def tearDown(self):
self.aminer_config = AMinerConfig.load_config(self.__configFilePath)
persistence_file_name = AMinerConfig.build_persistence_file_name(
self.aminer_config)
if os.path.exists(persistence_file_name):
shutil.rmtree(persistence_file_name)
if not os.path.exists(persistence_file_name):
os.makedirs(persistence_file_name)
def setUp(self):
self.aminer_config = AMinerConfig.load_config(self.__configFilePath)
self.analysis_context = AnalysisContext(self.aminer_config)
self.output_stream = StringIO()
self.stream_printer_event_handler = StreamPrinterEventHandler(
self.analysis_context, self.output_stream)
persistence_file_name = AMinerConfig.build_persistence_file_name(
self.aminer_config)
if os.path.exists(persistence_file_name):
shutil.rmtree(persistence_file_name)
if not os.path.exists(persistence_file_name):
os.makedirs(persistence_file_name)
def __init__(self,
aminerConfig,
propertyPath,
binDefinition,
reportInterval,
reportEventHandlers,
resetAfterReportFlag=True,
persistenceId='Default'):
"""Initialize the analysis component.
@param reportInterval delay in seconds between creation of two
reports. The parameter is applied to the parsed record data
time, not the system time. Hence reports can be delayed when
no data is received."""
self.lastReportTime = None
self.nextReportTime = 0.0
self.propertyPath = propertyPath
self.binDefinition = binDefinition
self.histogramData = {}
self.reportInterval = reportInterval
self.reportEventHandlers = reportEventHandlers
self.resetAfterReportFlag = resetAfterReportFlag
self.persistenceId = persistenceId
self.nextPersistTime = None
PersistencyUtil.addPersistableComponent(self)
self.persistenceFileName = AMinerConfig.buildPersistenceFileName(
aminerConfig, 'PathDependentHistogramAnalysis', persistenceId)
persistenceData = PersistencyUtil.loadJson(self.persistenceFileName)
if persistenceData is not None:
raise Exception('No data reading, def merge yet')
def __init__(self,
aminerConfig,
targetPathList,
anomalyEventHandlers,
persistenceId='Default',
allowMissingValuesFlag=False,
autoIncludeFlag=False,
outputLogLine=True):
"""Initialize the detector. This will also trigger reading
or creation of persistence storage location.
@param targetPathList the list of values to extract from each
match to create the value combination to be checked.
@param allowMissingValuesFlag when set to True, the detector
will also use matches, where one of the pathes from targetPathList
does not refer to an existing parsed data object.
@param autoIncludeFlag when set to True, this detector will
report a new value only the first time before including it
in the known values set automatically."""
self.targetPathList = targetPathList
self.anomalyEventHandlers = anomalyEventHandlers
self.allowMissingValuesFlag = allowMissingValuesFlag
self.autoIncludeFlag = autoIncludeFlag
self.outputLogLine = outputLogLine
self.aminerConfig = aminerConfig
self.persistenceFileName = AMinerConfig.buildPersistenceFileName(
aminerConfig, self.__class__.__name__, persistenceId)
self.nextPersistTime = None
self.loadPersistencyData()
PersistencyUtil.addPersistableComponent(self)
def __init__(self, aminerConfig, parallelCheckCount, correlationTestCount, \
maxFailCount, anomalyEventHandlers, persistenceId='Default', recordCountBeforeEvent=0x10000):
"""Initialize the detector. This will also trigger reading
or creation of persistence storage location.
@param parallelCheckCount number of rule detection checks
to run in parallel.
@param correlationTestCount number of tests to perform on a rule under
test.
@param maxFailCount maximal number of test failures so that
rule is still eligible for reporting."""
self.lastTimestamp = 0.0
self.parallelCheckCount = parallelCheckCount
self.correlationTestCount = correlationTestCount
self.maxFailCount = maxFailCount
self.anomalyEventHandlers = anomalyEventHandlers
self.maxRuleAttributes = 5
self.lastUnhandledMatch = None
self.nextPersistTime = None
self.totalRecords = 0
self.recordCountBeforeEvent = recordCountBeforeEvent
PersistencyUtil.addPersistableComponent(self)
self.persistenceFileName = AMinerConfig.buildPersistenceFileName(
aminerConfig, 'TimeCorrelationDetector', persistenceId)
persistenceData = PersistencyUtil.loadJson(self.persistenceFileName)
if persistenceData is None:
self.featureList = []
self.eventCountTable = [
0
] * parallelCheckCount * parallelCheckCount * 2
self.eventDeltaTable = [
0
] * parallelCheckCount * parallelCheckCount * 2
def __init__(self,
aminer_config,
target_path_list,
anomaly_event_handlers,
persistence_id='Default',
allow_missing_values_flag=False,
auto_include_flag=False,
output_log_line=True):
"""Initialize the detector. This will also trigger reading or creation of persistence storage location.
@param target_path_list the list of values to extract from each match to create the value combination to be checked.
@param allow_missing_values_flag when set to True, the detector will also use matches, where one of the pathes from targetPathList
does not refer to an existing parsed data object.
@param auto_include_flag when set to True, this detector will report a new value only the first time before including it
in the known values set automatically."""
self.target_path_list = target_path_list
self.anomaly_event_handlers = anomaly_event_handlers
self.allow_missing_values_flag = allow_missing_values_flag
self.auto_include_flag = auto_include_flag
self.output_log_line = output_log_line
self.aminer_config = aminer_config
self.persistence_id = persistence_id
self.persistence_file_name = AMinerConfig.build_persistence_file_name(
aminer_config, self.__class__.__name__, persistence_id)
self.next_persist_time = None
self.known_values_set = set()
self.load_persistency_data()
PersistencyUtil.add_persistable_component(self)
def __init__(self,
aminer_config,
target_path_list,
anomaly_event_handlers,
persistence_id='Default',
auto_include_flag=False,
output_log_line=True):
"""Initialize the detector. This will also trigger reading or creation of persistence storage location."""
self.target_path_list = target_path_list
self.anomaly_event_handlers = anomaly_event_handlers
self.auto_include_flag = auto_include_flag
self.next_persist_time = None
self.output_log_line = output_log_line
self.aminer_config = aminer_config
self.persistence_id = persistence_id
self.log_success = 0
self.log_total = 0
self.log_learned_path_values = 0
self.log_new_learned_values = []
PersistenceUtil.add_persistable_component(self)
self.persistence_file_name = AMinerConfig.build_persistence_file_name(
aminer_config, self.__class__.__name__, persistence_id)
persistence_data = PersistenceUtil.load_json(
self.persistence_file_name)
if persistence_data is None:
self.known_values_set = set()
else:
self.known_values_set = set(persistence_data)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).debug(
'%s loaded persistence data.', self.__class__.__name__)
def __init__(self,
aminer_config,
histogram_defs,
report_interval,
report_event_handlers,
reset_after_report_flag=True,
persistence_id='Default',
output_log_line=True):
"""Initialize the analysis component.
@param histogram_defs is a list of tuples containing the target property path to analyze and the BinDefinition to apply for
binning.
@param report_interval delay in seconds between creation of two reports. The parameter is applied to the parsed record data
time, not the system time. Hence reports can be delayed when no data is received."""
self.last_report_time = None
self.next_report_time = 0.0
self.histogram_data = []
for (path, bin_definition) in histogram_defs:
self.histogram_data.append(HistogramData(path, bin_definition))
self.report_interval = report_interval
self.report_event_handlers = report_event_handlers
self.reset_after_report_flag = reset_after_report_flag
self.persistence_id = persistence_id
self.next_persist_time = None
self.output_log_line = output_log_line
PersistencyUtil.add_persistable_component(self)
self.persistenceFileName = AMinerConfig.build_persistence_file_name(
aminer_config, 'HistogramAnalysis', persistence_id)
persistence_data = PersistencyUtil.load_json(self.persistenceFileName)
if persistence_data is not None:
raise Exception('No data reading, def merge yet')
def __init__(self, aminerConfig, anomalyEventHandlers, timestampPath,
analyzePathList, minBinElements, minBinTime, syncBinsFlag=True,
debugMode=False, persistenceId='Default'):
"""Initialize the detector. This will also trigger reading
or creation of persistence storage location.
@param timestampPath if not None, use this path value for
timestamp based bins.
@param analyzePathList list of match pathes to analyze in
this detector.
@param minBinElements evaluate the latest bin only after at
least that number of elements was added to it.
@param minBinTime evaluate the latest bin only when the first
element is received after minBinTime has elapsed.
@param syncBinsFlag if true the bins of all analyzed path values
have to be filled enough to trigger analysis.
@param debugMode if true, generate an analysis report even
when average of last bin was within expected range."""
self.anomalyEventHandlers = anomalyEventHandlers
self.timestampPath = timestampPath
self.minBinElements = minBinElements
self.minBinTime = minBinTime
self.syncBinsFlag = syncBinsFlag
self.debugMode = debugMode
self.nextPersistTime = None
PersistencyUtil.addPersistableComponent(self)
self.persistenceFileName = AMinerConfig.buildPersistenceFileName(aminerConfig, \
'MatchValueAverageChangeDetector', persistenceId)
persistenceData = PersistencyUtil.loadJson(self.persistenceFileName)
if persistenceData is None:
self.statData = []
for path in analyzePathList:
self.statData.append((path, [],))
def __init__(self, aminer_config, parallel_check_count, correlation_test_count, max_fail_count, anomaly_event_handlers,
persistence_id='Default', record_count_before_event=0x10000, output_log_line=True):
"""Initialize the detector. This will also trigger reading or creation of persistence storage location.
@param parallel_check_count number of rule detection checks to run in parallel.
@param correlation_test_count number of unit to perform on a rule under test.
@param max_fail_count maximal number of test failures so that rule is still eligible for reporting."""
self.last_timestamp = 0.0
self.parallel_check_count = parallel_check_count
self.correlation_test_count = correlation_test_count
self.max_fail_count = max_fail_count
self.anomaly_event_handlers = anomaly_event_handlers
self.max_rule_attributes = 5
self.last_unhandled_match = None
self.next_persist_time = None
self.total_records = 0
self.record_count_before_event = record_count_before_event
self.persistence_id = persistence_id
self.output_log_line = output_log_line
PersistencyUtil.add_persistable_component(self)
self.persistence_file_name = AMinerConfig.build_persistence_file_name(aminer_config, 'TimeCorrelationDetector', persistence_id)
persistence_data = PersistencyUtil.load_json(self.persistence_file_name)
if persistence_data is None:
self.feature_list = []
self.event_count_table = [0] * parallel_check_count * parallel_check_count * 2
self.event_delta_table = [0] * parallel_check_count * parallel_check_count * 2
def __init__(self, aminer_config, anomaly_event_handlers, timestamp_path, analyze_path_list, min_bin_elements, min_bin_time,
sync_bins_flag=True, debug_mode=False, persistence_id='Default', output_log_line=True):
"""Initialize the detector. This will also trigger reading or creation of persistence storage location.
@param timestamp_path if not None, use this path value for timestamp based bins.
@param analyze_path_list list of match pathes to analyze in this detector.
@param min_bin_elements evaluate the latest bin only after at least that number of elements was added to it.
@param min_bin_time evaluate the latest bin only when the first element is received after minBinTime has elapsed.
@param sync_bins_flag if true the bins of all analyzed path values have to be filled enough to trigger analysis.
@param debug_mode if true, generate an analysis report even when average of last bin was within expected range."""
self.anomaly_event_handlers = anomaly_event_handlers
self.timestamp_path = timestamp_path
self.min_bin_elements = min_bin_elements
self.min_bin_time = min_bin_time
self.sync_bins_flag = sync_bins_flag
self.debug_mode = debug_mode
self.next_persist_time = None
self.persistence_id = persistence_id
self.output_log_line = output_log_line
PersistencyUtil.add_persistable_component(self)
self.persistence_file_name = AMinerConfig.build_persistence_file_name(aminer_config, 'MatchValueAverageChangeDetector',
persistence_id)
persistence_data = PersistencyUtil.load_json(self.persistence_file_name)
if persistence_data is None:
self.stat_data = []
for path in analyze_path_list:
self.stat_data.append((path, [],))
def __init__(self,
aminer_config,
ruleset,
anomaly_event_handlers,
persistence_id='Default',
output_log_line=True):
"""
Initialize the detector. This will also trigger reading or creation of persistence storage location.
@param ruleset a list of MatchRule rules with appropriate CorrelationRules attached as actions.
"""
self.aminer_config = aminer_config
self.event_classification_ruleset = ruleset
self.anomaly_event_handlers = anomaly_event_handlers
self.next_persist_time = time.time(
) + self.aminer_config.config_properties.get(
AMinerConfig.KEY_PERSISTENCE_PERIOD,
AMinerConfig.DEFAULT_PERSISTENCE_PERIOD)
self.persistence_id = persistence_id
self.output_log_line = output_log_line
self.last_log_atom = None
event_correlation_set = set()
for rule in self.event_classification_ruleset:
if rule.match_action.artefact_a_rules is not None:
event_correlation_set |= set(
rule.match_action.artefact_a_rules)
if rule.match_action.artefact_b_rules is not None:
event_correlation_set |= set(
rule.match_action.artefact_b_rules)
self.event_correlation_ruleset = list(event_correlation_set)
PersistenceUtil.add_persistable_component(self)
self.persistence_file_name = AMinerConfig.build_persistence_file_name(
aminer_config, 'TimeCorrelationViolationDetector', persistence_id)
def __init__(self,
aminer_config,
target_path,
anomaly_event_handlers,
persistence_id='Default',
auto_include_flag=False,
default_interval=3600,
realert_interval=86400,
output_log_line=True):
"""Initialize the detector. This will also trigger reading or creation of persistence storage location.
@param target_path to extract a source identification value from each logatom."""
self.target_path = target_path
self.anomaly_event_handlers = anomaly_event_handlers
self.auto_include_flag = auto_include_flag
self.default_interval = default_interval
self.realert_interval = realert_interval
# This timestamps is compared with timestamp values from log atoms for activation of alerting logic. The first timestamp from logs
# above this value will trigger alerting.
self.next_check_timestamp = 0
self.last_seen_timestamp = 0
self.next_persist_time = None
self.output_log_line = output_log_line
self.aminer_config = aminer_config
self.persistence_id = persistence_id
PersistencyUtil.add_persistable_component(self)
self.persistence_file_name = AMinerConfig.build_persistence_file_name(
aminer_config, self.__class__.__name__, persistence_id)
persistence_data = PersistencyUtil.load_json(
self.persistence_file_name)
if persistence_data is None:
self.expected_values_dict = {}
else:
self.expected_values_dict = persistence_data
self.analysis_string = 'Analysis.%s'
def __init__(self,
aminer_config,
target_path_list,
anomaly_event_handlers,
persistence_id='Default',
auto_include_flag=False,
output_log_line=True):
"""Initialize the detector. This will also trigger reading or creation of persistence storage location."""
self.target_path_list = target_path_list
self.anomaly_event_handlers = anomaly_event_handlers
self.auto_include_flag = auto_include_flag
self.next_persist_time = None
self.output_log_line = output_log_line
self.aminer_config = aminer_config
self.persistence_id = persistence_id
PersistencyUtil.add_persistable_component(self)
self.persistence_file_name = AMinerConfig.build_persistence_file_name(
aminer_config, self.__class__.__name__, persistence_id)
persistence_data = PersistencyUtil.load_json(
self.persistence_file_name)
if persistence_data is None:
self.known_path_set = set()
else:
self.known_path_set = set(persistence_data)
def __init__(self,
aminerConfig,
targetPath,
anomalyEventHandlers,
persistenceId='Default',
autoIncludeFlag=False,
defaultInterval=3600,
realertInterval=86400,
outputLogLine=True):
"""Initialize the detector. This will also trigger reading
or creation of persistence storage location.
@param targetPath to extract a source identification value
from each logatom."""
self.targetPath = targetPath
self.anomalyEventHandlers = anomalyEventHandlers
self.autoIncludeFlag = autoIncludeFlag
self.defaultInterval = defaultInterval
self.realertInterval = realertInterval
# This timestamps is compared with timestamp values from log atoms
# for activation of alerting logic. The first timestamp from logs
# above this value will trigger alerting.
self.nextCheckTimestamp = 0
self.lastSeenTimestamp = 0
self.nextPersistTime = None
self.outputLogLine = outputLogLine
self.aminerConfig = aminerConfig
PersistencyUtil.addPersistableComponent(self)
self.persistenceFileName = AMinerConfig.buildPersistenceFileName(
aminerConfig, self.__class__.__name__, persistenceId)
persistenceData = PersistencyUtil.loadJson(self.persistenceFileName)
if persistenceData is None:
self.expectedValuesDict = {}
else:
self.expectedValuesDict = persistenceData
def __init__(self,
aminer_config,
property_path,
bin_definition,
report_interval,
report_event_handlers,
reset_after_report_flag=True,
persistence_id='Default',
output_log_line=True):
"""Initialize the analysis component.
@param report_interval delay in seconds between creation of two reports. The parameter is applied to the parsed record data
time, not the system time. Hence reports can be delayed when no data is received."""
self.last_report_time = None
self.next_report_time = 0.0
self.property_path = property_path
self.bin_definition = bin_definition
self.histogram_data = {}
self.report_interval = report_interval
self.report_event_handlers = report_event_handlers
self.reset_after_report_flag = reset_after_report_flag
self.persistence_id = persistence_id
self.next_persist_time = None
self.output_log_line = output_log_line
PersistencyUtil.add_persistable_component(self)
self.persistence_file_name = AMinerConfig.build_persistence_file_name(
aminer_config, 'PathDependentHistogramAnalysis', persistence_id)
persistence_data = PersistencyUtil.load_json(
self.persistence_file_name)
if persistence_data is not None:
raise Exception('No data reading, def merge yet')
def __init__(self,
aminerConfig,
histogramDefs,
reportInterval,
reportEventHandlers,
resetAfterReportFlag=True,
persistenceId='Default'):
"""Initialize the analysis component.
@param histogramDefs is a list of tuples containing the target
property path to analyze and the BinDefinition to apply for
binning.
@param reportInterval delay in seconds between creation of two
reports. The parameter is applied to the parsed record data
time, not the system time. Hence reports can be delayed when
no data is received."""
self.lastReportTime = None
self.nextReportTime = 0.0
self.histogramData = []
for (path, binDefinition) in histogramDefs:
self.histogramData.append(HistogramData(path, binDefinition))
self.reportInterval = reportInterval
self.reportEventHandlers = reportEventHandlers
self.resetAfterReportFlag = resetAfterReportFlag
self.persistenceId = persistenceId
self.nextPersistTime = None
PersistencyUtil.addPersistableComponent(self)
self.persistenceFileName = AMinerConfig.buildPersistenceFileName(
aminerConfig, 'HistogramAnalysis', persistenceId)
persistenceData = PersistencyUtil.loadJson(self.persistenceFileName)
if persistenceData != None:
raise Exception('No data reading, def merge yet')
def save_current_config(self, analysis_context, destination_file):
"""
Save the current live config into a file.
@param analysis_context the analysis context of the AMiner.
@param destination_file the path to the file in which the config is saved.
"""
msg = AMinerConfig.save_config(analysis_context, destination_file)
self.REMOTE_CONTROL_RESPONSE = msg
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).info(msg)
def __init__(self,
aminer_config,
target_path,
anomaly_event_handlers,
persistence_id='Default',
auto_include_flag=False,
default_interval=3600,
realert_interval=86400,
output_log_line=True):
"""
Initialize the detector. This will also trigger reading or creation of persistence storage location.
@param target_path to extract a source identification value from each logatom.
"""
self.target_path = target_path
self.anomaly_event_handlers = anomaly_event_handlers
self.auto_include_flag = auto_include_flag
self.default_interval = default_interval
self.realert_interval = realert_interval
# This timestamps is compared with timestamp values from log atoms for activation of alerting logic. The first timestamp from logs
# above this value will trigger alerting.
self.next_check_timestamp = 0
self.last_seen_timestamp = 0
self.next_persist_time = None
self.output_log_line = output_log_line
self.aminer_config = aminer_config
self.persistence_id = persistence_id
self.log_success = 0
self.log_total = 0
self.log_learned_values = 0
self.log_new_learned_values = []
PersistenceUtil.add_persistable_component(self)
self.persistence_file_name = AMinerConfig.build_persistence_file_name(
aminer_config, self.__class__.__name__, persistence_id)
persistence_data = PersistenceUtil.load_json(
self.persistence_file_name)
self.expected_values_dict = {}
if persistence_data is not None:
for key in persistence_data:
value = persistence_data[key]
if self.target_path is not None:
if value[3] != self.target_path:
continue
elif self.target_path_list is not None:
if value[3] not in self.target_path_list:
continue
if value[1] != default_interval:
value[1] = default_interval
value[2] = value[0] + default_interval
self.expected_values_dict[key] = value
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).debug(
'%s loaded persistence data.', self.__class__.__name__)
self.analysis_string = 'Analysis.%s'
def __init__(self,
aminer_config,
anomaly_event_handlers,
parallel_check_count,
persistence_id='Default',
record_count_before_event=10000,
output_log_line=True,
use_path_match=True,
use_value_match=True,
min_rule_attributes=1,
max_rule_attributes=5):
"""
Initialize the detector. This will also trigger reading or creation of persistence storage location.
@param parallel_check_count number of rule detection checks to run in parallel.
@param record_count_before_event number of events used to calculate statistics (i.e., window size)
@param min_rule_attributes minimum number of attributes forming a rule
@param max_rule_attributes maximum number of attributes forming a rule
@param use_path_match if true rules are build based on path existance
@param use_value_match if true rules are built based on actual values
"""
self.last_timestamp = 0.0
self.parallel_check_count = parallel_check_count
self.anomaly_event_handlers = anomaly_event_handlers
self.min_rule_attributes = min_rule_attributes
self.max_rule_attributes = max_rule_attributes
self.last_unhandled_match = None
self.next_persist_time = None
self.total_records = 0
self.record_count_before_event = record_count_before_event
self.persistence_id = persistence_id
self.output_log_line = output_log_line
self.use_path_match = use_path_match
self.use_value_match = use_value_match
self.aminer_config = aminer_config
PersistenceUtil.add_persistable_component(self)
self.persistence_file_name = AMinerConfig.build_persistence_file_name(
aminer_config, 'TimeCorrelationDetector', persistence_id)
persistence_data = PersistenceUtil.load_json(
self.persistence_file_name)
if persistence_data is None:
self.feature_list = []
self.event_count_table = [
0
] * parallel_check_count * parallel_check_count * 2
self.event_delta_table = [
0
] * parallel_check_count * parallel_check_count * 2
else:
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).debug(
'%s loaded persistence data.', self.__class__.__name__)
def __init__(self,
aminer_config,
target_path_list,
anomaly_event_handlers,
id_path_list,
min_allowed_time_diff,
persistence_id='Default',
allow_missing_values_flag=False,
auto_include_flag=False,
output_log_line=True):
"""
Initialize the detector. This will also trigger reading or creation of persistence storage location.
@param target_path_list the list of values to extract from each match to create the value combination to be checked.
@param id_path_list the list of pathes where id values can be stored in all relevant log event types.
@param min_allowed_time_diff the minimum amount of time in seconds after the first appearance of a log atom with a specific id
that is waited for other log atoms with the same id to occur. The maximum possible time to keep an incomplete combo
is 2*min_allowed_time_diff
@param allow_missing_values_flag when set to True, the detector will also use matches, where one of the pathes from target_path_list
does not refer to an existing parsed data object.
@param auto_include_flag when set to True, this detector will report a new value only the first time before including it
in the known values set automatically.
"""
self.target_path_list = target_path_list
self.anomaly_event_handlers = anomaly_event_handlers
self.id_path_list = id_path_list
self.min_allowed_time_diff = min_allowed_time_diff
self.allow_missing_values_flag = allow_missing_values_flag
self.auto_include_flag = auto_include_flag
self.output_log_line = output_log_line
self.aminer_config = aminer_config
self.persistence_id = persistence_id
self.log_success = 0
self.log_total = 0
self.log_learned_path_value_combos = 0
self.log_new_learned_values = []
self.persistence_file_name = AMinerConfig.build_persistence_file_name(
aminer_config, self.__class__.__name__, persistence_id)
self.next_persist_time = None
self.load_persistence_data()
PersistenceUtil.add_persistable_component(self)
self.id_dict_current = {}
self.id_dict_old = {}
self.next_shift_time = None
def __init__(self, aminerConfig, anomalyEventHandlers, \
persistenceId='Default', autoIncludeFlag=False, outputLogLine=True):
"""Initialize the detector. This will also trigger reading
or creation of persistence storage location."""
self.anomalyEventHandlers = anomalyEventHandlers
self.autoIncludeFlag = autoIncludeFlag
self.nextPersistTime = None
self.outputLogLine = outputLogLine
self.aminerConfig = aminerConfig
PersistencyUtil.addPersistableComponent(self)
self.persistenceFileName = AMinerConfig.buildPersistenceFileName(
aminerConfig, self.__class__.__name__, persistenceId)
persistenceData = PersistencyUtil.loadJson(self.persistenceFileName)
if persistenceData is None:
self.knownPathSet = set()
else:
self.knownPathSet = set(persistenceData)
def __init__(self, aminer_config, anomaly_event_handlers, timestamp_path, analyze_path_list, min_bin_elements, min_bin_time,
debug_mode=False, persistence_id='Default', output_log_line=True):
"""
Initialize the detector. This will also trigger reading or creation of persistence storage location.
@param timestamp_path if not None, use this path value for timestamp based bins.
@param analyze_path_list list of match paths to analyze in this detector.
@param min_bin_elements evaluate the latest bin only after at least that number of elements was added to it.
@param min_bin_time evaluate the latest bin only when the first element is received after minBinTime has elapsed.
@param debug_mode if true, generate an analysis report even when average of last bin was within expected range.
"""
self.anomaly_event_handlers = anomaly_event_handlers
self.timestamp_path = timestamp_path
self.min_bin_elements = min_bin_elements
self.min_bin_time = min_bin_time
self.debug_mode = debug_mode
self.next_persist_time = None
self.persistence_id = persistence_id
self.output_log_line = output_log_line
self.aminer_config = aminer_config
PersistenceUtil.add_persistable_component(self)
self.persistence_file_name = AMinerConfig.build_persistence_file_name(aminer_config, 'MatchValueAverageChangeDetector',
persistence_id)
persistence_data = PersistenceUtil.load_json(self.persistence_file_name)
self.stat_data = []
for path in analyze_path_list:
self.stat_data.append((path, [],))
if persistence_data is not None:
for val in persistence_data:
if isinstance(val, str):
val = val.strip('[').strip(']').split(',', 2)
path = val[0].strip('"')
values = val[1].strip(' ').strip('[').strip(']')
else:
path = val[0]
values = val[1]
index = 0
for p, _ in self.stat_data:
if p == path:
break
index += 1
for value in values:
self.stat_data[index][1].append(value)
def __init__(self, programName, aminerConfig):
self.programName = programName
self.analysisContext = AnalysisContext(aminerConfig)
self.runAnalysisLoopFlag = True
self.logStreamsByName = {}
self.persistenceFileName = AMinerConfig.buildPersistenceFileName(
self.analysisContext.aminerConfig,
self.__class__.__name__ + '/RepositioningData')
self.nextPersistTime = time.time() + 600
self.repositioningDataDict = {}
self.masterControlSocket = None
self.remoteControlSocket = None
# This dictionary provides a lookup list from file descriptor
# to associated object for handling the data to and from the given
# descriptor. Currently supported handler objects are:
# * Parent process socket
# * Remote control listening socket
# * LogStreams
# * Remote control connections
self.trackedFdsDict = {}
# Override the signal handler to allow graceful shutdown.
def gracefulShutdownHandler(_signo, _stackFrame):
"""This is the signal handler function to react on typical
shutdown signals."""
print('%s: caught signal, shutting down' % programName,
file=sys.stderr)
self.runAnalysisLoopFlag = False
import signal
signal.signal(signal.SIGHUP, gracefulShutdownHandler)
signal.signal(signal.SIGINT, gracefulShutdownHandler)
signal.signal(signal.SIGTERM, gracefulShutdownHandler)
# Do this on at the end of the initialization to avoid having
# partially initialized objects inside the registry.
self.analysisContext.addTimeTriggeredComponent(self)
def __init__(self, aminerConfig, ruleset, anomalyEventHandlers, persistenceId='Default'):
"""Initialize the detector. This will also trigger reading
or creation of persistence storage location.
@param ruleset a list of MatchRule rules with appropriate
CorrelationRules attached as actions."""
self.eventClassificationRuleset = ruleset
self.anomalyEventHandlers = anomalyEventHandlers
self.nextPersistTime = time.time()+600.0
self.historyAEvents = []
self.historyBEvents = []
eventCorrelationSet = set()
for rule in self.eventClassificationRuleset:
if rule.matchAction.artefactARules is not None:
eventCorrelationSet |= set(rule.matchAction.artefactARules)
if rule.matchAction.artefactBRules is not None:
eventCorrelationSet |= set(rule.matchAction.artefactBRules)
self.eventCorrelationRuleset = list(eventCorrelationSet)
PersistencyUtil.addPersistableComponent(self)
self.persistenceFileName = AMinerConfig.buildPersistenceFileName(
aminerConfig, 'TimeCorrelationViolationDetector', persistenceId)
def __init__(self, program_name, aminer_config):
self.program_name = program_name
self.analysis_context = AnalysisContext(aminer_config)
self.run_analysis_loop_flag = True
self.log_streams_by_name = {}
self.persistence_file_name = AMinerConfig.build_persistence_file_name(
self.analysis_context.aminer_config, self.__class__.__name__ + '/RepositioningData')
self.next_persist_time = time.time() + 600
self.repositioning_data_dict = {}
self.master_control_socket = None
self.remote_control_socket = None
# This dictionary provides a lookup list from file descriptor to associated object for handling the data to and from the given
# descriptor. Currently supported handler objects are:
# * Parent process socket
# * Remote control listening socket
# * LogStreams
# * Remote control connections
self.tracked_fds_dict = {}
# Override the signal handler to allow graceful shutdown.
def graceful_shutdown_handler(_signo, _stack_frame):
"""React on typical shutdown signals."""
msg = '%s: caught signal, shutting down' % program_name
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).info(msg)
self.run_analysis_loop_flag = False
import signal
signal.signal(signal.SIGHUP, graceful_shutdown_handler)
signal.signal(signal.SIGINT, graceful_shutdown_handler)
signal.signal(signal.SIGTERM, graceful_shutdown_handler)
# Do this on at the end of the initialization to avoid having partially initialized objects inside the registry.
self.analysis_context.add_time_triggered_component(self)
def __init__(self,
aminer_config,
anomaly_event_handlers,
persistence_id='Default',
path_list=None,
min_num_vals=1000,
max_num_vals=1500,
save_values=True,
track_time_for_TSA=False,
waiting_time_for_TSA=300,
num_sections_waiting_time_for_TSA=10):
"""Initialize the detector. This will also trigger reading or creation of persistence storage location."""
self.next_persist_time = time.time() + 600.0
self.anomaly_event_handlers = anomaly_event_handlers
self.num_events = 0
# List of the longest path of the events
self.longest_path = []
# List of the keys corresponding to the events
self.found_keys = []
# List of the keys, which take values in the log-line
self.variable_key_list = []
# List of the values of the log-lines. If the lenght reaches max_num_vals the list gets reduced to min_num_vals values per variable
self.values = []
# Saves the number of lines of the event types
self.num_eventlines = []
# Saves the number of total log-lines
self.total_records = 0
# List of the modules which follow the event_type_detector. The implemented modules are form the list
# [variableTypeDetector, variableCorrelationDetector]
self.following_modules = []
# List of paths, which variables are being tracked. All other paths will not get tracked. If None all paths are being tracked.
self.path_list = path_list
# List of bools, which state if the variables of variable_key_list are updated.
self.check_variables = []
# List ot the time trigger. The first list states the times when something should be triggered, the second list states the indices
# of the eventtyps, or a list of the evnettype, a path and a value which should be counted (-1 for an initialization)
# the third list states, the length of the time window (-1 for a one time trigger)
self.etd_time_trigger = [[], [], []]
# Reference containing the number of lines of the events for the TSA
self.num_eventlines_TSA_ref = []
# Index of the eventtype of the current log line
self.current_index = 0
# Number of the values which the list is being reduced to. Be cautious that this is higher than 'num_min_values'
# in VarTypeD/Cor!!!
self.min_num_vals = min_num_vals
# Maximum number of lines in the value list before it is reduced. > min_num_vals.
self.max_num_vals = max_num_vals
# If False the values of the Token are not saved for further analysis. Disables self.values, and self.check_variables
self.save_values = save_values
# States if the time windows should be tracked for the time series analysis
self.track_time_for_TSA = track_time_for_TSA
# Time in seconds, until the time windows are being initialized
self.waiting_time_for_TSA = waiting_time_for_TSA
# Number of subdivisions of the initialization window. The length of the input-list of the function_Init-funtion is numSubd+1
self.num_sections_waiting_time_for_TSA = num_sections_waiting_time_for_TSA
self.aminer_config = aminer_config
# Loads the persistence
PersistenceUtil.add_persistable_component(self)
self.persistence_file_name = AMinerConfig.build_persistence_file_name(
aminer_config, self.__class__.__name__, persistence_id)
persistence_data = PersistenceUtil.load_json(
self.persistence_file_name)
# Imports the persistence
if persistence_data is not None:
for key in persistence_data[0]:
self.found_keys.append(set(key))
self.variable_key_list = persistence_data[1]
self.values = persistence_data[2]
self.longest_path = persistence_data[3]
self.check_variables = persistence_data[4]
self.num_eventlines = persistence_data[5]
self.etd_time_trigger = persistence_data[6]
self.num_eventlines_TSA_ref = persistence_data[7]
self.num_events = len(self.found_keys)
else:
if self.track_time_for_TSA:
self.etd_time_trigger[0].append(-1)
self.etd_time_trigger[1].append(-1)
self.etd_time_trigger[2].append(-1)
def save_current_config(self, analysis_context, destination_file):
self.REMOTE_CONTROL_RESPONSE = AMinerConfig.save_config(analysis_context, destination_file)
def test17_demo_yaml_config_equals_python_config(self):
"""This test checks if the yaml demo config is the same as the python version."""
spec = importlib.util.spec_from_file_location(
'aminer_config',
'/usr/lib/logdata-anomaly-miner/aminer/YamlConfig.py')
aminer_config = importlib.util.module_from_spec(spec)
spec.loader.exec_module(aminer_config)
aminer_config.load_yaml('demo/AMiner/demo-config.yml')
yml_context = AnalysisContext(aminer_config)
yml_context.build_analysis_pipeline()
aminer_config = AMinerConfig.load_config('demo/AMiner/demo-config.py')
py_context = AnalysisContext(aminer_config)
py_context.build_analysis_pipeline()
import copy
yml_config_properties = copy.deepcopy(
yml_context.aminer_config.config_properties)
del yml_config_properties['Parser']
del yml_config_properties['Input']
del yml_config_properties['Analysis']
del yml_config_properties['EventHandlers']
del yml_config_properties['LearnMode']
del yml_config_properties['SuppressNewMatchPathDetector']
# remove SimpleUnparsedAtomHandler, VerboseUnparsedAtomHandler and NewMatchPathDetector as they are added by the YamlConfig.
py_registered_components = copy.copy(py_context.registered_components)
del py_registered_components[0]
del py_registered_components[1]
del py_registered_components[2]
del py_registered_components[10]
yml_registered_components = copy.copy(
yml_context.registered_components)
del yml_registered_components[0]
del yml_registered_components[1]
tmp = {}
keys = list(py_registered_components.keys())
for i in range(1, len(py_registered_components) + 1):
tmp[i] = py_registered_components[keys[i - 1]]
py_registered_components = tmp
py_registered_components_by_name = copy.copy(
py_context.registered_components_by_name)
del py_registered_components_by_name['SimpleUnparsedHandler']
del py_registered_components_by_name['VerboseUnparsedHandler']
del py_registered_components_by_name['NewMatchPath']
del py_registered_components_by_name['SimpleMonotonicTimestampAdjust']
yml_registered_components_by_name = copy.copy(
yml_context.registered_components_by_name)
del yml_registered_components_by_name['DefaultNewMatchPathDetector']
del yml_registered_components_by_name['AtomFilter']
self.assertEqual(yml_config_properties,
py_context.aminer_config.config_properties)
# there actually is no easy way to compare AMiner components as they do not implement the __eq__ method.
self.assertEqual(len(yml_registered_components),
len(py_registered_components))
for i in range(2, len(yml_registered_components)):
self.assertEqual(type(yml_registered_components[i]),
type(py_registered_components[i]))
self.assertEqual(yml_registered_components_by_name.keys(),
py_registered_components_by_name.keys())
for name in yml_registered_components_by_name.keys():
self.assertEqual(type(yml_registered_components_by_name[name]),
type(py_registered_components_by_name[name]))
self.assertEqual(len(yml_context.real_time_triggered_components),
len(py_context.real_time_triggered_components))
# the atom_handler_list is not equal as the python version uses a SimpleMonotonicTimestampAdjust.
self.assertEqual(yml_context.atomizer_factory.default_timestamp_paths,
py_context.atomizer_factory.default_timestamp_paths)
self.assertEqual(type(yml_context.atomizer_factory.event_handler_list),
type(py_context.atomizer_factory.event_handler_list))
def main():
"""Run the aminer main program."""
# Extract program name, but only when sure to contain no problematic characters.
warnings.filterwarnings('ignore', category=ImportWarning)
program_name = sys.argv[0].split('/')[-1]
if (program_name == '.') or (program_name == '..') or (re.match(
'^[a-zA-Z0-9._-]+$', program_name) is None):
print('Invalid program name, check your execution args',
file=sys.stderr)
sys.exit(1)
# We will not read stdin from here on, so get rid of it immediately, thus aberrant child cannot manipulate caller's stdin using it.
stdin_fd = os.open('/dev/null', os.O_RDONLY)
os.dup2(stdin_fd, 0)
os.close(stdin_fd)
help_message = 'aminer - logdata-anomaly-miner\n'
if supports_color():
help_message += colflame
else:
help_message += flame
parser = argparse.ArgumentParser(
description=help_message,
formatter_class=argparse.RawTextHelpFormatter)
parser.add_argument('-v',
'--version',
action='version',
version=__version_string__)
parser.add_argument('-c',
'--config',
default='/etc/aminer/config.yml',
type=str,
help='path to the config-file')
parser.add_argument('-D',
'--daemon',
action='store_false',
help='run as a daemon process')
parser.add_argument(
'-s',
'--stat',
choices=[0, 1, 2],
type=int,
help='set the stat level. Possible stat-levels are 0 for no statistics'
', 1 for normal statistic level and 2 for verbose statistics.')
parser.add_argument(
'-d',
'--debug',
choices=[0, 1, 2],
type=int,
help='set the debug level. Possible debug-levels are 0 for no '
'debugging, 1 for normal output (INFO and above), 2 for printing'
' all debug information.')
parser.add_argument('--run-analysis',
action='store_true',
help='enable/disable analysis')
parser.add_argument('-C',
'--clear',
action='store_true',
help='removes all persistence directories')
parser.add_argument('-r',
'--remove',
action='append',
type=str,
help='removes a specific persistence directory')
parser.add_argument('-R',
'--restore',
type=str,
help='restore a persistence backup')
parser.add_argument(
'-f',
'--from-begin',
action='store_true',
help='removes RepositioningData before starting the aminer')
args = parser.parse_args()
config_file_name = args.config
run_in_foreground_flag = args.daemon
run_analysis_child_flag = args.run_analysis
clear_persistence_flag = args.clear
remove_persistence_dirs = args.remove
from_begin_flag = args.from_begin
if args.restore is not None and ('.' in args.restore
or '/' in args.restore):
parser.error('The restore path %s must not contain any . or /' %
args.restore)
if args.remove is not None:
for remove in args.remove:
if '.' in remove or '/' in remove:
parser.error('The remove path %s must not contain any . or /' %
remove)
restore_relative_persistence_path = args.restore
stat_level = 1
debug_level = 1
stat_level_console_flag = False
debug_level_console_flag = False
if args.stat is not None:
stat_level = args.stat
stat_level_console_flag = True
if args.debug is not None:
debug_level = args.stat
debug_level_console_flag = True
# Load the main configuration file.
if not os.path.exists(config_file_name):
print('%s: config "%s" not (yet) available!' %
(program_name, config_file_name),
file=sys.stderr)
sys.exit(1)
# Minimal import to avoid loading too much within the privileged process.
try:
aminer_config = AMinerConfig.load_config(config_file_name)
except ValueError as e:
print("Config-Error: %s" % e)
sys.exit(1)
persistence_dir = aminer_config.config_properties.get(
AMinerConfig.KEY_PERSISTENCE_DIR, AMinerConfig.DEFAULT_PERSISTENCE_DIR)
child_user_name = aminer_config.config_properties.get(
AMinerConfig.KEY_AMINER_USER)
child_group_name = aminer_config.config_properties.get(
AMinerConfig.KEY_AMINER_GROUP)
child_user_id = -1
child_group_id = -1
try:
if child_user_name is not None:
from pwd import getpwnam
child_user_id = getpwnam(child_user_name).pw_uid
if child_group_name is not None:
from grp import getgrnam
child_group_id = getgrnam(child_user_name).gr_gid
except: # skipcq: FLK-E722
print('Failed to resolve %s or %s' %
(AMinerConfig.KEY_AMINER_USER, AMinerConfig.KEY_AMINER_GROUP),
file=sys.stderr)
sys.exit(1)
initialize_loggers(aminer_config, child_user_name, child_group_name)
if restore_relative_persistence_path is not None and (
clear_persistence_flag or remove_persistence_dirs):
msg = 'The --restore parameter removes all persistence files. Do not use this parameter with --Clear or --Remove!'
print(msg, sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
sys.exit(1)
if not stat_level_console_flag and AMinerConfig.KEY_LOG_STAT_LEVEL in aminer_config.config_properties:
stat_level = aminer_config.config_properties[
AMinerConfig.KEY_LOG_STAT_LEVEL]
if not debug_level_console_flag and AMinerConfig.KEY_LOG_DEBUG_LEVEL in aminer_config.config_properties:
debug_level = aminer_config.config_properties[
AMinerConfig.KEY_LOG_DEBUG_LEVEL]
AMinerConfig.STAT_LEVEL = stat_level
AMinerConfig.DEBUG_LEVEL = debug_level
if clear_persistence_flag:
if remove_persistence_dirs:
msg = 'The --clear and --remove arguments must not be used together!'
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
sys.exit(1)
clear_persistence(persistence_dir)
if remove_persistence_dirs:
persistence_dir_name = aminer_config.config_properties.get(
AMinerConfig.KEY_PERSISTENCE_DIR,
AMinerConfig.DEFAULT_PERSISTENCE_DIR)
for filename in os.listdir(persistence_dir_name):
file_path = os.path.join(persistence_dir_name, filename)
try:
if not os.path.isdir(file_path):
msg = 'The aminer persistence directory should not contain any files.'
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).warning(msg)
continue
shutil.rmtree(file_path)
except OSError as e:
msg = 'Failed to delete %s. Reason: %s' % (file_path, e)
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
for filename in remove_persistence_dirs:
file_path = os.path.join(persistence_dir, filename)
try:
if not os.path.exists(file_path):
continue
if not os.path.isdir(file_path):
msg = 'The aminer persistence directory should not contain any files.'
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).warning(msg)
continue
shutil.rmtree(file_path)
except OSError as e:
msg = 'Failed to delete %s. Reason: %s' % (file_path, e)
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
if restore_relative_persistence_path is not None:
absolute_persistence_path = os.path.join(
persistence_dir, 'backup', restore_relative_persistence_path)
if not os.path.exists(absolute_persistence_path):
msg = '%s does not exist. Continuing without restoring persistence.' % absolute_persistence_path
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).warning(msg)
else:
clear_persistence(persistence_dir)
copytree(absolute_persistence_path, persistence_dir)
aminer_user = aminer_config.config_properties.get(
AMinerConfig.KEY_AMINER_USER)
aminer_grp = aminer_config.config_properties.get(
AMinerConfig.KEY_AMINER_GROUP)
for dirpath, _dirnames, filenames in os.walk(persistence_dir):
shutil.chown(dirpath, aminer_user, aminer_grp)
for filename in filenames:
shutil.chown(os.path.join(dirpath, filename), aminer_user,
aminer_grp)
if from_begin_flag:
repositioning_data_path = os.path.join(
aminer_config.config_properties.get(
AMinerConfig.KEY_PERSISTENCE_DIR,
AMinerConfig.DEFAULT_PERSISTENCE_DIR), 'AnalysisChild',
'RepositioningData')
if os.path.exists(repositioning_data_path):
os.remove(repositioning_data_path)
if run_analysis_child_flag:
# Call analysis process, this function will never return.
run_analysis_child(aminer_config, program_name)
# Start importing of aminer specific components after reading of "config.py" to allow replacement of components via sys.path
# from within configuration.
from aminer.util import SecureOSFunctions
from aminer.util import decode_string_as_byte_string
log_sources_list = aminer_config.config_properties.get(
AMinerConfig.KEY_LOG_SOURCES_LIST)
if (log_sources_list is None) or not log_sources_list:
msg = '%s: %s not defined' % (program_name,
AMinerConfig.KEY_LOG_SOURCES_LIST)
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
sys.exit(1)
# Now create the management entries for each logfile.
log_data_resource_dict = {}
for log_resource_name in log_sources_list:
# From here on log_resource_name is a byte array.
log_resource_name = decode_string_as_byte_string(log_resource_name)
log_resource = None
if log_resource_name.startswith(b'file://'):
from aminer.input.LogStream import FileLogDataResource
log_resource = FileLogDataResource(log_resource_name, -1)
elif log_resource_name.startswith(b'unix://'):
from aminer.input.LogStream import UnixSocketLogDataResource
log_resource = UnixSocketLogDataResource(log_resource_name, -1)
else:
msg = 'Unsupported schema in %s: %s' % (
AMinerConfig.KEY_LOG_SOURCES_LIST, repr(log_resource_name))
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
sys.exit(1)
if not os.path.exists(log_resource_name[7:].decode()):
msg = "WARNING: file or socket '%s' does not exist (yet)!" % log_resource_name[
7:].decode()
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).warning(msg)
try:
log_resource.open()
except OSError as open_os_error:
if open_os_error.errno == errno.EACCES:
msg = '%s: no permission to access %s' % (
program_name, repr(log_resource_name))
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
sys.exit(1)
else:
msg = '%s: unexpected error opening %s: %d (%s)' % (
program_name, repr(log_resource_name), open_os_error.errno,
os.strerror(open_os_error.errno))
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
sys.exit(1)
log_data_resource_dict[log_resource_name] = log_resource
# Create the remote control socket, if any. Do this in privileged mode to allow binding it at arbitrary locations and support restricted
# permissions of any type for current (privileged) uid.
remote_control_socket_name = aminer_config.config_properties.get(
AMinerConfig.KEY_REMOTE_CONTROL_SOCKET_PATH, None)
remote_control_socket = None
if remote_control_socket_name is not None:
if os.path.exists(remote_control_socket_name):
try:
os.unlink(remote_control_socket_name)
except OSError:
msg = 'Failed to clean up old remote control socket at %s' % remote_control_socket_name
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
sys.exit(1)
# Create the local socket: there is no easy way to create it with correct permissions, hence a fork is needed, setting umask,
# bind the socket. It is also recommended to create the socket in a directory having the correct permissions already.
remote_control_socket = socket.socket(socket.AF_UNIX,
socket.SOCK_STREAM)
remote_control_socket.setblocking(False)
bind_child_pid = os.fork()
if bind_child_pid == 0:
os.umask(0o177)
remote_control_socket.bind(remote_control_socket_name)
# Do not perform any cleanup, flushing of streams. Use _exit(0) to avoid interference with fork.
os._exit(0) # skipcq: PYL-W0212
os.waitpid(bind_child_pid, 0)
remote_control_socket.listen(4)
# Now have checked all we can get from the configuration in the privileged process. Detach from the TTY when in daemon mode.
if not run_in_foreground_flag:
child_pid = 0
try:
# Fork a child to make sure, we are not the process group leader already.
child_pid = os.fork()
except Exception as fork_exception: # skipcq: PYL-W0703
msg = 'Failed to daemonize: %s' % fork_exception
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
sys.exit(1)
if child_pid != 0:
# This is the parent.
os._exit(0) # skipcq: PYL-W0212
# This is the child. Create a new session and become process group leader. Here we get rid of the controlling tty.
os.setsid()
# Fork again to become an orphaned process not being session leader, hence not able to get a controlling tty again.
try:
child_pid = os.fork()
except Exception as fork_exception: # skipcq: PYL-W0703
msg = 'Failed to daemonize: %s' % fork_exception
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
sys.exit(1)
if child_pid != 0:
# This is the parent.
os._exit(0) # skipcq: PYL-W0212
# Move to root directory to avoid lingering in some cwd someone else might want to unmount.
os.chdir('/')
# Change the umask here to clean all group/other mask bits so that accidentially created files are not accessible by other.
os.umask(0o77)
# Install a signal handler catching common stop signals and relaying it to all children for sure.
# skipcq: PYL-W0603
global child_termination_triggered_flag
child_termination_triggered_flag = False
def graceful_shutdown_handler(_signo, _stackFrame):
"""React on typical shutdown signals."""
msg = '%s: caught signal, shutting down' % program_name
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).info(msg)
# Just set the flag. It is likely, that child received same signal also so avoid multiple signaling, which could interrupt the
# shutdown procedure again.
# skipcq: PYL-W0603
global child_termination_triggered_flag
child_termination_triggered_flag = True
import signal
signal.signal(signal.SIGHUP, graceful_shutdown_handler)
signal.signal(signal.SIGINT, graceful_shutdown_handler)
signal.signal(signal.SIGTERM, graceful_shutdown_handler)
# Now create the socket to connect the analysis child.
(parent_socket, child_socket) = socket.socketpair(socket.AF_UNIX,
socket.SOCK_DGRAM, 0)
# Have it nonblocking from here on.
parent_socket.setblocking(False)
child_socket.setblocking(False)
# Use normal fork, we should have been detached from TTY already. Flush stderr to avoid duplication of output if both child and
# parent want to write something.
sys.stderr.flush()
child_pid = os.fork()
if child_pid == 0:
# Relocate the child socket fd to 3 if needed
if child_socket.fileno() != 3:
os.dup2(child_socket.fileno(), 3)
child_socket.close()
# Clear the supplementary groups before dropping privileges. This makes only sense when changing the uid or gid.
if os.getuid() == 0:
if ((child_user_id != -1) and (child_user_id != os.getuid())) or (
(child_group_id != -1) and (child_group_id != os.getgid())):
os.setgroups([])
# Drop privileges before executing child. setuid/gid will raise an exception when call has failed.
if child_group_id != -1:
os.setgid(child_group_id)
if child_user_id != -1:
os.setuid(child_user_id)
else:
msg = 'INFO: No privilege separation when started as unprivileged user'
print(msg, file=sys.stderr)
initialize_loggers(aminer_config, 'aminer', 'aminer')
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).info(msg)
# Now resolve the specific analysis configuration file (if any).
analysis_config_file_name = aminer_config.config_properties.get(
AMinerConfig.KEY_ANALYSIS_CONFIG_FILE, None)
if analysis_config_file_name is None:
analysis_config_file_name = config_file_name
elif not os.path.isabs(analysis_config_file_name):
analysis_config_file_name = os.path.join(
os.path.dirname(config_file_name), analysis_config_file_name)
# This is the child. Close all parent file descriptors, we do not need. Perhaps this could be done more elegantly.
for close_fd in range(4, 1 << 16):
try:
os.close(close_fd)
except OSError as open_os_error:
if open_os_error.errno == errno.EBADF:
continue
msg = '%s: unexpected exception closing file descriptors: %s' % (
program_name, open_os_error)
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
# Flush stderr before exit without any cleanup.
sys.stderr.flush()
os._exit(1) # skipcq: PYL-W0212
# Now execute the very same program again, but user might have moved or renamed it meanwhile. This would be problematic with
# SUID-binaries (which we do not yet support). Do NOT just fork but also exec to avoid child circumventing
# parent's ALSR due to cloned kernel VMA.
execArgs = [
'AMinerChild', '--run-analysis', '--config',
analysis_config_file_name, '--stat',
str(stat_level), '--debug',
str(debug_level)
]
os.execve(sys.argv[0], execArgs, {}) # skipcq: BAN-B606
msg = 'Failed to execute child process'
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
sys.stderr.flush()
os._exit(1) # skipcq: PYL-W0212
child_socket.close()
# Send all log resource information currently available to child process.
for log_resource_name, log_resource in log_data_resource_dict.items():
if (log_resource
is not None) and (log_resource.get_file_descriptor() >= 0):
SecureOSFunctions.send_logstream_descriptor(
parent_socket, log_resource.get_file_descriptor(),
log_resource_name)
log_resource.close()
# Send the remote control server socket, if any and close it afterwards. It is not needed any more on parent side.
if remote_control_socket is not None:
SecureOSFunctions.send_annotated_file_descriptor(
parent_socket, remote_control_socket.fileno(), 'remotecontrol', '')
remote_control_socket.close()
exit_status = 0
child_termination_triggered_count = 0
while True:
if child_termination_triggered_flag:
if child_termination_triggered_count == 0:
time.sleep(1)
elif child_termination_triggered_count < 5:
os.kill(child_pid, signal.SIGTERM)
else:
os.kill(0, signal.SIGKILL)
child_termination_triggered_count += 1
(sig_child_pid, sig_status) = os.waitpid(-1, os.WNOHANG)
if sig_child_pid != 0:
if sig_child_pid == child_pid:
if child_termination_triggered_flag:
# This was expected, just terminate.
break
msg = '%s: Analysis child process %d terminated unexpectedly with signal 0x%x' % (
program_name, sig_child_pid, sig_status)
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
exit_status = 1
break
# So the child has been cloned, the clone has terminated. This should not happen either.
msg = '%s: untracked child %d terminated with with signal 0x%x' % (
program_name, sig_child_pid, sig_status)
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
exit_status = 1
# Child information handled, scan for rotated logfiles or other resources, where reopening might make sense.
for log_resouce_name, log_data_resource in log_data_resource_dict.items(
):
try:
if not log_data_resource.open(reopen_flag=True):
continue
except OSError as open_os_error:
if open_os_error.errno == errno.EACCES:
msg = '%s: no permission to access %s' % (program_name,
log_resouce_name)
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
else:
msg = '%s: unexpected error reopening %s: %d (%s)' % (
program_name, log_resouce_name, open_os_error.errno,
os.strerror(open_os_error.errno))
print(msg, file=sys.stderr)
logging.getLogger(AMinerConfig.DEBUG_LOG_NAME).error(msg)
exit_status = 2
continue
SecureOSFunctions.send_logstream_descriptor(
parent_socket, log_data_resource.get_file_descriptor(),
log_resouce_name)
log_data_resource.close()
time.sleep(1)
parent_socket.close()
SecureOSFunctions.close_base_directory()
sys.exit(exit_status)
