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 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 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 put_conf(self, conf):
if self.con is None:
self.create_connection()
#print "Connection is OK, now we put conf", conf
#print "Try to put conf:", conf
try:
pyro.set_timeout(self.con, self.data_timeout)
self.con.put_conf(conf)
pyro.set_timeout(self.con, self.timeout)
return True
except Pyro_exp_pack , exp:
self.con = None
#print ''.join(Pyro.util.getPyroTraceback(exp))
return False
def put_conf(self, conf):
if self.con is None:
self.create_connexion()
#print "Connexion is OK, now we put conf", conf
#print "Try to put conf:", conf
try:
pyro.set_timeout(self.con, self.data_timeout)
print "DBG: put conf to", self.con.__dict__
self.con.put_conf(conf)
pyro.set_timeout(self.con, self.timeout)
return True
except Pyro_exp_pack, exp:
self.con = None
#print ''.join(Pyro.util.getPyroTraceback(exp))
return False
def create_connection(self):
try:
self.uri = pyro.create_uri(self.address, self.port, "ForArbiter", self.__class__.use_ssl)
# By default Pyro got problem in connect() function that can take
# long seconds to raise a timeout. And even with the _setTimeout()
# call. So we change the whole default connect() timeout
socket.setdefaulttimeout(self.timeout)
self.con = pyro.getProxy(self.uri)
# But the multiprocessing module is not copatible with it!
# so we must disable it imadiatly after
socket.setdefaulttimeout(None)
pyro.set_timeout(self.con, self.timeout)
except Pyro_exp_pack , exp:
# But the multiprocessing module is not copatible with it!
# so we must disable it imadiatly after
socket.setdefaulttimeout(None)
self.con = None
logger.log('Error : in creation connection for %s : %s' % (self.get_name(), str(exp)))
def put_conf(self, conf):
if self.con is None:
self.create_connection()
#print "Connection is OK, now we put conf", conf
#print "Try to put conf:", conf
# Maybe the connexion was not ok, bail out
if not self.con:
return False
try:
pyro.set_timeout(self.con, self.data_timeout)
self.con.put_conf(conf)
pyro.set_timeout(self.con, self.timeout)
return True
except Pyro_exp_pack, exp:
self.con = None
logger.error(''.join(PYRO_VERSION < "4.0" and Pyro.util.getPyroTraceback(exp) or Pyro.util.getPyroTraceback()))
return False
def create_connection(self):
try:
self.uri = pyro.create_uri(self.arb_satmap['address'],
self.arb_satmap['port'], "ForArbiter",
self.__class__.use_ssl)
# By default Pyro got problem in connect() function that can take
# long seconds to raise a timeout. And even with the _setTimeout()
# call. So we change the whole default connect() timeout
socket.setdefaulttimeout(self.timeout)
self.con = pyro.getProxy(self.uri)
# But the multiprocessing module is not copatible with it!
# so we must disable it imadiatly after
socket.setdefaulttimeout(None)
pyro.set_timeout(self.con, self.timeout)
except Pyro_exp_pack, exp:
# But the multiprocessing module is not compatible with it!
# so we must disable it imadiatly after
socket.setdefaulttimeout(None)
self.con = None
logger.error("Creating connection for %s: %s" %
(self.get_name(), str(exp)))
def put_conf(self, conf):
if self.con is None:
self.create_connection()
#print "Connection is OK, now we put conf", conf
#print "Try to put conf:", conf
# Maybe the connexion was not ok, bail out
if not self.con:
return False
try:
pyro.set_timeout(self.con, self.data_timeout)
self.con.put_conf(conf)
pyro.set_timeout(self.con, self.timeout)
return True
except Pyro_exp_pack, exp:
self.con = None
logger.error(''.join(
PYRO_VERSION < "4.0" and Pyro.util.getPyroTraceback(exp)
or Pyro.util.getPyroTraceback()))
return False
raise SystemExit(CRITICAL)
elif options.target not in daemon_types:
print 'CRITICAL - target', options.target, 'is not a Shinken daemon!'
parser.print_help()
raise SystemExit(CRITICAL)
uri = pyro.create_uri(options.hostname, options.portnum, PYRO_OBJECT,
options.ssl)
# Set the default socket connection to the timeout, by default it's 10s
socket.setdefaulttimeout(float(options.timeout))
con = None
try:
con = Pyro.core.getProxyForURI(uri)
pyro.set_timeout(con, options.timeout)
except Exception, exp:
print "CRITICAL : the Arbiter is not reachable : (%s)." % exp
raise SystemExit(CRITICAL)
if options.daemon:
# We just want a check for a single satellite daemon
# Only OK or CRITICAL here
daemon_name = options.daemon
try:
result = con.get_satellite_status(options.target, daemon_name)
except Exception, exp:
print "CRITICAL : the Arbiter is not reachable : (%s)." % exp
raise SystemExit(CRITICAL)
def create_connection(self):
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)
class Broker(BaseSatellite):
properties = BaseSatellite.properties.copy()
properties.update({
'pidfile': PathProp(default='brokerd.pid'),
'port': IntegerProp(default='7772'),
'local_log': PathProp(default='brokerd.log'),
})
def __init__(self, config_file, is_daemon, do_replace, debug, debug_file):
super(Broker, self).__init__('broker', config_file, is_daemon,
do_replace, debug, debug_file)
# Our arbiters
self.arbiters = {}
# Our pollers and reactionners
self.pollers = {}
self.reactionners = {}
# Modules are load one time
self.have_modules = False
# Can have a queue of external_commands given by modules
# will be processed by arbiter
self.external_commands = []
# All broks to manage
self.broks = [] # broks to manage
# broks raised this turn and that needs to be put in self.broks
self.broks_internal_raised = []
self.timeout = 1.0
# Schedulers have some queues. We can simplify the call by adding
# elements into the proper queue just by looking at their type
# Brok -> self.broks
# TODO: better tag ID?
# External commands -> self.external_commands
def add(self, elt):
cls_type = elt.__class__.my_type
if cls_type == 'brok':
# For brok, we TAG brok with our instance_id
elt.instance_id = 0
self.broks_internal_raised.append(elt)
return
elif cls_type == 'externalcommand':
logger.debug("Enqueuing an external command '%s'" %
str(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!
logger.debug(str(elt.__dict__))
if elt.get_type() == 'NeedData':
data = elt.get_data()
# Full instance id means: I got no data for this scheduler
# so give me all dumbass!
if 'full_instance_id' in data:
c_id = data['full_instance_id']
source = elt.source
logger.info(
'The module %s is asking me to get all initial data from the scheduler %d'
% (source, c_id))
# so we just reset the connection and 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
logger.warning(
"the module %s ask me a full_instance_id for an unknown ID (%d)!"
% (source, c_id))
# Maybe a module tells me that it's dead, I must log it's last words...
if elt.get_type() == 'ICrash':
data = elt.get_data()
logger.error(
'the module %s just crash! Please look at the traceback:' %
data['name'])
logger.error(data['trace'])
# The module death will be looked for elsewhere and restarted.
# Get the good tabs for links by the kind. If unknown, return None
def get_links_from_type(self, type):
t = {'scheduler': self.schedulers, 'arbiter': self.arbiters, \
'poller': self.pollers, 'reactionner': self.reactionners}
if type in t:
return t[type]
return None
# Call by arbiter to get our external commands
def get_external_commands(self):
res = self.external_commands
self.external_commands = []
return res
# Check if we do not connect to often to this
def is_connection_try_too_close(self, elt):
now = time.time()
last_connection = elt['last_connection']
if now - last_connection < 5:
return True
return False
# initialize or re-initialize connection with scheduler or
# arbiter if type == arbiter
def pynag_con_init(self, id, type='scheduler'):
# Get the good links tab for looping..
links = self.get_links_from_type(type)
if links is None:
logger.debug('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']
try:
socket.setdefaulttimeout(3)
links[id]['con'] = Pyro.core.getProxyForURI(uri)
socket.setdefaulttimeout(None)
except Pyro_exp_pack, exp:
# But the multiprocessing module is not compatible with it!
# so we must disable it immediately after
socket.setdefaulttimeout(None)
logger.info("Connection problem to the %s %s: %s" %
(type, links[id]['name'], str(exp)))
links[id]['con'] = None
return
try:
# initial 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 transfer can be longer
pyro.set_timeout(links[id]['con'], 120)
# The schedulers have been restarted: it has a new run_id.
# So we clear all verifs, they are obsolete now.
if new_run_id != running_id:
logger.debug("[%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 new full broks if
# it's a scheduler
if type == 'scheduler':
logger.debug(
"[%s] I ask for a broks generation to the scheduler %s"
% (self.name, links[id]['name']))
links[id]['con'].fill_initial_broks(self.name)
# Ok all is done, we can save this new running id
links[id]['running_id'] = new_run_id
except Pyro_exp_pack, exp:
logger.info("Connection problem to the %s %s: %s" %
(type, links[id]['name'], str(exp)))
links[id]['con'] = None
return
parser.print_help()
raise SystemExit, CRITICAL
elif options.target not in daemon_types:
print 'CRITICAL - target %s is not a Shinken daemon!' % options.target
parser.print_help()
raise SystemExit, CRITICAL
uri = pyro.create_uri(options.hostname, options.portnum, PYRO_OBJECT , options.ssl)
# Set the default socekt connexion to the timeout, by default it's 10s
socket.setdefaulttimeout(float(options.timeout))
con = None
try:
con = Pyro.core.getProxyForURI(uri)
pyro.set_timeout(con, float(options.timeout))
except Exception, exp:
print "CRITICAL : the Arbiter is not reachable : (%s)." % exp
sys.exit(CRITICAL)
if options.daemon:
# We just want a check for a single satellite daemon
# Only OK or CRITICAL here
daemon_name = options.daemon
try:
result = con.get_satellite_status(options.target, daemon_name)
except Exception, exp:
print "CRITICAL : the Arbiter is not reachable : (%s)." % exp
raise SystemExit, CRITICAL
def create_connexion(self):
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)
raise SystemExit(CRITICAL)
elif options.target not in daemon_types:
print 'CRITICAL - target', options.target, 'is not a Shinken daemon!'
parser.print_help()
raise SystemExit(CRITICAL)
uri = pyro.create_uri(options.hostname, options.portnum, PYRO_OBJECT,
options.ssl)
# Set the default socket connection to the timeout, by default it's 10s
socket.setdefaulttimeout(float(options.timeout))
con = None
try:
con = Pyro.core.getProxyForURI(uri)
pyro.set_timeout(con, options.timeout)
except Exception, exp:
print "CRITICAL : the Arbiter is not reachable : (%s)." % exp
raise SystemExit(CRITICAL)
if options.daemon:
# We just want a check for a single satellite daemon
# Only OK or CRITICAL here
daemon_name = options.daemon
try:
result = con.get_satellite_status(options.target, daemon_name)
except Exception, exp:
print "CRITICAL : the Arbiter is not reachable : (%s)." % exp
raise SystemExit(CRITICAL)
raise SystemExit(CRITICAL)
elif options.target not in daemon_types:
print 'CRITICAL - target', options.target, 'is not a Shinken daemon!'
parser.print_help()
raise SystemExit(CRITICAL)
uri = pyro.create_uri(options.hostname, options.portnum, PYRO_OBJECT,
options.ssl)
# Set the default socket connection to the timeout, by default it's 10s
socket.setdefaulttimeout(float(options.timeout))
con = None
try:
con = Pyro.core.getProxyForURI(uri)
pyro.set_timeout(con, options.timeout)
except Exception, exp:
print "CRITICAL : the Arbiter is not reachable : (%s)." % exp
raise SystemExit(CRITICAL)
if options.daemon:
# We just want a check for a single satellite daemon
# Only OK or CRITICAL here
daemon_name = options.daemon
try:
result = con.get_satellite_status(options.target, daemon_name)
except Exception, exp:
print "CRITICAL : the Arbiter is not reachable : (%s)." % exp
raise SystemExit(CRITICAL)
if result:
