def add(self, elt):
cls_type = elt.__class__.my_type
if cls_type == 'brok':
# For brok, we TAG brok with our instance_id
elt.data['instance_id'] = 0
self.broks_internal_raised.append(elt)
return
elif cls_type == 'externalcommand':
print "Adding in queue an external command", ExternalCommand.__dict__
self.external_commands.append(elt)
# Maybe we got a Message from the modules, it's way to ask something
#like from now a full data from a scheduler for example.
elif cls_type == 'message':
# We got a message, great!
print elt.__dict__
if elt.get_type() == 'NeedData':
data = elt.get_data()
# Full instance id mean : I got no data for this scheduler
# so give me all dumbass!
if 'full_instance_id' in data:
c_id = data['full_instance_id']
logger.log('A module is asking me to get all initial data from the scheduler %d' % c_id)
# so we just reset the connexion adn the running_id, it will just get all new things
self.schedulers[c_id]['con'] = None
self.schedulers[c_id]['running_id'] = 0
def load(self):
now = int(time.time())
""" Try to import the requested modules ; put the imported modules in self.imported_modules.
The previous imported modules, if any, are cleaned before. """
# We get all modules file with .py
modules_files = [ fname[:-3] for fname in os.listdir(self.modules_path)
if fname.endswith(".py") ]
# And directories
modules_files.extend([ fname for fname in os.listdir(self.modules_path)
if os.path.isdir(os.path.join(self.modules_path, fname)) ])
# Now we try to load thems
# So first we add their dir into the sys.path
if not self.modules_path in sys.path:
sys.path.append(self.modules_path)
# We try to import them, but we keep only the one of
# our type
del self.imported_modules[:]
for fname in modules_files:
#print "Try to load", fname
try:
m = __import__(fname)
if not hasattr(m, 'properties'):
continue
# We want to keep only the modules of our type
if self.modules_type in m.properties['daemons']:
self.imported_modules.append(m)
except Exception , exp:
logger.log("Warning in importing module : %s" % exp)
def no_loop_in_parents(self):
# Ok, we say "from now, no loop :) "
r = True
# Create parent graph
parents = Graph()
# With all hosts as nodes
for h in self:
if h is not None:
parents.add_node(h)
# And now fill edges
for h in self:
for p in h.parents:
if p is not None:
parents.add_edge(p, h)
# Now get the list of all hosts in a loop
host_in_loops = parents.loop_check()
# and raise errors about it
for h in host_in_loops:
logger.log("Error: The host '%s' is part of a circular parent/child chain!" % h.get_name())
r = False
return r
def is_correct(self):
state = True #guilty or not? :)
cls = self.__class__
#All of the above are checks in the notificationways part
for prop, entry in cls.properties.items():
if prop not in _special_properties:
if not hasattr(self, prop) and entry.required:
print self.get_name(), " : I do not have", prop
state = False #Bad boy...
#There is a case where there is no nw : when there is not special_prop defined
#at all!!
if self.notificationways == []:
for p in _special_properties:
print self.get_name(), " : I'm missing the property %s" % p
state = False
if hasattr(self, 'contact_name'):
for c in cls.illegal_object_name_chars:
if c in self.contact_name:
logger.log(
"%s : My contact_name got the caracter %s that is not allowed."
% (self.get_name(), c))
state = False
else:
if hasattr(self,
'alias'): #take the alias if we miss the contact_name
self.contact_name = self.alias
return state
def is_correct(self):
state = True #guilty or not? :)
cls = self.__class__
#All of the above are checks in the notificationways part
for prop, entry in cls.properties.items():
if prop not in _special_properties:
if not hasattr(self, prop) and entry.required:
print self.get_name(), " : I do not have", prop
state = False #Bad boy...
#There is a case where there is no nw : when there is not special_prop defined
#at all!!
if self.notificationways == []:
for p in _special_properties:
print self.get_name()," : I'm missing the property %s" % p
state = False
if hasattr(self, 'contact_name'):
for c in cls.illegal_object_name_chars:
if c in self.contact_name:
logger.log("%s : My contact_name got the caracter %s that is not allowed." % (self.get_name(), c))
state = False
else:
if hasattr(self, 'alias'): #take the alias if we miss the contact_name
self.contact_name = self.alias
return state
def search(look_at):
# Now really publish it
proxy = CONFIG['shinken.io']['proxy']
api_key = CONFIG['shinken.io']['api_key']
# Ok we will push the file with a 10s timeout
c = pycurl.Curl()
c.setopt(c.POST, 0)
c.setopt(c.CONNECTTIMEOUT, 10)
c.setopt(c.TIMEOUT, 10)
if proxy:
c.setopt(c.PROXY, proxy)
args = {'keywords': ','.join(look_at)}
c.setopt(c.URL, str('shinken.io/searchcli?' + urllib.urlencode(args)))
response = StringIO()
c.setopt(pycurl.WRITEFUNCTION, response.write)
#c.setopt(c.VERBOSE, 1)
c.perform()
r = c.getinfo(pycurl.HTTP_CODE)
c.close()
if r != 200:
logger.error("There was a critical error : %s" % response.getvalue())
sys.exit(2)
else:
ret = json.loads(response.getvalue().replace('\\/', '/'))
status = ret.get('status')
result = ret.get('result')
if status != 200:
logger.log(result)
return []
return result
def publish_archive(archive):
# Now really publish it
proxy = CONFIG['shinken.io']['proxy']
api_key = CONFIG['shinken.io']['api_key']
# Ok we will push the file with a 10s timeout
c = pycurl.Curl()
c.setopt(c.POST, 1)
c.setopt(c.CONNECTTIMEOUT, 10)
c.setopt(c.TIMEOUT, 10)
if proxy:
c.setopt(c.PROXY, proxy)
c.setopt(c.URL, "http://shinken.io/push")
c.setopt(c.HTTPPOST, [("api_key", api_key),
("data",
(c.FORM_FILE, str(archive),
c.FORM_CONTENTTYPE, "application/x-gzip"))
])
response = StringIO()
c.setopt(pycurl.WRITEFUNCTION, response.write)
c.setopt(c.VERBOSE, 1)
c.perform()
r = c.getinfo(pycurl.HTTP_CODE)
c.close()
if r != 200:
logger.error("There was a critical error : %s" % response.getvalue())
sys.exit(2)
else:
ret = json.loads(response.getvalue().replace('\\/', '/'))
status = ret.get('status')
text = ret.get('text')
if status == 200:
logger.log(text)
else:
logger.error(text)
def get_scheduler_ordered_list(self, r):
# get scheds, alive and no spare first
scheds = []
for s in r.schedulers:
scheds.append(s)
# now the spare scheds of higher realms
# they are after the sched of realm, so
# they will be used after the spare of
# the realm
for higher_r in r.higher_realms:
for s in higher_r.schedulers:
if s.spare:
scheds.append(s)
# Now we sort the scheds so we take master, then spare
# the dead, but we do not care about thems
scheds.sort(alive_then_spare_then_deads)
scheds.reverse() #pop is last, I need first
#DBG: dump
print_sched = [s.get_name() for s in scheds]
print_sched.reverse()
print_string = '[%s] Schedulers order : ' % r.get_name()
for s in print_sched:
print_string += '%s ' % s
logger.log(print_string)
#END DBG
return scheds
def add(self, elt):
cls_type = elt.__class__.my_type
if cls_type == "brok":
# For brok, we TAG brok with our instance_id
elt.data["instance_id"] = 0
self.broks_internal_raised.append(elt)
return
elif cls_type == "externalcommand":
print "Adding in queue an external command", ExternalCommand.__dict__
self.external_commands.append(elt)
# Maybe we got a Message from the modules, it's way to ask something
# like from now a full data from a scheduler for example.
elif cls_type == "message":
# We got a message, great!
print elt.__dict__
if elt.get_type() == "NeedData":
data = elt.get_data()
# Full instance id mean : I got no data for this scheduler
# so give me all dumbass!
if "full_instance_id" in data:
c_id = data["full_instance_id"]
logger.log("A module is asking me to get all initial data from the scheduler %d" % c_id)
# so we just reset the connection adn the running_id, it will just get all new things
try:
self.schedulers[c_id]["con"] = None
self.schedulers[c_id]["running_id"] = 0
except KeyError: # maybe this instance was not known, forget it
print "WARNING: a module ask me a full_instance_id for an unknown ID!", c_id
def init(self):
logger.log("I connect to NDO database")
self.db = DBMysql(self.host, self.user, self.password, self.database,
self.character_set, table_prefix='nagios_', port=self.port)
self.connect_database()
# Cache for hosts and services
# The structure is as follow:
# First the instance id then the host / (host,service desc) to access the wanted data
self.services_cache_sync = {}
self.hosts_cache_sync = {}
# We need to search for centreon_specific fields, like long_output
query = u"select TABLE_NAME from information_schema.columns where TABLE_SCHEMA='ndo' and TABLE_NAME='nagios_servicestatus' and COLUMN_NAME='long_output';"
self.db.execute_query(query)
row = self.db.fetchone()
if row is None or len(row) < 1:
self.centreon_version = False
else:
self.centreon_version = True
logger.log("[MySQL/NDO] Using the centreon version")
# Cache for database id
# In order not to query the database every time
self.database_id_cache = {}
# Mapping service_id in Shinken and in database
# Because can't acces host_name from a service everytime :(
self.mapping_service_id = {}
# Todo list to manage brok
self.todo = []
def get_new_broks(self, type='scheduler'):
# Get the good links tab for looping..
links = self.get_links_from_type(type)
if links is None:
logger.log('DBG: Type unknown for connection! %s' % type)
return
# We check for new check in each schedulers and put
# the result in new_checks
for sched_id in links:
try:
con = links[sched_id]['con']
if con is not None: # None = not initilized
tmp_broks = con.get_broks()
for b in tmp_broks.values():
b.instance_id = links[sched_id]['instance_id']
# Ok, we can add theses broks to our queues
self.add_broks_to_queue(tmp_broks.values())
else: # no con? make the connection
self.pynag_con_init(sched_id, type=type)
# Ok, con is not known, so we create it
except KeyError , exp:
print exp
self.pynag_con_init(sched_id, type=type)
except Pyro.errors.ProtocolError , exp:
logger.log("[%s] Connection problem to the %s %s : %s" % (self.name, type, links[sched_id]['name'], str(exp)))
links[sched_id]['con'] = None
def search(look_at):
# Now really publish it
proxy = CONFIG['shinken.io']['proxy']
api_key = CONFIG['shinken.io']['api_key']
# Ok we will push the file with a 10s timeout
c = pycurl.Curl()
c.setopt(c.POST, 0)
c.setopt(c.CONNECTTIMEOUT, 10)
c.setopt(c.TIMEOUT, 10)
if proxy:
c.setopt(c.PROXY, proxy)
args = {'keywords':','.join(look_at)}
c.setopt(c.URL, str('shinken.io/searchcli?'+urllib.urlencode(args)))
response = StringIO()
c.setopt(pycurl.WRITEFUNCTION, response.write)
#c.setopt(c.VERBOSE, 1)
c.perform()
r = c.getinfo(pycurl.HTTP_CODE)
c.close()
if r != 200:
logger.error("There was a critical error : %s" % response.getvalue())
sys.exit(2)
else:
ret = json.loads(response.getvalue().replace('\\/', '/'))
status = ret.get('status')
result = ret.get('result')
if status != 200:
logger.log(result)
return []
return result
def hook_save_retention(self, daemon):
log_mgr = logger
logger.log(
"[PickleRetentionGeneric] asking me to update the retention objects"
)
#Now the flat file method
try:
# Open a file near the path, with .tmp extension
# so in cae or problem, we do not lost the old one
f = open(self.path + '.tmp', 'wb')
# We get interesting retention data from the daemon it self
all_data = daemon.get_retention_data()
# And we save it on file :)
#s = cPickle.dumps(all_data)
#s_compress = zlib.compress(s)
cPickle.dump(all_data, f, protocol=cPickle.HIGHEST_PROTOCOL)
#f.write(s_compress)
f.close()
# Now move the .tmp fiel to the real path
shutil.move(self.path + '.tmp', self.path)
except IOError, exp:
log_mgr.log("Error: retention file creation failed, %s" % str(exp))
return
def get_scheduler_ordered_list(self, r):
# get scheds, alive and no spare first
scheds = []
for s in r.schedulers:
scheds.append(s)
# now the spare scheds of higher realms
# they are after the sched of realm, so
# they will be used after the spare of
# the realm
for higher_r in r.higher_realms:
for s in higher_r.schedulers:
if s.spare:
scheds.append(s)
# Now we sort the scheds so we take master, then spare
# the dead, but we do not care about thems
scheds.sort(alive_then_spare_then_deads)
scheds.reverse() #pop is last, I need first
#DBG: dump
print_sched = [s.get_name() for s in scheds]
print_sched.reverse()
print_string = '[%s] Schedulers order : ' % r.get_name()
for s in print_sched:
print_string += '%s ' % s
logger.log(print_string)
#END DBG
return scheds
def try_instance_init(self, inst):
""" Try to "init" the given module instance.
Returns: True on successfull init. False if instance init method raised any Exception. """
try:
print "Trying to init module", inst.get_name()
inst.init_try += 1
# Maybe it's a retry
if inst.init_try > 1:
# Do not try until 5 sec, or it's too loopy
if inst.last_init_try > time.time() - 5:
return False
inst.last_init_try = time.time()
# If it's an external, create/update Queues()
if inst.is_external:
inst.create_queues()
inst.init()
except Exception, e:
logger.log(
"Error : the instance %s raised an exception %s, I remove it!"
% (inst.get_name(), str(e)))
output = cStringIO.StringIO()
traceback.print_exc(file=output)
logger.log("Back trace of this remove : %s" % (output.getvalue()))
output.close()
return False
def hook_save_retention(self, daemon):
log_mgr = logger
logger.log("[PickleRetentionGeneric] asking me to update the retention objects")
#Now the flat file method
try:
# Open a file near the path, with .tmp extension
# so in cae or problem, we do not lost the old one
f = open(self.path+'.tmp', 'wb')
# We get interesting retention data from the daemon it self
all_data = daemon.get_retention_data()
# And we save it on file :)
#s = cPickle.dumps(all_data)
#s_compress = zlib.compress(s)
cPickle.dump(all_data, f, protocol=cPickle.HIGHEST_PROTOCOL)
#f.write(s_compress)
f.close()
# Now move the .tmp fiel to the real path
shutil.move(self.path+'.tmp', self.path)
except IOError , exp:
log_mgr.log("Error: retention file creation failed, %s" % str(exp))
return
def try_instance_init(self, inst):
""" Try to "init" the given module instance.
Returns: True on successfull init. False if instance init method raised any Exception. """
try:
print "Trying to init module", inst.get_name()
inst.init_try += 1
# Maybe it's a retry
if inst.init_try > 1:
# Do not try until 5 sec, or it's too loopy
if inst.last_init_try > time.time() - 5:
return False
inst.last_init_try = time.time()
# If it's an external, create/update Queues()
if inst.is_external:
inst.create_queues()
inst.init()
except Exception, e:
logger.log("Error : the instance %s raised an exception %s, I remove it!" % (inst.get_name(), str(e)))
output = cStringIO.StringIO()
traceback.print_exc(file=output)
logger.log("Back trace of this remove : %s" % (output.getvalue()))
output.close()
return False
def add(self, elt):
cls_type = elt.__class__.my_type
if cls_type == 'brok':
# For brok, we TAG brok with our instance_id
elt.data['instance_id'] = 0
self.broks_internal_raised.append(elt)
return
elif cls_type == 'externalcommand':
print "Adding in queue an external command", ExternalCommand.__dict__
self.external_commands.append(elt)
# Maybe we got a Message from the modules, it's way to ask something
#like from now a full data from a scheduler for example.
elif cls_type == 'message':
# We got a message, great!
print elt.__dict__
if elt.get_type() == 'NeedData':
data = elt.get_data()
# Full instance id mean : I got no data for this scheduler
# so give me all dumbass!
if 'full_instance_id' in data:
c_id = data['full_instance_id']
logger.log(
'A module is asking me to get all initial data from the scheduler %d'
% c_id)
# so we just reset the connexion adn the running_id, it will just get all new things
self.schedulers[c_id]['con'] = None
self.schedulers[c_id]['running_id'] = 0
def compensate_system_time_change(self, difference):
""" Compensate a system time change of difference for all hosts/services/checks/notifs """
logger.log('Warning: A system time change of %d has been detected. Compensating...' % difference)
# We only need to change some value
self.program_start = max(0, self.program_start + difference)
# Then we compasate all host/services
for h in self.sched.hosts:
h.compensate_system_time_change(difference)
for s in self.sched.services:
s.compensate_system_time_change(difference)
# Now all checks and actions
for c in self.sched.checks.values():
# Already launch checks should not be touch
if c.status == 'scheduled':
t_to_go = c.t_to_go
ref = c.ref
new_t = max(0, t_to_go + difference)
# But it's no so simple, we must match the timeperiod
new_t = ref.check_period.get_next_valid_time_from_t(new_t)
# But maybe no there is no more new value! Not good :(
# Say as error, with error output
if new_t is None:
c.state = 'waitconsume'
c.exit_status = 2
c.output = '(Error: there is no available check time after time change!)'
c.check_time = time.time()
c.execution_time = 0
else:
c.t_to_go = new_t
ref.next_chk = new_t
# Now all checks and actions
for c in self.sched.actions.values():
# Already launch checks should not be touch
if c.status == 'scheduled':
t_to_go = c.t_to_go
# Event handler do not have ref
ref = getattr(c, 'ref', None)
new_t = max(0, t_to_go + difference)
# Notification should be check with notification_period
if c.is_a == 'notification':
# But it's no so simple, we must match the timeperiod
new_t = ref.notification_period.get_next_valid_time_from_t(new_t)
# And got a creation_time variable too
c.creation_time = c.creation_time + difference
# But maybe no there is no more new value! Not good :(
# Say as error, with error output
if new_t is None:
c.state = 'waitconsume'
c.exit_status = 2
c.output = '(Error: there is no available check time after time change!)'
c.check_time = time.time()
c.execution_time = 0
else:
c.t_to_go = new_t
def setup_new_conf(self):
conf = self.new_conf
self.new_conf = None
self.cur_conf = conf
# Got our name from the globals
if 'receiver_name' in conf['global']:
name = conf['global']['receiver_name']
else:
name = 'Unnamed receiver'
self.name = name
self.log.load_obj(self, name)
print "[%s] Sending us configuration %s" % (self.name, conf)
if not self.have_modules:
self.modules = mods = conf['global']['modules']
self.have_modules = True
logger.log("[%s] We received modules %s " % (self.name, mods))
# Set our giving timezone from arbiter
use_timezone = conf['global']['use_timezone']
if use_timezone != 'NOTSET':
logger.log("[%s] Setting our timezone to" % (self.name, use_timezone))
os.environ['TZ'] = use_timezone
time.tzset()
def get_new_broks(self, type='scheduler'):
# Get teh good links tab for looping..
links = self.get_links_from_type(type)
if links is None:
logger.log('DBG: Type unknown for connexion! %s' % type)
return
# We check for new check in each schedulers and put
# the result in new_checks
for sched_id in links:
try:
con = links[sched_id]['con']
if con is not None: # None = not initilized
tmp_broks = con.get_broks()
for b in tmp_broks.values():
b.instance_id = links[sched_id]['instance_id']
# Ok, we can add theses broks to our queues
self.add_broks_to_queue(tmp_broks.values())
else: # no con? make the connexion
self.pynag_con_init(sched_id, type=type)
# Ok, con is not know, so we create it
except KeyError, exp:
print exp
self.pynag_con_init(sched_id, type=type)
except Pyro.errors.ProtocolError, exp:
logger.log(
"[%s] Connexion problem to the %s %s : %s" %
(self.name, type, links[sched_id]['name'], str(exp)))
links[sched_id]['con'] = None
def setup_new_conf(self):
conf = self.new_conf
self.new_conf = None
self.cur_conf = conf
# Got our name from the globals
if 'receiver_name' in conf['global']:
name = conf['global']['receiver_name']
else:
name = 'Unnamed receiver'
self.name = name
self.log.load_obj(self, name)
print "[%s] Sending us configuration %s" % (self.name, conf)
if not self.have_modules:
self.modules = mods = conf['global']['modules']
self.have_modules = True
logger.log("[%s] We received modules %s " % (self.name, mods))
# Set our giving timezone from arbiter
use_timezone = conf['global']['use_timezone']
if use_timezone != 'NOTSET':
logger.log("[%s] Setting our timezone to" %
(self.name, use_timezone))
os.environ['TZ'] = use_timezone
time.tzset()
def is_me(self):
logger.log(
"And arbiter is launched with the hostname:%s from an arbiter point of view of addr :%s"
% (self.host_name, socket.getfqdn()),
print_it=False,
)
return self.host_name == socket.getfqdn() or self.host_name == socket.gethostname()
def publish_archive(archive):
# Now really publish it
proxy = CONFIG['shinken.io']['proxy']
api_key = CONFIG['shinken.io']['api_key']
# Ok we will push the file with a 10s timeout
c = pycurl.Curl()
c.setopt(c.POST, 1)
c.setopt(c.CONNECTTIMEOUT, 10)
c.setopt(c.TIMEOUT, 10)
if proxy:
c.setopt(c.PROXY, proxy)
c.setopt(c.URL, "http://shinken.io/push")
c.setopt(c.HTTPPOST,
[("api_key", api_key),
("data", (c.FORM_FILE, str(archive), c.FORM_CONTENTTYPE,
"application/x-gzip"))])
response = StringIO()
c.setopt(pycurl.WRITEFUNCTION, response.write)
c.setopt(c.VERBOSE, 1)
c.perform()
r = c.getinfo(pycurl.HTTP_CODE)
c.close()
if r != 200:
logger.error("There was a critical error : %s" % response.getvalue())
sys.exit(2)
else:
ret = json.loads(response.getvalue().replace('\\/', '/'))
status = ret.get('status')
text = ret.get('text')
if status == 200:
logger.log(text)
else:
logger.error(text)
def main(self):
self.set_signal_handler()
logger.log("[%s[%d]]: Now running.." % (self.name, os.getpid()))
while not self.interrupted:
self.do_loop_turn()
self.do_stop()
logger.log("[%s]: exiting now.." % (self.name))
def main(self):
self.load_config_file()
for line in self.get_header():
self.log.log(line)
logger.log("[Broker] Using working directory : %s" % os.path.abspath(self.workdir))
self.do_daemon_init_and_start()
self.uri2 = self.pyro_daemon.register(self.interface, "ForArbiter")
print "The Arbtier uri it at", self.uri2
# We wait for initial conf
self.wait_for_initial_conf()
if not self.new_conf:
return
self.setup_new_conf()
# Set modules, init them and start external ones
self.modules_manager.set_modules(self.modules)
self.do_load_modules()
self.modules_manager.start_external_instances()
# Do the modules part, we have our modules in self.modules
# REF: doc/broker-modules.png (1)
self.hook_point('load_retention')
# Now the main loop
self.do_mainloop()
def main(self):
self.load_config_file()
for line in self.get_header():
self.log.log(line)
logger.log("[Receiver] Using working directory : %s" %
os.path.abspath(self.workdir))
self.do_daemon_init_and_start()
self.uri2 = self.pyro_daemon.register(self.interface, "ForArbiter")
print "The Arbtier uri it at", self.uri2
# We wait for initial conf
self.wait_for_initial_conf()
if not self.new_conf:
return
self.setup_new_conf()
self.modules_manager.set_modules(self.modules)
self.do_load_modules()
# and start external modules too
self.modules_manager.start_external_instances()
# Do the modules part, we have our modules in self.modules
# REF: doc/receiver-modules.png (1)
# Now the main loop
self.do_mainloop()
def main(self):
""" module "main" method. Only used by external modules. """
self.set_signal_handler()
logger.log("[%s[%d]]: Now running.." % (self.name, os.getpid()))
while not self.interrupted:
self.do_loop_turn()
self.do_stop()
logger.log("[%s]: exiting now.." % (self.name))
def create_connection(self):
try:
self.uri = pyro.create_uri(self.address, self.port, "ForArbiter", self.__class__.use_ssl)
self.con = pyro.getProxy(self.uri)
pyro.set_timeout(self.con, self.timeout)
except Pyro_exp_pack , exp:
self.con = None
logger.log('Error : in creation connexion for %s : %s' % (self.get_name(), str(exp)))
def main(self):
""" module "main" method. Only used by external modules. """
self.set_signal_handler()
logger.log("[%s[%d]]: Now running.." % (self.name, os.getpid()))
while not self.interrupted:
self.do_loop_turn()
self.do_stop()
logger.log("[%s]: exiting now.." % (self.name))
def hook_early_configuration(self, arb):
logger.log("[IpTag] in hook late config")
for h in arb.conf.hosts:
if not hasattr(h, 'address') and not hasattr(h, 'host_name'):
continue
# The address to resolve
addr = None
#By default take the address, if not, take host_name
if not hasattr(h, 'address'):
addr = h.host_name
else:
addr = h.address
print "Looking for h", h.get_name()
print addr
h_ip = None
try:
IP(addr)
# If we reach here, it's it was a real IP :)
h_ip = addr
except:
pass
# Ok, try again with name resolution
if not h_ip:
try:
h_ip = socket.gethostbyname(addr)
except:
pass
# Ok, maybe we succeed :)
print "Host ip is:", h_ip
# If we got an ip that match and the object do not already got
# the property, tag it!
if h_ip and h_ip in self.ip_range:
print "Is in the range"
# 2 cases : append or replace.
# append will join with the value if exist
# replace will replace it if NOT existing
if self.method == 'append':
orig_v = getattr(h, self.property, '')
print "Orig_v", orig_v
new_v = ','.join([orig_v, self.value])
print "Newv", new_v
setattr(h, self.property, new_v)
# If it's a poller_tag, remember to also tag commands!
if(self.property == 'poller_tag'):
h.check_command.poller_tag = self.value
if self.method == 'replace':
if not hasattr(h, self.property):
# Ok, set the value!
setattr(h, self.property, self.value)
# If it's a poller_tag, remember to also tag commands!
if(self.property == 'poller_tag'):
h.check_command.poller_tag = self.value
def pynag_con_init(self, id, type="scheduler"):
# Get teh good links tab for looping..
links = self.get_links_from_type(type)
if links is None:
logger.log("DBG: Type unknown for connection! %s" % type)
return
if type == "scheduler":
# If sched is not active, I do not try to init
# it is just useless
is_active = links[id]["active"]
if not is_active:
return
# If we try to connect too much, we slow down our tests
if self.is_connection_try_too_close(links[id]):
return
# Ok, we can now update it
links[id]["last_connection"] = time.time()
# DBG: print "Init connection with", links[id]['uri']
running_id = links[id]["running_id"]
# DBG: print "Running id before connection", running_id
uri = links[id]["uri"]
links[id]["con"] = Pyro.core.getProxyForURI(uri)
try:
# intial ping must be quick
pyro.set_timeout(links[id]["con"], 5)
links[id]["con"].ping()
new_run_id = links[id]["con"].get_running_id()
# data transfert can be longer
pyro.set_timeout(links[id]["con"], 120)
# The schedulers have been restart : it has a new run_id.
# So we clear all verifs, they are obsolete now.
if new_run_id != running_id:
print "[%s] New running id for the %s %s : %s (was %s)" % (
self.name,
type,
links[id]["name"],
new_run_id,
running_id,
)
links[id]["broks"].clear()
# we must ask for a enw full broks if
# it's a scheduler
if type == "scheduler":
print "[%s] I ask for a broks generation to the scheduler %s" % (self.name, links[id]["name"])
links[id]["con"].fill_initial_broks()
# else:
# print "I do nto ask for brok generation"
links[id]["running_id"] = new_run_id
except (Pyro.errors.ProtocolError, Pyro.errors.CommunicationError), exp:
logger.log("[%s] Connexion problem to the %s %s : %s" % (self.name, type, links[id]["name"], str(exp)))
links[id]["con"] = None
return
def register_local_log(self):
# The arbiter don't have such an attribute
if hasattr(self, 'use_local_log') and self.use_local_log:
try:
self.local_log_fd = self.log.register_local_log(self.local_log)
except IOError, exp:
print "Error : opening the log file '%s' failed with '%s'" % (self.local_log, exp)
sys.exit(2)
logger.log("Using the local log file '%s'" % self.local_log)
def stop_process(self):
if self.process:
logger.log("I'm stopping module '%s' process pid:%s " % (self.get_name(), self.process.pid))
self.process.terminate()
self.process.join(timeout=1)
if self.process.is_alive():
logger.log("The process is still alive, I help it to die")
self.__kill()
self.process = None
def pynag_con_init(self, id, type='scheduler'):
# Get teh good links tab for looping..
links = self.get_links_from_type(type)
if links is None:
logger.log('DBG: Type unknown for connexion! %s' % type)
return
if type == 'scheduler':
# If sched is not active, I do not try to init
# it is just useless
is_active = links[id]['active']
if not is_active:
return
# If we try to connect too much, we slow down our tests
if self.is_connexion_try_too_close(links[id]):
return
# Ok, we can now update it
links[id]['last_connexion'] = time.time()
# DBG: print "Init connexion with", links[id]['uri']
running_id = links[id]['running_id']
# DBG: print "Running id before connexion", running_id
uri = links[id]['uri']
links[id]['con'] = Pyro.core.getProxyForURI(uri)
try:
# intial ping must be quick
pyro.set_timeout(links[id]['con'], 5)
links[id]['con'].ping()
new_run_id = links[id]['con'].get_running_id()
# data transfert can be longer
pyro.set_timeout(links[id]['con'], 120)
# The schedulers have been restart : it has a new run_id.
# So we clear all verifs, they are obsolete now.
if new_run_id != running_id:
print "[%s] New running id for the %s %s : %s (was %s)" % (
self.name, type, links[id]['name'], new_run_id, running_id)
links[id]['broks'].clear()
# we must ask for a enw full broks if
# it's a scheduler
if type == 'scheduler':
print "[%s] I ask for a broks generation to the scheduler %s" % (
self.name, links[id]['name'])
links[id]['con'].fill_initial_broks()
# else:
# print "I do nto ask for brok generation"
links[id]['running_id'] = new_run_id
except (Pyro.errors.ProtocolError,
Pyro.errors.CommunicationError), exp:
logger.log("[%s] Connexion problem to the %s %s : %s" %
(self.name, type, links[id]['name'], str(exp)))
links[id]['con'] = None
return
def start(self):
""" Start this module process if it's external. if not -> donothing """
if not self.is_external:
return
self.stop_process()
logger.log("Starting external process for instance %s" % (self.name))
p = self.process = Process(target=self.main, args=())
self.properties['process'] = p ## TODO: temporary
p.start()
logger.log("%s is now started ; pid=%d" % (self.name, p.pid))
def register_local_log(self):
# The arbiter don't have such an attribute
if hasattr(self, 'use_local_log') and self.use_local_log:
try:
self.local_log_fd = self.log.register_local_log(self.local_log)
except IOError, exp:
print "Error : opening the log file '%s' failed with '%s'" % (
self.local_log, exp)
sys.exit(2)
logger.log("Using the local log file '%s'" % self.local_log)
def hook_point(self, hook_name):
for inst in self.modules_manager.instances:
full_hook_name = 'hook_' + hook_name
if hasattr(inst, full_hook_name):
f = getattr(inst, full_hook_name)
try :
f(self)
except Exception, exp:
logger.log('The instance %s raise an exception %s. I disable, and set it to restart later' % (inst.get_name(), str(exp)))
self.modules_manager.set_to_restart(inst)
def start(self):
""" Start this module process if it's external. if not -> donothing """
if not self.is_external:
return
self.stop_process()
logger.log("Starting external process for instance %s" % (self.name))
p = self.process = Process(target=self.main, args=())
self.properties['process'] = p ## TODO: temporary
p.start()
logger.log("%s is now started ; pid=%d" % (self.name, p.pid))
def raise_notification_log_entry(self, n):
contact = n.contact
command = n.command_call
if n.type in ('DOWNTIMESTART', 'DOWNTIMEEND', 'CUSTOM', 'ACKNOWLEDGEMENT', 'FLAPPINGSTART', 'FLAPPINGSTOP', 'FLAPPINGDISABLED'):
state = '%s (%s)' % (n.type, self.state)
else:
state = self.state
if self.__class__.log_notifications:
logger.log("HOST NOTIFICATION: %s;%s;%s;%s;%s" % (contact.get_name(), self.get_name(), state, \
command.get_name(), self.output))
def set_dead(self):
was_alive = self.alive
self.alive = False
self.con = None
# We are dead now. Must raise
# a brok to say it
if was_alive:
logger.log("Warning : Setting the satellite %s to a dead state." % self.get_name())
b = self.get_update_status_brok()
self.broks.append(b)
def add_failed_check_attempt(self, reason=''):
self.reachable = False
self.attempt += 1
self.attempt = min(self.attempt, self.max_check_attempts)
# Don't need to warn again and again if the satellite is already dead
if self.alive:
s = "Info : Add failed attempt to %s (%d/%d) %s" % (self.get_name(), self.attempt, self.max_check_attempts, reason)
logger.log(s)
# check when we just go HARD (dead)
if self.attempt == self.max_check_attempts:
self.set_dead()
def stop_process(self):
""" Request the module process to stop and release it """
if self.process:
logger.log("I'm stopping process pid:%s " % self.process.pid)
self.process.terminate()
self.process.join(timeout=1)
print dir(self.process)
if self.process.is_alive():
logger.log("The process is still alive, I help it to die")
self.__kill()
self.process = None
def stop_process(self):
""" Request the module process to stop and release it """
if self.process:
logger.log("I'm stopping process pid:%s " % self.process.pid)
self.process.terminate()
self.process.join(timeout=1)
print dir(self.process)
if self.process.is_alive():
logger.log("The process is still alive, I help it to die")
self.__kill()
self.process = None
def start_external_instances(self):
for inst in [inst for inst in self.instances if inst.is_external]:
# But maybe the init failed a bit, so bypass this ones from now
if not self.try_instance_init(inst):
logger.log("Warning : the module '%s' failed to init, I will try to restart it later" % inst.get_name())
self.to_restart.append(inst)
continue
# ok, init succeed
logger.log("Starting external module %s" % inst.get_name())
inst.start()
def hook_point(self, hook_name):
for inst in self.modules_manager.instances:
full_hook_name = 'hook_' + hook_name
if hasattr(inst, full_hook_name):
f = getattr(inst, full_hook_name)
try:
f(self)
except Exception, exp:
logger.log(
'The instance %s raise an exception %s. I disable, and set it to restart later'
% (inst.get_name(), str(exp)))
self.modules_manager.set_to_restart(inst)
def start_external_instances(self):
for inst in [inst for inst in self.instances if inst.is_external]:
# But maybe the init failed a bit, so bypass this ones from now
if not self.try_instance_init(inst):
logger.log(
"Warning : the module '%s' failed to init, I will try to restart it later"
% inst.get_name())
self.to_restart.append(inst)
continue
# ok, init succeed
print "Starting external module %s" % inst.get_name(), inst.from_q
inst.start()
def do_stop(self):
if self.modules_manager:
# We save what we can but NOT for the scheduler
# because the current sched object is a dummy one
# and the old one aleady do it!
if not hasattr(self, 'sched'):
self.hook_point('save_retention')
# And we quit
logger.log('Stopping all modules')
self.modules_manager.stop_all()
if self.pyro_daemon:
pyro.shutdown(self.pyro_daemon) #.shutdown(True)
logger.quit()
def wait_for_initial_conf(self, timeout=1.0):
logger.log("Waiting for initial configuration")
cur_timeout = timeout
# Arbiter do not already set our have_conf param
while not self.new_conf and not self.interrupted:
elapsed, _, _ = self.handleRequests(cur_timeout)
if elapsed:
cur_timeout -= elapsed
if cur_timeout > 0:
continue
cur_timeout = timeout
sys.stdout.write(".")
sys.stdout.flush()
def check_alive_instances(self):
#to_del = []
#Only for external
for inst in self.instances:
if not inst in self.to_restart:
if inst.is_external and not inst.process.is_alive():
logger.log(
"Error : the external module %s goes down unexpectly!"
% inst.get_name())
logger.log("Setting the module %s to restart" %
inst.get_name())
# We clean its queues, they are no more useful
inst.clear_queues()
self.to_restart.append(inst)
def is_correct(self):
state = True
properties = self.__class__.properties
# Raised all previously saw errors like unknown contacts and co
if self.configuration_errors != []:
state = False
for err in self.configuration_errors:
logger.log(err)
for prop, entry in properties.items():
if not hasattr(self, prop) and entry.required:
print self.get_name(), "missing property :", prop
state = False
return state
def check_and_do_archive(self, first_pass=False):
now = int(time.time())
#first check if the file last mod (or creation) was
#not our day
try:
t_last_mod = int(float(str(os.path.getmtime(self.path))))
except OSError: #there should be no path from now, so no move :)
return False
#print "Ctime %d" % os.path.getctime(self.path)
t_last_mod_day = get_day(t_last_mod)
today = get_day(now)
#print "Dates: t_last_mod : %d, t_last_mod_day: %d, today : %d" % (t_last_mod, t_last_mod_day, today)
if t_last_mod_day != today:
logger.log("We are archiving the old log file")
#For the first pass, it's not already open
if not first_pass:
self.file.close()
#Now we move it
#Get a new name like MM
#f_name is like nagios.log
f_name = os.path.basename(self.path)
#remove the ext -> (nagios,.log)
(f_base_name, ext) = os.path.splitext(f_name)
#make the good looking day for archive name
#like -05-09-2010-00
d = datetime.datetime.fromtimestamp(today)
s_day = d.strftime("-%m-%d-%Y-00")
archive_name = f_base_name + s_day + ext
file_archive_path = os.path.join(self.archive_path, archive_name)
logger.log("Moving the old log file from %s to %s" %
(self.path, file_archive_path))
shutil.move(self.path, file_archive_path)
#and we overwrite it
print "I open the log file %s" % self.path
self.file = open(self.path, 'a')
return True
return False
def test_utf8log(self):
sutf = 'h\351h\351' # Latin Small Letter E with acute in Latin-1
logger.log(sutf)
sutf8 = u'I love myself $£¤' # dollar, pound, currency
logger.log(sutf8)
s = unichr(40960) + u'abcd' + unichr(1972)
logger.log(s)
def dump_memory(self):
logger.log("I dump my memory, it can ask some seconds to do")
try:
from guppy import hpy
hp = hpy()
logger.log(hp.heap())
except ImportError:
logger.log(
'I do not have the module guppy for memory dump, please install it'
)
def check_bad_dispatch(self):
for elt in self.elements:
if hasattr(elt, 'conf'):
# If element have a conf, I do not care, it's a good dispatch
# If die : I do not ask it something, it won't respond..
if elt.conf is None and elt.reachable:
# print "Ask", elt.get_name() , 'if it got conf'
if elt.have_conf():
logger.log(
'Warning : The element %s have a conf and should not have one! I ask it to idle now'
% elt.get_name())
elt.active = False
elt.wait_new_conf()
# I do not care about order not send or not. If not,
# The next loop wil resent it
# else:
# print "No conf"
# I ask satellite witch sched_id they manage. If I am not agree, I ask
# them to remove it
for satellite in self.satellites:
kind = satellite.get_my_type()
if satellite.reachable:
cfg_ids = satellite.what_i_managed()
# I do nto care about satellites that do nothing, it already
# do what I want :)
if len(cfg_ids) != 0:
id_to_delete = []
for cfg_id in cfg_ids:
# DBG print kind, ":", satellite.get_name(), "manage cfg id:", cfg_id
# Ok, we search for realm that have the conf
for r in self.realms:
if cfg_id in r.confs:
# Ok we've got the realm, we check it's to_satellites_managed_by
# to see if reactionner is in. If not, we remove he sched_id for it
if not satellite in r.to_satellites_managed_by[
kind][cfg_id]:
id_to_delete.append(cfg_id)
# Maybe we removed all cfg_id of this reactionner
# We can make it idle, no active and wait_new_conf
if len(id_to_delete) == len(cfg_ids):
satellite.active = False
logger.log("I ask %s to wait a new conf" %
satellite.get_name())
satellite.wait_new_conf()
else: #It is not fully idle, just less cfg
for id in id_to_delete:
logger.log(
"I ask to remove configuration N%d from %s" %
(cfg_id, satellite.get_name()))
satellite.remove_from_conf(cfg_id)
def manage_brok(self, b):
# Call all modules if they catch the call
for mod in self.modules_manager.get_internal_instances():
try:
mod.manage_brok(b)
except Exception, exp:
print exp.__dict__
logger.log(
"[%s] Warning : The mod %s raise an exception: %s, I'm tagging it to restart later"
% (self.name, mod.get_name(), str(exp)))
logger.log("[%s] Exception type : %s" % (self.name, type(exp)))
logger.log("Back trace of this kill: %s" %
(traceback.format_exc()))
self.modules_manager.set_to_restart(inst)
def manage_brok(self, b):
to_del = []
# Call all modules if they catch the call
for mod in self.modules_manager.get_internal_instances():
try:
mod.manage_brok(b)
except Exception, exp:
print exp.__dict__
logger.log(
"[%s] Warning : The mod %s raise an exception: %s, I kill it"
% (self.name, mod.get_name(), str(exp)))
logger.log("[%s] Exception type : %s" % (self.name, type(exp)))
logger.log("Back trace of this kill: %s" %
(traceback.format_exc()))
to_del.append(mod)
def get_instances(self):
""" Create, init and then returns the list of module instances that the caller needs.
If an instance can't be created or init'ed then only log is done. That instance is skipped.
The previous modules instance(s), if any, are all cleaned. """
self.clear_instances()
for (mod_conf, module) in self.modules_assoc:
try:
mod_conf.properties = module.properties.copy()
inst = module.get_instance(mod_conf)
if inst is None: #None = Bad thing happened :)
logger.log("get_instance for module %s returned None !" %
(mod_conf.get_name()))
continue
assert (isinstance(inst, BaseModule))
self.instances.append(inst)
except Exception, exp:
logger.log(
"Error : the module %s raised an exception %s, I remove it!"
% (mod_conf.get_name(), str(exp)))
output = cStringIO.StringIO()
traceback.print_exc(file=output)
logger.log("Back trace of this remove : %s" %
(output.getvalue()))
output.close()
def hook_save_retention(self, daemon):
logger.log(
"[NagiosRetention] asking me to update the retention objects, but I won't do it."
)