def _load_json_file(self, key):
""" Load dict obj from json in given absolute file path"""
value = None
absfilepath = key
# Check if directory exists
directory_path = os.path.join(os.path.dirname(absfilepath), "")
if not os.path.isdir(directory_path):
logger.critical("Path doesn't exists: {0}".format(directory_path))
return
try:
fh = open(absfilepath,"rb")
try:
value = pickle.load(fh)
except:
value = fh.read()
except IOError as err:
logger.warn("I/O error[{0}] while loading data from file {1}): {2}"\
.format(err.errno,absfilepath,err))
except ValueError as jsonerr:
logger.warn("JSON error{0} while loading from {1}".format(jsonerr, absfilepath))
value = None
except OSError as oserr:
logger.warn("OS error{0} while loading from {1}".format(oserr, absfilepath))
except Exception as gerr:
logger.warn("Error{0} while reading data from file {1}"\
.format(gerr, absfilepath))
else:
fh.close()
return value
def put(self, value, key, pickled=True):
""" Dump value to given absolute file path"""
absfilepath = key
directory_path = os.path.join(os.path.dirname(absfilepath), "")
# If directory does not exists, create
if not os.path.isdir(directory_path):
try:
os.makedirs(directory_path, exist_ok=True)
except OSError as exc:
if exc.errno == errno.EACCES:
logger.critical(f"Permission denied while creating dir: {directory_path}")
except Exception as err:
logger.warn(f"{directory_path} creation failed with error {err}, alerts \
may get missed on sspl restart or failover!!")
try:
fh = open(absfilepath,"wb")
if pickled:
pickle.dump(value, fh)
else:
fh.write(value)
except IOError as err:
logger.warn("I/O error[{0}] while dumping data to file {1}): {2}"\
.format(err.errno,absfilepath,err))
except Exception as gerr:
logger.warn("Error[{0}] while dumping data to file {1}"\
.format(gerr, absfilepath))
else:
fh.close()
def _run_thread_capture_errors(curr_module, msgQlist, conf_reader, product):
"""Run the given thread and log any errors that happen on it.
Will stop all sspl_modules if one of them fails."""
try:
# Each module is passed a reference list to message queues so it can transmit
# internal messages to other modules as desired
curr_module.initialize(conf_reader, msgQlist, product)
curr_module.start()
except BaseException as ex:
logger.critical(
"SSPL-Tests encountered a fatal error, terminating service Error: %s"
% ex)
logger.exception(ex)
# Populate an actuator response message and transmit back to HAlon
error_msg = "SSPL-Tests encountered an error, terminating service Error: " + \
", Exception: " + logger.exception(ex)
jsonMsg = ThreadControllerMsg(curr_module, error_msg).getJson()
curr_module._write_internal_msgQ(EgressProcessorTests.name(), jsonMsg)
# Shut it down, error is non-recoverable
for name, other_module in list(world.sspl_modules.items()):
if other_module is not curr_module:
other_module.shutdown()
def _parse_fru_info(self, fru):
"""Parses fan information"""
specific_info = None
specifics = []
for sensor_id, fru_info in self.fru_specific_info.items():
specific_info = dict()
for fru_key,fru_value in fru_info.items():
specific_info[fru_key] = fru_value
specific_info["resource_id"] = sensor_id
specifics.append(specific_info)
if (fru == "Power Supply") or (fru == "Fan") or (fru == "Drive Slot / Bay"):
if not specifics:
manufacturer = self._executor.get_manufacturer_name()
msg = "'%s' sensors not seen in %s node server" % (
fru, manufacturer)
specifics.append({"ERROR": msg})
logger.critical(msg)
else:
for each in specifics:
if each.get('States Asserted'):
each['States Asserted'] = ' '.join(
x.strip() for x in each['States Asserted'].split())
self.fru_specific_info = {}
return specifics
def _run_thread_capture_errors(curr_module, sspl_modules, msgQlist,
conf_reader, product):
"""Run the given thread and log any errors that happen on it.
Will stop all sspl_modules if one of them fails."""
try:
# Each module is passed a reference list to message queues so it can transmit
# internal messages to other modules as desired
curr_module.start_thread(conf_reader, msgQlist, product)
except BaseException as ex:
logger.critical(
"SSPL-LL encountered a fatal error, terminating service Error: %s"
% ex)
logger.exception(ex)
# Populate an actuator response message and transmit back to HAlon
error_msg = "SSPL-LL encountered an error, terminating service Error: " + \
", Exception: " + logger.exception(ex)
json_msg = ThreadControllerMsg(curr_module.name(), error_msg).getJson()
if product.lower() in [x.lower() for x in enabled_products]:
self._write_internal_msgQ(RabbitMQegressProcessor.name(), json_msg)
elif product.lower() in [x.lower() for x in cs_legacy_products]:
self._write_internal_msgQ(PlaneCntrlRMQegressProcessor.name(),
json_msg)
# Shut it down, error is non-recoverable
for name, other_module in list(sspl_modules.items()):
if other_module is not curr_module:
other_module.shutdown()
def execute_thread(module, msgQlist, conf_reader, product, resume=True):
"""
Run module as a thread. Recover the module if any error during
initialization and run time of the module.
If recovery count>0,
module will be recovered from failure until the maximum recovery
attempt. If not recoverable, corresponding module will be shutdown
and failure alert will be raised due to its impact.
If recovery count=0,
no recovery attempt will be made.
"""
module_name = module.name()
# Suspend module threads
if resume == False:
module.suspend()
# Initialize persistent cache for sensor status
per_data_path = os.path.join(
module_cache_dir, f"{module_name.upper()}_{node_id}")
if not os.path.isfile(per_data_path):
module_persistent_data[module_name] = {}
store.put(module_persistent_data[module_name], per_data_path)
is_sensor_thread = False
recovery_count = recovery_interval = 0
if isinstance(module, SensorThread):
recovery_count, recovery_interval = _get_recovery_config(module_name)
is_sensor_thread = True
attempt = 0
while attempt <= recovery_count:
attempt += 1
try:
# Each module is passed a reference list to message queues so it
# can transmit internal messages to other modules as desired
module.start_thread(conf_reader, msgQlist, product)
except Exception as err:
curr_state = "fault"
err_msg = f"{module_name}, {err}"
logger.error(err_msg)
if attempt > recovery_count:
logger.debug(traceback.format_exc())
description = f"{module_name} is stopped and unrecoverable. {err_msg}"
impact = module.impact()
recommendation = "Restart SSPL service"
logger.critical(
f"{description}. Impact: {impact} Recommendation: {recommendation}")
# Check previous state of the module and send fault alert
if os.path.isfile(per_data_path):
module_persistent_data[module_name] = store.get(per_data_path)
prev_state = module_persistent_data[module_name].get('prev_state')
if is_sensor_thread and curr_state != prev_state:
module_persistent_data[module_name] = {"prev_state": curr_state}
store.put(module_persistent_data[module_name], per_data_path)
specific_info = Conf.get(SSPL_CONF, f"{module_name.upper()}")
info = {
"module_name": module_name,
"alert_type": curr_state,
"description": description,
"impact": impact,
"recommendation": recommendation,
"severity": "critical",
"specific_info": specific_info
}
jsonMsg = ThreadMonitorMsg(info).getJson()
module._write_internal_msgQ(EgressProcessor.name(), jsonMsg)
else:
logger.debug(f"Recovering {module_name} from failure, "
f"attempt: {attempt}")
time.sleep(recovery_interval)
# Shutdown if no recovery attempt
logger.info(f"Terminating monitoring thread {module_name}")
module.shutdown()
retry = 5
while module.is_running():
module.shutdown()
retry -= 1
if not retry:
break
time.sleep(2)
