral_db_capacity
)
ephemeral_index_capacity_gauge.set_value(stats.ephemeral_index_capacity)
ephemeral_index_size_gauge.set_value(
stats.ephemeral_index_total_capacity - stats.ephemeral_index_capacity
)
cache_db_item_count_gauge.set_value(stats.cache_db_count)
cache_db_capacity_gauge.set_value(stats.cache_db_capacity)
cache_db_size_gauge.set_value(stats.cache_db_total_capacity - stats.cache_db_capacity)
cache_index_capacity_gauge.set_value(stats.cache_index_capacity)
cache_index_size_gauge.set_value(
stats.cache_index_total_capacity - stats.cache_index_capacity
)
async def _collect_system_metrics(self, frequency: int) -> None:
cpu_sysload_gauge = self._registry.gauge('alexandria.system/cpu/sysload.gauge')
cpu_syswait_gauge = self._registry.gauge('alexandria.system/cpu/syswait.gauge')
memory_used_gauge = self._registry.gauge('alexandria.system/memory/used.gauge')
memory_free_gauge = self._registry.gauge('alexandria.system/memory/free.gauge')
disk_readdata_meter = self._registry.meter('alexandria.system/disk/readdata.meter')
disk_writedata_meter = self._registry.meter('alexandria.system/disk/writedata.meter')
network_in_packets_meter = self._registry.meter('alexandria.network/in/packets/total.meter')
network_out_packets_meter = self._registry.meter('alexandria.network/out/packets/total.meter') # noqa: E501
previous = read_system_stats()
async for _ in every(frequency, initial_delay=frequency):
current = read_system_stats()
global_time = current.cpu_stats.global_time - previous.cpu_stats.global_time
cpu_sysload_gauge.set_value(global_time / frequency)
global_wait = current.cpu_stats.global_wait_io - previous.cpu_stats.global_wait_io
cpu_syswait_gauge.set_value(global_wait / frequency)
memory_used_gauge.set_value(current.memory_stats.used)
memory_free_gauge.set_value(current.memory_stats.free)
read_bytes = current.disk_stats.read_bytes - previous.disk_stats.read_bytes
disk_readdata_meter.mark(read_bytes)
write_bytes = current.disk_stats.write_bytes - previous.disk_stats.write_bytes
disk_writedata_meter.mark(write_bytes)
in_packets = current.network_stats.in_packets - previous.network_stats.in_packets
network_in_packets_meter.mark(in_packets)
out_packets = current.network_stats.out_packets - previous.network_stats.out_packets
network_out_packets_meter.mark(out_packets)
previous = current
