def format():
if DiskHeartbeat.is_in_use():
nodes=DiskHeartbeat().get_all_ts().keys()
print "Warning: Heartbeat disk is in use by:", ", ".join(nodes)
if(raw_input("Proceed ? [y/N]:").upper() != "Y"):
print "Aborded by user."
raise SystemExit(0)
DiskHeartbeat.format()
print "Device", core.cfg['HB_DISK'], "formatted."
def joinCluster(self):
def startHeartbeats():
self._startSlave()
self.s_rpc.startService()
if self.role == MasterService.RL_ACTIVE:
self._startMaster()
def joinRefused(reason):
reason.trap(NodeRefusedError, RPCRefusedError)
log.err("Join to cluster %s failed: Master %s has refused me: %s" %
(core.cfg['CLUSTER_NAME'], self.master, reason.getErrorMessage()))
self.stopService()
def joinAccepted(result):
self.role=MasterService.RL_PASSIVE
log.info("Join successfull, I'm now part of cluster %s." % (core.cfg['CLUSTER_NAME']))
startHeartbeats()
def masterConnected(obj):
d = obj.callRemote("register",DNSCache.getInstance().name)
d.addCallbacks(joinAccepted,joinRefused)
d.addErrback(log.err)
d.addBoth(lambda _: rpcConnector.disconnect())
return d
try:
if self.master is None:
# New active master
if DNSCache.getInstance().name not in core.cfg['ALLOWED_NODES']:
log.warn("I'm not allowed to create a new cluster. Exiting.")
raise Exception("Cluster creation not allowed")
if DiskHeartbeat.is_in_use():
log.err("Heartbeat disk is in use but we are alone !")
raise Exception("Heartbeat disk already in use")
log.info("No master found. I'm now the new master of %s." % (core.cfg['CLUSTER_NAME']))
self.role=MasterService.RL_ACTIVE
self.master=DNSCache.getInstance().name
self.status[self.master]={'timestamp': 0, 'offset': 0, 'vms': []}
self.disk.make_slot(DNSCache.getInstance().name)
startHeartbeats()
else:
# Passive master
self.role=MasterService.RL_JOINING
log.info("Trying to join cluster %s..." % (core.cfg['CLUSTER_NAME']))
factory = pb.PBClientFactory()
rpcConnector = reactor.connectTCP(self.master, core.cfg['TCP_PORT'], factory)
d = factory.getRootObject()
d.addCallback(masterConnected)
d.addErrback(log.err)
except Exception, e:
log.err("Startup failed: %s. Shutting down." % (e))
self.stopService()
def __init__(self):
self.role = MasterService.RL_ALONE # Current role of this node
self.state = MasterService.ST_NORMAL # Current cluster error status
self.master = None # Name of the active master
self.masterLastSeen = 0 # Timestamp for master failover
self.status = dict() # Whole cluster status
self.localNode = Node(DNSCache.getInstance().name)
self.disk = DiskHeartbeat()
self.s_slaveHb = SlaveHearbeatService(self)
self.s_masterHb = MasterHeartbeatService(self)
self.s_rpc = RPCService(self)
# Watchdogs for failover
self.l_slaveDog = task.LoopingCall(self.checkMasterHeartbeat)
self.l_masterDog = task.LoopingCall(self.checkSlaveHeartbeats)
# Election Stuff
self.ballotBox = None # All received votes
self.currentElection = None # Election name, none if no pending election
self.f_tally = None # IDelayedCall used to trigger countVotes()
self.lastTallyDate = 0 # Timestamp for debbuging elections
self.panicRequested = False # True if panic is requested during election
def joinCluster(self):
def startHeartbeats():
self._startSlave()
self.s_rpc.startService()
if self.role == MasterService.RL_ACTIVE:
self._startMaster()
def joinRefused(reason):
reason.trap(NodeRefusedError, RPCRefusedError)
log.err("Join to cluster %s failed: Master %s has refused me: %s" %
(core.cfg['CLUSTER_NAME'], self.master,
reason.getErrorMessage()))
self.stopService()
def joinAccepted(result):
self.role = MasterService.RL_PASSIVE
log.info("Join successfull, I'm now part of cluster %s." %
(core.cfg['CLUSTER_NAME']))
startHeartbeats()
def masterConnected(obj):
d = obj.callRemote("register", DNSCache.getInstance().name)
d.addCallbacks(joinAccepted, joinRefused)
d.addErrback(log.err)
d.addBoth(lambda _: rpcConnector.disconnect())
return d
try:
if self.master is None:
# New active master
if DNSCache.getInstance(
).name not in core.cfg['ALLOWED_NODES']:
log.warn(
"I'm not allowed to create a new cluster. Exiting.")
raise Exception("Cluster creation not allowed")
if DiskHeartbeat.is_in_use():
log.err("Heartbeat disk is in use but we are alone !")
raise Exception("Heartbeat disk already in use")
log.info("No master found. I'm now the new master of %s." %
(core.cfg['CLUSTER_NAME']))
self.role = MasterService.RL_ACTIVE
self.master = DNSCache.getInstance().name
self.status[self.master] = {
'timestamp': 0,
'offset': 0,
'vms': []
}
self.disk.make_slot(DNSCache.getInstance().name)
startHeartbeats()
else:
# Passive master
self.role = MasterService.RL_JOINING
log.info("Trying to join cluster %s..." %
(core.cfg['CLUSTER_NAME']))
factory = pb.PBClientFactory()
rpcConnector = reactor.connectTCP(self.master,
core.cfg['TCP_PORT'],
factory)
d = factory.getRootObject()
d.addCallback(masterConnected)
d.addErrback(log.err)
except Exception, e:
log.err("Startup failed: %s. Shutting down." % (e))
self.stopService()
