4 # Copyright (C) 2006, 2007, 2008, 2010, 2011, 2012 Google Inc.
6 # This program is free software; you can redistribute it and/or modify
7 # it under the terms of the GNU General Public License as published by
8 # the Free Software Foundation; either version 2 of the License, or
9 # (at your option) any later version.
11 # This program is distributed in the hope that it will be useful, but
12 # WITHOUT ANY WARRANTY; without even the implied warranty of
13 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 # General Public License for more details.
16 # You should have received a copy of the GNU General Public License
17 # along with this program; if not, write to the Free Software
18 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
22 """Functions to bootstrap a new cluster.
32 from ganeti import rpc
33 from ganeti import ssh
34 from ganeti import utils
35 from ganeti import errors
36 from ganeti import config
37 from ganeti import constants
38 from ganeti import objects
39 from ganeti import ssconf
40 from ganeti import serializer
41 from ganeti import hypervisor
42 from ganeti import bdev
43 from ganeti import netutils
44 from ganeti import backend
45 from ganeti import luxi
46 from ganeti import jstore
49 # ec_id for InitConfig's temporary reservation manager
50 _INITCONF_ECID = "initconfig-ecid"
52 #: After how many seconds daemon must be responsive
53 _DAEMON_READY_TIMEOUT = 10.0
57 """Setup the SSH configuration for the cluster.
59 This generates a dsa keypair for root, adds the pub key to the
60 permitted hosts and adds the hostkey to its own known hosts.
63 priv_key, pub_key, auth_keys = ssh.GetUserFiles(constants.GANETI_RUNAS)
65 for name in priv_key, pub_key:
66 if os.path.exists(name):
67 utils.CreateBackup(name)
68 utils.RemoveFile(name)
70 result = utils.RunCmd(["ssh-keygen", "-t", "dsa",
74 raise errors.OpExecError("Could not generate ssh keypair, error %s" %
77 utils.AddAuthorizedKey(auth_keys, utils.ReadFile(pub_key))
80 def GenerateHmacKey(file_name):
81 """Writes a new HMAC key.
84 @param file_name: Path to output file
87 utils.WriteFile(file_name, data="%s\n" % utils.GenerateSecret(), mode=0400,
91 def GenerateClusterCrypto(new_cluster_cert, new_rapi_cert, new_spice_cert,
92 new_confd_hmac_key, new_cds,
93 rapi_cert_pem=None, spice_cert_pem=None,
94 spice_cacert_pem=None, cds=None,
95 nodecert_file=constants.NODED_CERT_FILE,
96 rapicert_file=constants.RAPI_CERT_FILE,
97 spicecert_file=constants.SPICE_CERT_FILE,
98 spicecacert_file=constants.SPICE_CACERT_FILE,
99 hmackey_file=constants.CONFD_HMAC_KEY,
100 cds_file=constants.CLUSTER_DOMAIN_SECRET_FILE):
101 """Updates the cluster certificates, keys and secrets.
103 @type new_cluster_cert: bool
104 @param new_cluster_cert: Whether to generate a new cluster certificate
105 @type new_rapi_cert: bool
106 @param new_rapi_cert: Whether to generate a new RAPI certificate
107 @type new_spice_cert: bool
108 @param new_spice_cert: Whether to generate a new SPICE certificate
109 @type new_confd_hmac_key: bool
110 @param new_confd_hmac_key: Whether to generate a new HMAC key
112 @param new_cds: Whether to generate a new cluster domain secret
113 @type rapi_cert_pem: string
114 @param rapi_cert_pem: New RAPI certificate in PEM format
115 @type spice_cert_pem: string
116 @param spice_cert_pem: New SPICE certificate in PEM format
117 @type spice_cacert_pem: string
118 @param spice_cacert_pem: Certificate of the CA that signed the SPICE
119 certificate, in PEM format
121 @param cds: New cluster domain secret
122 @type nodecert_file: string
123 @param nodecert_file: optional override of the node cert file path
124 @type rapicert_file: string
125 @param rapicert_file: optional override of the rapi cert file path
126 @type spicecert_file: string
127 @param spicecert_file: optional override of the spice cert file path
128 @type spicecacert_file: string
129 @param spicecacert_file: optional override of the spice CA cert file path
130 @type hmackey_file: string
131 @param hmackey_file: optional override of the hmac key file path
134 # noded SSL certificate
135 cluster_cert_exists = os.path.exists(nodecert_file)
136 if new_cluster_cert or not cluster_cert_exists:
137 if cluster_cert_exists:
138 utils.CreateBackup(nodecert_file)
140 logging.debug("Generating new cluster certificate at %s", nodecert_file)
141 utils.GenerateSelfSignedSslCert(nodecert_file)
144 if new_confd_hmac_key or not os.path.exists(hmackey_file):
145 logging.debug("Writing new confd HMAC key to %s", hmackey_file)
146 GenerateHmacKey(hmackey_file)
149 rapi_cert_exists = os.path.exists(rapicert_file)
152 # Assume rapi_pem contains a valid PEM-formatted certificate and key
153 logging.debug("Writing RAPI certificate at %s", rapicert_file)
154 utils.WriteFile(rapicert_file, data=rapi_cert_pem, backup=True)
156 elif new_rapi_cert or not rapi_cert_exists:
158 utils.CreateBackup(rapicert_file)
160 logging.debug("Generating new RAPI certificate at %s", rapicert_file)
161 utils.GenerateSelfSignedSslCert(rapicert_file)
164 spice_cert_exists = os.path.exists(spicecert_file)
165 spice_cacert_exists = os.path.exists(spicecacert_file)
167 # spice_cert_pem implies also spice_cacert_pem
168 logging.debug("Writing SPICE certificate at %s", spicecert_file)
169 utils.WriteFile(spicecert_file, data=spice_cert_pem, backup=True)
170 logging.debug("Writing SPICE CA certificate at %s", spicecacert_file)
171 utils.WriteFile(spicecacert_file, data=spice_cacert_pem, backup=True)
172 elif new_spice_cert or not spice_cert_exists:
173 if spice_cert_exists:
174 utils.CreateBackup(spicecert_file)
175 if spice_cacert_exists:
176 utils.CreateBackup(spicecacert_file)
178 logging.debug("Generating new self-signed SPICE certificate at %s",
180 (_, cert_pem) = utils.GenerateSelfSignedSslCert(spicecert_file)
182 # Self-signed certificate -> the public certificate is also the CA public
184 logging.debug("Writing the public certificate to %s",
186 utils.io.WriteFile(spicecacert_file, mode=0400, data=cert_pem)
188 # Cluster domain secret
190 logging.debug("Writing cluster domain secret to %s", cds_file)
191 utils.WriteFile(cds_file, data=cds, backup=True)
193 elif new_cds or not os.path.exists(cds_file):
194 logging.debug("Generating new cluster domain secret at %s", cds_file)
195 GenerateHmacKey(cds_file)
198 def _InitGanetiServerSetup(master_name):
199 """Setup the necessary configuration for the initial node daemon.
201 This creates the nodepass file containing the shared password for
202 the cluster, generates the SSL certificate and starts the node daemon.
204 @type master_name: str
205 @param master_name: Name of the master node
208 # Generate cluster secrets
209 GenerateClusterCrypto(True, False, False, False, False)
211 result = utils.RunCmd([constants.DAEMON_UTIL, "start", constants.NODED])
213 raise errors.OpExecError("Could not start the node daemon, command %s"
214 " had exitcode %s and error %s" %
215 (result.cmd, result.exit_code, result.output))
217 _WaitForNodeDaemon(master_name)
220 def _WaitForNodeDaemon(node_name):
221 """Wait for node daemon to become responsive.
224 def _CheckNodeDaemon():
225 # Pylint bug <http://www.logilab.org/ticket/35642>
226 # pylint: disable=E1101
227 result = rpc.BootstrapRunner().call_version([node_name])[node_name]
229 raise utils.RetryAgain()
232 utils.Retry(_CheckNodeDaemon, 1.0, _DAEMON_READY_TIMEOUT)
233 except utils.RetryTimeout:
234 raise errors.OpExecError("Node daemon on %s didn't answer queries within"
235 " %s seconds" % (node_name, _DAEMON_READY_TIMEOUT))
238 def _WaitForMasterDaemon():
239 """Wait for master daemon to become responsive.
242 def _CheckMasterDaemon():
245 (cluster_name, ) = cl.QueryConfigValues(["cluster_name"])
247 raise utils.RetryAgain()
249 logging.debug("Received cluster name %s from master", cluster_name)
252 utils.Retry(_CheckMasterDaemon, 1.0, _DAEMON_READY_TIMEOUT)
253 except utils.RetryTimeout:
254 raise errors.OpExecError("Master daemon didn't answer queries within"
255 " %s seconds" % _DAEMON_READY_TIMEOUT)
258 def _InitFileStorage(file_storage_dir):
259 """Initialize if needed the file storage.
261 @param file_storage_dir: the user-supplied value
262 @return: either empty string (if file storage was disabled at build
263 time) or the normalized path to the storage directory
266 file_storage_dir = os.path.normpath(file_storage_dir)
268 if not os.path.isabs(file_storage_dir):
269 raise errors.OpPrereqError("File storage directory '%s' is not an absolute"
270 " path" % file_storage_dir, errors.ECODE_INVAL)
272 if not os.path.exists(file_storage_dir):
274 os.makedirs(file_storage_dir, 0750)
276 raise errors.OpPrereqError("Cannot create file storage directory"
277 " '%s': %s" % (file_storage_dir, err),
278 errors.ECODE_ENVIRON)
280 if not os.path.isdir(file_storage_dir):
281 raise errors.OpPrereqError("The file storage directory '%s' is not"
282 " a directory." % file_storage_dir,
283 errors.ECODE_ENVIRON)
284 return file_storage_dir
287 def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914
288 master_netmask, master_netdev, file_storage_dir,
289 shared_file_storage_dir, candidate_pool_size, secondary_ip=None,
290 vg_name=None, beparams=None, nicparams=None, ndparams=None,
291 hvparams=None, diskparams=None, enabled_hypervisors=None,
292 modify_etc_hosts=True, modify_ssh_setup=True,
293 maintain_node_health=False, drbd_helper=None, uid_pool=None,
294 default_iallocator=None, primary_ip_version=None, ipolicy=None,
295 prealloc_wipe_disks=False, use_external_mip_script=False,
296 hv_state=None, disk_state=None):
297 """Initialise the cluster.
299 @type candidate_pool_size: int
300 @param candidate_pool_size: master candidate pool size
303 # TODO: complete the docstring
304 if config.ConfigWriter.IsCluster():
305 raise errors.OpPrereqError("Cluster is already initialised",
308 if not enabled_hypervisors:
309 raise errors.OpPrereqError("Enabled hypervisors list must contain at"
310 " least one member", errors.ECODE_INVAL)
311 invalid_hvs = set(enabled_hypervisors) - constants.HYPER_TYPES
313 raise errors.OpPrereqError("Enabled hypervisors contains invalid"
314 " entries: %s" % invalid_hvs,
318 ipcls = netutils.IPAddress.GetClassFromIpVersion(primary_ip_version)
319 except errors.ProgrammerError:
320 raise errors.OpPrereqError("Invalid primary ip version: %d." %
323 hostname = netutils.GetHostname(family=ipcls.family)
324 if not ipcls.IsValid(hostname.ip):
325 raise errors.OpPrereqError("This host's IP (%s) is not a valid IPv%d"
326 " address." % (hostname.ip, primary_ip_version))
328 if ipcls.IsLoopback(hostname.ip):
329 raise errors.OpPrereqError("This host's IP (%s) resolves to a loopback"
330 " address. Please fix DNS or %s." %
331 (hostname.ip, constants.ETC_HOSTS),
332 errors.ECODE_ENVIRON)
334 if not ipcls.Own(hostname.ip):
335 raise errors.OpPrereqError("Inconsistency: this host's name resolves"
336 " to %s,\nbut this ip address does not"
337 " belong to this host" %
338 hostname.ip, errors.ECODE_ENVIRON)
340 clustername = netutils.GetHostname(name=cluster_name, family=ipcls.family)
342 if netutils.TcpPing(clustername.ip, constants.DEFAULT_NODED_PORT, timeout=5):
343 raise errors.OpPrereqError("Cluster IP already active",
344 errors.ECODE_NOTUNIQUE)
347 if primary_ip_version == constants.IP6_VERSION:
348 raise errors.OpPrereqError("When using a IPv6 primary address, a valid"
349 " IPv4 address must be given as secondary",
351 secondary_ip = hostname.ip
353 if not netutils.IP4Address.IsValid(secondary_ip):
354 raise errors.OpPrereqError("Secondary IP address (%s) has to be a valid"
355 " IPv4 address." % secondary_ip,
358 if not netutils.IP4Address.Own(secondary_ip):
359 raise errors.OpPrereqError("You gave %s as secondary IP,"
360 " but it does not belong to this host." %
361 secondary_ip, errors.ECODE_ENVIRON)
363 if master_netmask is not None:
364 if not ipcls.ValidateNetmask(master_netmask):
365 raise errors.OpPrereqError("CIDR netmask (%s) not valid for IPv%s " %
366 (master_netmask, primary_ip_version))
368 master_netmask = ipcls.iplen
370 if vg_name is not None:
371 # Check if volume group is valid
372 vgstatus = utils.CheckVolumeGroupSize(utils.ListVolumeGroups(), vg_name,
373 constants.MIN_VG_SIZE)
375 raise errors.OpPrereqError("Error: %s\nspecify --no-lvm-storage if"
376 " you are not using lvm" % vgstatus,
379 if drbd_helper is not None:
381 curr_helper = bdev.BaseDRBD.GetUsermodeHelper()
382 except errors.BlockDeviceError, err:
383 raise errors.OpPrereqError("Error while checking drbd helper"
384 " (specify --no-drbd-storage if you are not"
385 " using drbd): %s" % str(err),
386 errors.ECODE_ENVIRON)
387 if drbd_helper != curr_helper:
388 raise errors.OpPrereqError("Error: requiring %s as drbd helper but %s"
389 " is the current helper" % (drbd_helper,
393 if constants.ENABLE_FILE_STORAGE:
394 file_storage_dir = _InitFileStorage(file_storage_dir)
396 file_storage_dir = ""
398 if constants.ENABLE_SHARED_FILE_STORAGE:
399 shared_file_storage_dir = _InitFileStorage(shared_file_storage_dir)
401 shared_file_storage_dir = ""
403 if not re.match("^[0-9a-z]{2}:[0-9a-z]{2}:[0-9a-z]{2}$", mac_prefix):
404 raise errors.OpPrereqError("Invalid mac prefix given '%s'" % mac_prefix,
407 result = utils.RunCmd(["ip", "link", "show", "dev", master_netdev])
409 raise errors.OpPrereqError("Invalid master netdev given (%s): '%s'" %
411 result.output.strip()), errors.ECODE_INVAL)
413 dirs = [(constants.RUN_GANETI_DIR, constants.RUN_DIRS_MODE)]
414 utils.EnsureDirs(dirs)
416 objects.UpgradeBeParams(beparams)
417 utils.ForceDictType(beparams, constants.BES_PARAMETER_TYPES)
418 utils.ForceDictType(nicparams, constants.NICS_PARAMETER_TYPES)
420 objects.NIC.CheckParameterSyntax(nicparams)
422 full_ipolicy = objects.FillIPolicy(constants.IPOLICY_DEFAULTS, ipolicy)
424 if ndparams is not None:
425 utils.ForceDictType(ndparams, constants.NDS_PARAMETER_TYPES)
427 ndparams = dict(constants.NDC_DEFAULTS)
429 # This is ugly, as we modify the dict itself
430 # FIXME: Make utils.ForceDictType pure functional or write a wrapper
433 for hvname, hvs_data in hv_state.items():
434 utils.ForceDictType(hvs_data, constants.HVSTS_PARAMETER_TYPES)
435 hv_state[hvname] = objects.Cluster.SimpleFillHvState(hvs_data)
437 hv_state = dict((hvname, constants.HVST_DEFAULTS)
438 for hvname in enabled_hypervisors)
440 # FIXME: disk_state has no default values yet
442 for storage, ds_data in disk_state.items():
443 if storage not in constants.DS_VALID_TYPES:
444 raise errors.OpPrereqError("Invalid storage type in disk state: %s" %
445 storage, errors.ECODE_INVAL)
446 for ds_name, state in ds_data.items():
447 utils.ForceDictType(state, constants.DSS_PARAMETER_TYPES)
448 ds_data[ds_name] = objects.Cluster.SimpleFillDiskState(state)
450 # hvparams is a mapping of hypervisor->hvparams dict
451 for hv_name, hv_params in hvparams.iteritems():
452 utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES)
453 hv_class = hypervisor.GetHypervisor(hv_name)
454 hv_class.CheckParameterSyntax(hv_params)
456 # diskparams is a mapping of disk-template->diskparams dict
457 for template, dt_params in diskparams.items():
458 param_keys = set(dt_params.keys())
459 default_param_keys = set(constants.DISK_DT_DEFAULTS[template].keys())
460 if not (param_keys <= default_param_keys):
461 unknown_params = param_keys - default_param_keys
462 raise errors.OpPrereqError("Invalid parameters for disk template %s:"
464 utils.CommaJoin(unknown_params)))
465 utils.ForceDictType(dt_params, constants.DISK_DT_TYPES)
467 # set up ssh config and /etc/hosts
468 sshline = utils.ReadFile(constants.SSH_HOST_RSA_PUB)
469 sshkey = sshline.split(" ")[1]
472 utils.AddHostToEtcHosts(hostname.name, hostname.ip)
477 if default_iallocator is not None:
478 alloc_script = utils.FindFile(default_iallocator,
479 constants.IALLOCATOR_SEARCH_PATH,
481 if alloc_script is None:
482 raise errors.OpPrereqError("Invalid default iallocator script '%s'"
483 " specified" % default_iallocator,
485 elif constants.HTOOLS:
486 # htools was enabled at build-time, we default to it
487 if utils.FindFile(constants.IALLOC_HAIL,
488 constants.IALLOCATOR_SEARCH_PATH,
490 default_iallocator = constants.IALLOC_HAIL
494 # init of cluster config file
495 cluster_config = objects.Cluster(
497 rsahostkeypub=sshkey,
498 highest_used_port=(constants.FIRST_DRBD_PORT - 1),
499 mac_prefix=mac_prefix,
500 volume_group_name=vg_name,
501 tcpudp_port_pool=set(),
502 master_node=hostname.name,
503 master_ip=clustername.ip,
504 master_netmask=master_netmask,
505 master_netdev=master_netdev,
506 cluster_name=clustername.name,
507 file_storage_dir=file_storage_dir,
508 shared_file_storage_dir=shared_file_storage_dir,
509 enabled_hypervisors=enabled_hypervisors,
510 beparams={constants.PP_DEFAULT: beparams},
511 nicparams={constants.PP_DEFAULT: nicparams},
514 diskparams=diskparams,
515 candidate_pool_size=candidate_pool_size,
516 modify_etc_hosts=modify_etc_hosts,
517 modify_ssh_setup=modify_ssh_setup,
521 maintain_node_health=maintain_node_health,
522 drbd_usermode_helper=drbd_helper,
523 default_iallocator=default_iallocator,
524 primary_ip_family=ipcls.family,
525 prealloc_wipe_disks=prealloc_wipe_disks,
526 use_external_mip_script=use_external_mip_script,
527 ipolicy=full_ipolicy,
528 hv_state_static=hv_state,
529 disk_state_static=disk_state,
531 master_node_config = objects.Node(name=hostname.name,
532 primary_ip=hostname.ip,
533 secondary_ip=secondary_ip,
535 master_candidate=True,
536 offline=False, drained=False,
537 ctime=now, mtime=now,
539 InitConfig(constants.CONFIG_VERSION, cluster_config, master_node_config)
540 cfg = config.ConfigWriter(offline=True)
541 ssh.WriteKnownHostsFile(cfg, constants.SSH_KNOWN_HOSTS_FILE)
542 cfg.Update(cfg.GetClusterInfo(), logging.error)
543 backend.WriteSsconfFiles(cfg.GetSsconfValues())
545 # set up the inter-node password and certificate
546 _InitGanetiServerSetup(hostname.name)
548 logging.debug("Starting daemons")
549 result = utils.RunCmd([constants.DAEMON_UTIL, "start-all"])
551 raise errors.OpExecError("Could not start daemons, command %s"
552 " had exitcode %s and error %s" %
553 (result.cmd, result.exit_code, result.output))
555 _WaitForMasterDaemon()
558 def InitConfig(version, cluster_config, master_node_config,
559 cfg_file=constants.CLUSTER_CONF_FILE):
560 """Create the initial cluster configuration.
562 It will contain the current node, which will also be the master
563 node, and no instances.
566 @param version: configuration version
567 @type cluster_config: L{objects.Cluster}
568 @param cluster_config: cluster configuration
569 @type master_node_config: L{objects.Node}
570 @param master_node_config: master node configuration
571 @type cfg_file: string
572 @param cfg_file: configuration file path
575 uuid_generator = config.TemporaryReservationManager()
576 cluster_config.uuid = uuid_generator.Generate([], utils.NewUUID,
578 master_node_config.uuid = uuid_generator.Generate([], utils.NewUUID,
581 master_node_config.name: master_node_config,
583 default_nodegroup = objects.NodeGroup(
584 uuid=uuid_generator.Generate([], utils.NewUUID, _INITCONF_ECID),
585 name=constants.INITIAL_NODE_GROUP_NAME,
586 members=[master_node_config.name],
590 default_nodegroup.uuid: default_nodegroup,
593 config_data = objects.ConfigData(version=version,
594 cluster=cluster_config,
595 nodegroups=nodegroups,
599 ctime=now, mtime=now)
600 utils.WriteFile(cfg_file,
601 data=serializer.Dump(config_data.ToDict()),
605 def FinalizeClusterDestroy(master):
606 """Execute the last steps of cluster destroy
608 This function shuts down all the daemons, completing the destroy
609 begun in cmdlib.LUDestroyOpcode.
612 cfg = config.ConfigWriter()
613 modify_ssh_setup = cfg.GetClusterInfo().modify_ssh_setup
614 runner = rpc.BootstrapRunner()
616 master_params = cfg.GetMasterNetworkParameters()
617 master_params.name = master
618 ems = cfg.GetUseExternalMipScript()
619 result = runner.call_node_deactivate_master_ip(master_params.name,
622 msg = result.fail_msg
624 logging.warning("Could not disable the master IP: %s", msg)
626 result = runner.call_node_stop_master(master)
627 msg = result.fail_msg
629 logging.warning("Could not disable the master role: %s", msg)
631 result = runner.call_node_leave_cluster(master, modify_ssh_setup)
632 msg = result.fail_msg
634 logging.warning("Could not shutdown the node daemon and cleanup"
635 " the node: %s", msg)
638 def SetupNodeDaemon(cluster_name, node, ssh_key_check):
639 """Add a node to the cluster.
641 This function must be called before the actual opcode, and will ssh
642 to the remote node, copy the needed files, and start ganeti-noded,
643 allowing the master to do the rest via normal rpc calls.
645 @param cluster_name: the cluster name
646 @param node: the name of the new node
647 @param ssh_key_check: whether to do a strict key check
650 family = ssconf.SimpleStore().GetPrimaryIPFamily()
651 sshrunner = ssh.SshRunner(cluster_name,
652 ipv6=(family == netutils.IP6Address.family))
654 bind_address = constants.IP4_ADDRESS_ANY
655 if family == netutils.IP6Address.family:
656 bind_address = constants.IP6_ADDRESS_ANY
658 # set up inter-node password and certificate and restarts the node daemon
659 # and then connect with ssh to set password and start ganeti-noded
660 # note that all the below variables are sanitized at this point,
661 # either by being constants or by the checks above
662 sshrunner.CopyFileToNode(node, constants.NODED_CERT_FILE)
663 sshrunner.CopyFileToNode(node, constants.RAPI_CERT_FILE)
664 sshrunner.CopyFileToNode(node, constants.SPICE_CERT_FILE)
665 sshrunner.CopyFileToNode(node, constants.SPICE_CACERT_FILE)
666 sshrunner.CopyFileToNode(node, constants.CONFD_HMAC_KEY)
667 mycommand = ("%s stop-all; %s start %s -b %s" %
668 (constants.DAEMON_UTIL, constants.DAEMON_UTIL, constants.NODED,
669 utils.ShellQuote(bind_address)))
671 result = sshrunner.Run(node, "root", mycommand, batch=False,
672 ask_key=ssh_key_check,
673 use_cluster_key=True,
674 strict_host_check=ssh_key_check)
676 raise errors.OpExecError("Remote command on node %s, error: %s,"
678 (node, result.fail_reason, result.output))
680 _WaitForNodeDaemon(node)
683 def MasterFailover(no_voting=False):
684 """Failover the master node.
686 This checks that we are not already the master, and will cause the
687 current master to cease being master, and the non-master to become
690 @type no_voting: boolean
691 @param no_voting: force the operation without remote nodes agreement
695 sstore = ssconf.SimpleStore()
697 old_master, new_master = ssconf.GetMasterAndMyself(sstore)
698 node_list = sstore.GetNodeList()
699 mc_list = sstore.GetMasterCandidates()
701 if old_master == new_master:
702 raise errors.OpPrereqError("This commands must be run on the node"
703 " where you want the new master to be."
704 " %s is already the master" %
705 old_master, errors.ECODE_INVAL)
707 if new_master not in mc_list:
708 mc_no_master = [name for name in mc_list if name != old_master]
709 raise errors.OpPrereqError("This node is not among the nodes marked"
710 " as master candidates. Only these nodes"
711 " can become masters. Current list of"
712 " master candidates is:\n"
713 "%s" % ("\n".join(mc_no_master)),
717 vote_list = GatherMasterVotes(node_list)
720 voted_master = vote_list[0][0]
721 if voted_master is None:
722 raise errors.OpPrereqError("Cluster is inconsistent, most nodes did"
723 " not respond.", errors.ECODE_ENVIRON)
724 elif voted_master != old_master:
725 raise errors.OpPrereqError("I have a wrong configuration, I believe"
726 " the master is %s but the other nodes"
727 " voted %s. Please resync the configuration"
729 (old_master, voted_master),
735 logging.info("Setting master to %s, old master: %s", new_master, old_master)
738 # instantiate a real config writer, as we now know we have the
740 cfg = config.ConfigWriter(accept_foreign=True)
742 cluster_info = cfg.GetClusterInfo()
743 cluster_info.master_node = new_master
744 # this will also regenerate the ssconf files, since we updated the
746 cfg.Update(cluster_info, logging.error)
747 except errors.ConfigurationError, err:
748 logging.error("Error while trying to set the new master: %s",
752 # if cfg.Update worked, then it means the old master daemon won't be
753 # able now to write its own config file (we rely on locking in both
754 # backend.UploadFile() and ConfigWriter._Write(); hence the next
755 # step is to kill the old master
757 logging.info("Stopping the master daemon on node %s", old_master)
759 runner = rpc.BootstrapRunner()
760 master_params = cfg.GetMasterNetworkParameters()
761 master_params.name = old_master
762 ems = cfg.GetUseExternalMipScript()
763 result = runner.call_node_deactivate_master_ip(master_params.name,
766 msg = result.fail_msg
768 logging.warning("Could not disable the master IP: %s", msg)
770 result = runner.call_node_stop_master(old_master)
771 msg = result.fail_msg
773 logging.error("Could not disable the master role on the old master"
774 " %s, please disable manually: %s", old_master, msg)
776 logging.info("Checking master IP non-reachability...")
778 master_ip = sstore.GetMasterIP()
781 # Here we have a phase where no master should be running
783 if netutils.TcpPing(master_ip, constants.DEFAULT_NODED_PORT):
784 raise utils.RetryAgain()
787 utils.Retry(_check_ip, (1, 1.5, 5), total_timeout)
788 except utils.RetryTimeout:
789 logging.warning("The master IP is still reachable after %s seconds,"
790 " continuing but activating the master on the current"
791 " node will probably fail", total_timeout)
793 if jstore.CheckDrainFlag():
794 logging.info("Undraining job queue")
795 jstore.SetDrainFlag(False)
797 logging.info("Starting the master daemons on the new master")
799 result = rpc.BootstrapRunner().call_node_start_master_daemons(new_master,
801 msg = result.fail_msg
803 logging.error("Could not start the master role on the new master"
804 " %s, please check: %s", new_master, msg)
807 logging.info("Master failed over from %s to %s", old_master, new_master)
812 """Returns the current master node.
814 This is a separate function in bootstrap since it's needed by
815 gnt-cluster, and instead of importing directly ssconf, it's better
816 to abstract it in bootstrap, where we do use ssconf in other
820 sstore = ssconf.SimpleStore()
822 old_master, _ = ssconf.GetMasterAndMyself(sstore)
827 def GatherMasterVotes(node_list):
828 """Check the agreement on who is the master.
830 This function will return a list of (node, number of votes), ordered
831 by the number of votes. Errors will be denoted by the key 'None'.
833 Note that the sum of votes is the number of nodes this machine
834 knows, whereas the number of entries in the list could be different
835 (if some nodes vote for another master).
837 We remove ourselves from the list since we know that (bugs aside)
838 since we use the same source for configuration information for both
839 backend and boostrap, we'll always vote for ourselves.
841 @type node_list: list
842 @param node_list: the list of nodes to query for master info; the current
843 node will be removed if it is in the list
845 @return: list of (node, votes)
848 myself = netutils.Hostname.GetSysName()
850 node_list.remove(myself)
854 # no nodes left (eventually after removing myself)
856 results = rpc.BootstrapRunner().call_master_info(node_list)
857 if not isinstance(results, dict):
858 # this should not happen (unless internal error in rpc)
859 logging.critical("Can't complete rpc call, aborting master startup")
860 return [(None, len(node_list))]
868 logging.warning("Error contacting node %s: %s", node, msg)
870 # for now we accept both length 3, 4 and 5 (data[3] is primary ip version
871 # and data[4] is the master netmask)
872 elif not isinstance(data, (tuple, list)) or len(data) < 3:
873 logging.warning("Invalid data received from node %s: %s", node, data)
876 if None not in votes:
880 master_node = data[2]
881 if master_node not in votes:
882 votes[master_node] = 0
883 votes[master_node] += 1
885 vote_list = [v for v in votes.items()]
886 # sort first on number of votes then on name, since we want None
887 # sorted later if we have the half of the nodes not responding, and
888 # half voting all for the same master
889 vote_list.sort(key=lambda x: (x[1], x[0]), reverse=True)