4 # Copyright (C) 2006, 2007 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 """Configuration management for Ganeti
24 This module provides the interface to the Ganeti cluster configuration.
26 The configuration data is stored on every node but is updated on the master
27 only. After each update, the master distributes the data to the other nodes.
29 Currently, the data storage format is JSON. YAML was slow and consuming too
39 from ganeti import errors
40 from ganeti import locking
41 from ganeti import utils
42 from ganeti import constants
43 from ganeti import rpc
44 from ganeti import objects
45 from ganeti import serializer
48 _config_lock = locking.SharedLock()
51 def _ValidateConfig(data):
52 """Verifies that a configuration objects looks valid.
54 This only verifies the version of the configuration.
56 @raise errors.ConfigurationError: if the version differs from what
60 if data.version != constants.CONFIG_VERSION:
61 raise errors.ConfigurationError("Cluster configuration version"
62 " mismatch, got %s instead of %s" %
64 constants.CONFIG_VERSION))
68 """The interface to the cluster configuration.
71 def __init__(self, cfg_file=None, offline=False):
73 self._lock = _config_lock
74 self._config_data = None
75 self._offline = offline
77 self._cfg_file = constants.CLUSTER_CONF_FILE
79 self._cfg_file = cfg_file
80 self._temporary_ids = set()
81 self._temporary_drbds = {}
82 self._temporary_macs = set()
83 # Note: in order to prevent errors when resolving our name in
84 # _DistributeConfig, we compute it here once and reuse it; it's
85 # better to raise an error before starting to modify the config
86 # file than after it was modified
87 self._my_hostname = utils.HostInfo().name
88 self._last_cluster_serial = -1
91 # this method needs to be static, so that we can call it on the class
94 """Check if the cluster is configured.
97 return os.path.exists(constants.CLUSTER_CONF_FILE)
99 @locking.ssynchronized(_config_lock, shared=1)
100 def GenerateMAC(self):
101 """Generate a MAC for an instance.
103 This should check the current instances for duplicates.
106 prefix = self._config_data.cluster.mac_prefix
107 all_macs = self._AllMACs()
110 byte1 = random.randrange(0, 256)
111 byte2 = random.randrange(0, 256)
112 byte3 = random.randrange(0, 256)
113 mac = "%s:%02x:%02x:%02x" % (prefix, byte1, byte2, byte3)
114 if mac not in all_macs and mac not in self._temporary_macs:
118 raise errors.ConfigurationError("Can't generate unique MAC")
119 self._temporary_macs.add(mac)
122 @locking.ssynchronized(_config_lock, shared=1)
123 def IsMacInUse(self, mac):
124 """Predicate: check if the specified MAC is in use in the Ganeti cluster.
126 This only checks instances managed by this cluster, it does not
127 check for potential collisions elsewhere.
130 all_macs = self._AllMACs()
131 return mac in all_macs or mac in self._temporary_macs
133 @locking.ssynchronized(_config_lock, shared=1)
134 def GenerateDRBDSecret(self):
135 """Generate a DRBD secret.
137 This checks the current disks for duplicates.
140 all_secrets = self._AllDRBDSecrets()
143 secret = utils.GenerateSecret()
144 if secret not in all_secrets:
148 raise errors.ConfigurationError("Can't generate unique DRBD secret")
151 def _ComputeAllLVs(self):
152 """Compute the list of all LVs.
156 for instance in self._config_data.instances.values():
157 node_data = instance.MapLVsByNode()
158 for lv_list in node_data.values():
159 lvnames.update(lv_list)
162 @locking.ssynchronized(_config_lock, shared=1)
163 def GenerateUniqueID(self, exceptions=None):
164 """Generate an unique disk name.
166 This checks the current node, instances and disk names for
169 @param exceptions: a list with some other names which should be checked
170 for uniqueness (used for example when you want to get
171 more than one id at one time without adding each one in
172 turn to the config file)
175 @return: the unique id
179 existing.update(self._temporary_ids)
180 existing.update(self._ComputeAllLVs())
181 existing.update(self._config_data.instances.keys())
182 existing.update(self._config_data.nodes.keys())
183 if exceptions is not None:
184 existing.update(exceptions)
187 unique_id = utils.NewUUID()
188 if unique_id not in existing and unique_id is not None:
191 raise errors.ConfigurationError("Not able generate an unique ID"
192 " (last tried ID: %s" % unique_id)
193 self._temporary_ids.add(unique_id)
197 """Return all MACs present in the config.
200 @return: the list of all MACs
204 for instance in self._config_data.instances.values():
205 for nic in instance.nics:
206 result.append(nic.mac)
210 def _AllDRBDSecrets(self):
211 """Return all DRBD secrets present in the config.
214 @return: the list of all DRBD secrets
217 def helper(disk, result):
218 """Recursively gather secrets from this disk."""
219 if disk.dev_type == constants.DT_DRBD8:
220 result.append(disk.logical_id[5])
222 for child in disk.children:
223 helper(child, result)
226 for instance in self._config_data.instances.values():
227 for disk in instance.disks:
232 def _CheckDiskIDs(self, disk, l_ids, p_ids):
233 """Compute duplicate disk IDs
235 @type disk: L{objects.Disk}
236 @param disk: the disk at which to start searching
238 @param l_ids: list of current logical ids
240 @param p_ids: list of current physical ids
242 @return: a list of error messages
246 if disk.logical_id in l_ids:
247 result.append("duplicate logical id %s" % str(disk.logical_id))
249 l_ids.append(disk.logical_id)
250 if disk.physical_id in p_ids:
251 result.append("duplicate physical id %s" % str(disk.physical_id))
253 p_ids.append(disk.physical_id)
256 for child in disk.children:
257 result.extend(self._CheckDiskIDs(child, l_ids, p_ids))
260 def _UnlockedVerifyConfig(self):
264 @return: a list of error messages; a non-empty list signifies
271 data = self._config_data
274 for instance_name in data.instances:
275 instance = data.instances[instance_name]
276 if instance.primary_node not in data.nodes:
277 result.append("instance '%s' has invalid primary node '%s'" %
278 (instance_name, instance.primary_node))
279 for snode in instance.secondary_nodes:
280 if snode not in data.nodes:
281 result.append("instance '%s' has invalid secondary node '%s'" %
282 (instance_name, snode))
283 for idx, nic in enumerate(instance.nics):
284 if nic.mac in seen_macs:
285 result.append("instance '%s' has NIC %d mac %s duplicate" %
286 (instance_name, idx, nic.mac))
288 seen_macs.append(nic.mac)
290 # gather the drbd ports for duplicate checks
291 for dsk in instance.disks:
292 if dsk.dev_type in constants.LDS_DRBD:
293 tcp_port = dsk.logical_id[2]
294 if tcp_port not in ports:
296 ports[tcp_port].append((instance.name, "drbd disk %s" % dsk.iv_name))
297 # gather network port reservation
298 net_port = getattr(instance, "network_port", None)
299 if net_port is not None:
300 if net_port not in ports:
302 ports[net_port].append((instance.name, "network port"))
304 # instance disk verify
305 for idx, disk in enumerate(instance.disks):
306 result.extend(["instance '%s' disk %d error: %s" %
307 (instance.name, idx, msg) for msg in disk.Verify()])
308 result.extend(self._CheckDiskIDs(disk, seen_lids, seen_pids))
310 # cluster-wide pool of free ports
311 for free_port in data.cluster.tcpudp_port_pool:
312 if free_port not in ports:
313 ports[free_port] = []
314 ports[free_port].append(("cluster", "port marked as free"))
316 # compute tcp/udp duplicate ports
322 txt = ", ".join(["%s/%s" % val for val in pdata])
323 result.append("tcp/udp port %s has duplicates: %s" % (pnum, txt))
325 # highest used tcp port check
327 if keys[-1] > data.cluster.highest_used_port:
328 result.append("Highest used port mismatch, saved %s, computed %s" %
329 (data.cluster.highest_used_port, keys[-1]))
331 if not data.nodes[data.cluster.master_node].master_candidate:
332 result.append("Master node is not a master candidate")
334 # master candidate checks
335 mc_now, mc_max = self._UnlockedGetMasterCandidateStats()
337 result.append("Not enough master candidates: actual %d, target %d" %
341 for node in data.nodes.values():
342 if [node.master_candidate, node.drained, node.offline].count(True) > 1:
343 result.append("Node %s state is invalid: master_candidate=%s,"
344 " drain=%s, offline=%s" %
345 (node.name, node.master_candidate, node.drain,
349 d_map, duplicates = self._UnlockedComputeDRBDMap()
350 for node, minor, instance_a, instance_b in duplicates:
351 result.append("DRBD minor %d on node %s is assigned twice to instances"
352 " %s and %s" % (minor, node, instance_a, instance_b))
356 @locking.ssynchronized(_config_lock, shared=1)
357 def VerifyConfig(self):
360 This is just a wrapper over L{_UnlockedVerifyConfig}.
363 @return: a list of error messages; a non-empty list signifies
367 return self._UnlockedVerifyConfig()
369 def _UnlockedSetDiskID(self, disk, node_name):
370 """Convert the unique ID to the ID needed on the target nodes.
372 This is used only for drbd, which needs ip/port configuration.
374 The routine descends down and updates its children also, because
375 this helps when the only the top device is passed to the remote
378 This function is for internal use, when the config lock is already held.
382 for child in disk.children:
383 self._UnlockedSetDiskID(child, node_name)
385 if disk.logical_id is None and disk.physical_id is not None:
387 if disk.dev_type == constants.LD_DRBD8:
388 pnode, snode, port, pminor, sminor, secret = disk.logical_id
389 if node_name not in (pnode, snode):
390 raise errors.ConfigurationError("DRBD device not knowing node %s" %
392 pnode_info = self._UnlockedGetNodeInfo(pnode)
393 snode_info = self._UnlockedGetNodeInfo(snode)
394 if pnode_info is None or snode_info is None:
395 raise errors.ConfigurationError("Can't find primary or secondary node"
396 " for %s" % str(disk))
397 p_data = (pnode_info.secondary_ip, port)
398 s_data = (snode_info.secondary_ip, port)
399 if pnode == node_name:
400 disk.physical_id = p_data + s_data + (pminor, secret)
401 else: # it must be secondary, we tested above
402 disk.physical_id = s_data + p_data + (sminor, secret)
404 disk.physical_id = disk.logical_id
407 @locking.ssynchronized(_config_lock)
408 def SetDiskID(self, disk, node_name):
409 """Convert the unique ID to the ID needed on the target nodes.
411 This is used only for drbd, which needs ip/port configuration.
413 The routine descends down and updates its children also, because
414 this helps when the only the top device is passed to the remote
418 return self._UnlockedSetDiskID(disk, node_name)
420 @locking.ssynchronized(_config_lock)
421 def AddTcpUdpPort(self, port):
422 """Adds a new port to the available port pool.
425 if not isinstance(port, int):
426 raise errors.ProgrammerError("Invalid type passed for port")
428 self._config_data.cluster.tcpudp_port_pool.add(port)
431 @locking.ssynchronized(_config_lock, shared=1)
432 def GetPortList(self):
433 """Returns a copy of the current port list.
436 return self._config_data.cluster.tcpudp_port_pool.copy()
438 @locking.ssynchronized(_config_lock)
439 def AllocatePort(self):
442 The port will be taken from the available port pool or from the
443 default port range (and in this case we increase
447 # If there are TCP/IP ports configured, we use them first.
448 if self._config_data.cluster.tcpudp_port_pool:
449 port = self._config_data.cluster.tcpudp_port_pool.pop()
451 port = self._config_data.cluster.highest_used_port + 1
452 if port >= constants.LAST_DRBD_PORT:
453 raise errors.ConfigurationError("The highest used port is greater"
454 " than %s. Aborting." %
455 constants.LAST_DRBD_PORT)
456 self._config_data.cluster.highest_used_port = port
461 def _UnlockedComputeDRBDMap(self):
462 """Compute the used DRBD minor/nodes.
465 @return: dictionary of node_name: dict of minor: instance_name;
466 the returned dict will have all the nodes in it (even if with
467 an empty list), and a list of duplicates; if the duplicates
468 list is not empty, the configuration is corrupted and its caller
469 should raise an exception
472 def _AppendUsedPorts(instance_name, disk, used):
474 if disk.dev_type == constants.LD_DRBD8 and len(disk.logical_id) >= 5:
475 nodeA, nodeB, dummy, minorA, minorB = disk.logical_id[:5]
476 for node, port in ((nodeA, minorA), (nodeB, minorB)):
477 assert node in used, ("Node '%s' of instance '%s' not found"
478 " in node list" % (node, instance_name))
479 if port in used[node]:
480 duplicates.append((node, port, instance_name, used[node][port]))
482 used[node][port] = instance_name
484 for child in disk.children:
485 duplicates.extend(_AppendUsedPorts(instance_name, child, used))
489 my_dict = dict((node, {}) for node in self._config_data.nodes)
490 for instance in self._config_data.instances.itervalues():
491 for disk in instance.disks:
492 duplicates.extend(_AppendUsedPorts(instance.name, disk, my_dict))
493 for (node, minor), instance in self._temporary_drbds.iteritems():
494 if minor in my_dict[node] and my_dict[node][minor] != instance:
495 duplicates.append((node, minor, instance, my_dict[node][minor]))
497 my_dict[node][minor] = instance
498 return my_dict, duplicates
500 @locking.ssynchronized(_config_lock)
501 def ComputeDRBDMap(self):
502 """Compute the used DRBD minor/nodes.
504 This is just a wrapper over L{_UnlockedComputeDRBDMap}.
506 @return: dictionary of node_name: dict of minor: instance_name;
507 the returned dict will have all the nodes in it (even if with
511 d_map, duplicates = self._UnlockedComputeDRBDMap()
513 raise errors.ConfigurationError("Duplicate DRBD ports detected: %s" %
517 @locking.ssynchronized(_config_lock)
518 def AllocateDRBDMinor(self, nodes, instance):
519 """Allocate a drbd minor.
521 The free minor will be automatically computed from the existing
522 devices. A node can be given multiple times in order to allocate
523 multiple minors. The result is the list of minors, in the same
524 order as the passed nodes.
526 @type instance: string
527 @param instance: the instance for which we allocate minors
530 assert isinstance(instance, basestring), \
531 "Invalid argument '%s' passed to AllocateDRBDMinor" % instance
533 d_map, duplicates = self._UnlockedComputeDRBDMap()
535 raise errors.ConfigurationError("Duplicate DRBD ports detected: %s" %
541 # no minors used, we can start at 0
544 self._temporary_drbds[(nname, 0)] = instance
548 ffree = utils.FirstFree(keys)
550 # return the next minor
551 # TODO: implement high-limit check
555 # double-check minor against current instances
556 assert minor not in d_map[nname], \
557 ("Attempt to reuse allocated DRBD minor %d on node %s,"
558 " already allocated to instance %s" %
559 (minor, nname, d_map[nname][minor]))
560 ndata[minor] = instance
561 # double-check minor against reservation
562 r_key = (nname, minor)
563 assert r_key not in self._temporary_drbds, \
564 ("Attempt to reuse reserved DRBD minor %d on node %s,"
565 " reserved for instance %s" %
566 (minor, nname, self._temporary_drbds[r_key]))
567 self._temporary_drbds[r_key] = instance
569 logging.debug("Request to allocate drbd minors, input: %s, returning %s",
573 def _UnlockedReleaseDRBDMinors(self, instance):
574 """Release temporary drbd minors allocated for a given instance.
576 @type instance: string
577 @param instance: the instance for which temporary minors should be
581 assert isinstance(instance, basestring), \
582 "Invalid argument passed to ReleaseDRBDMinors"
583 for key, name in self._temporary_drbds.items():
585 del self._temporary_drbds[key]
587 @locking.ssynchronized(_config_lock)
588 def ReleaseDRBDMinors(self, instance):
589 """Release temporary drbd minors allocated for a given instance.
591 This should be called on the error paths, on the success paths
592 it's automatically called by the ConfigWriter add and update
595 This function is just a wrapper over L{_UnlockedReleaseDRBDMinors}.
597 @type instance: string
598 @param instance: the instance for which temporary minors should be
602 self._UnlockedReleaseDRBDMinors(instance)
604 @locking.ssynchronized(_config_lock, shared=1)
605 def GetConfigVersion(self):
606 """Get the configuration version.
608 @return: Config version
611 return self._config_data.version
613 @locking.ssynchronized(_config_lock, shared=1)
614 def GetClusterName(self):
617 @return: Cluster name
620 return self._config_data.cluster.cluster_name
622 @locking.ssynchronized(_config_lock, shared=1)
623 def GetMasterNode(self):
624 """Get the hostname of the master node for this cluster.
626 @return: Master hostname
629 return self._config_data.cluster.master_node
631 @locking.ssynchronized(_config_lock, shared=1)
632 def GetMasterIP(self):
633 """Get the IP of the master node for this cluster.
638 return self._config_data.cluster.master_ip
640 @locking.ssynchronized(_config_lock, shared=1)
641 def GetMasterNetdev(self):
642 """Get the master network device for this cluster.
645 return self._config_data.cluster.master_netdev
647 @locking.ssynchronized(_config_lock, shared=1)
648 def GetFileStorageDir(self):
649 """Get the file storage dir for this cluster.
652 return self._config_data.cluster.file_storage_dir
654 @locking.ssynchronized(_config_lock, shared=1)
655 def GetHypervisorType(self):
656 """Get the hypervisor type for this cluster.
659 return self._config_data.cluster.default_hypervisor
661 @locking.ssynchronized(_config_lock, shared=1)
662 def GetHostKey(self):
663 """Return the rsa hostkey from the config.
666 @return: the rsa hostkey
669 return self._config_data.cluster.rsahostkeypub
671 @locking.ssynchronized(_config_lock)
672 def AddInstance(self, instance):
673 """Add an instance to the config.
675 This should be used after creating a new instance.
677 @type instance: L{objects.Instance}
678 @param instance: the instance object
681 if not isinstance(instance, objects.Instance):
682 raise errors.ProgrammerError("Invalid type passed to AddInstance")
684 if instance.disk_template != constants.DT_DISKLESS:
685 all_lvs = instance.MapLVsByNode()
686 logging.info("Instance '%s' DISK_LAYOUT: %s", instance.name, all_lvs)
688 instance.serial_no = 1
689 self._config_data.instances[instance.name] = instance
690 self._config_data.cluster.serial_no += 1
691 self._UnlockedReleaseDRBDMinors(instance.name)
692 for nic in instance.nics:
693 self._temporary_macs.discard(nic.mac)
696 def _SetInstanceStatus(self, instance_name, status):
697 """Set the instance's status to a given value.
700 assert isinstance(status, bool), \
701 "Invalid status '%s' passed to SetInstanceStatus" % (status,)
703 if instance_name not in self._config_data.instances:
704 raise errors.ConfigurationError("Unknown instance '%s'" %
706 instance = self._config_data.instances[instance_name]
707 if instance.admin_up != status:
708 instance.admin_up = status
709 instance.serial_no += 1
712 @locking.ssynchronized(_config_lock)
713 def MarkInstanceUp(self, instance_name):
714 """Mark the instance status to up in the config.
717 self._SetInstanceStatus(instance_name, True)
719 @locking.ssynchronized(_config_lock)
720 def RemoveInstance(self, instance_name):
721 """Remove the instance from the configuration.
724 if instance_name not in self._config_data.instances:
725 raise errors.ConfigurationError("Unknown instance '%s'" % instance_name)
726 del self._config_data.instances[instance_name]
727 self._config_data.cluster.serial_no += 1
730 @locking.ssynchronized(_config_lock)
731 def RenameInstance(self, old_name, new_name):
732 """Rename an instance.
734 This needs to be done in ConfigWriter and not by RemoveInstance
735 combined with AddInstance as only we can guarantee an atomic
739 if old_name not in self._config_data.instances:
740 raise errors.ConfigurationError("Unknown instance '%s'" % old_name)
741 inst = self._config_data.instances[old_name]
742 del self._config_data.instances[old_name]
745 for disk in inst.disks:
746 if disk.dev_type == constants.LD_FILE:
747 # rename the file paths in logical and physical id
748 file_storage_dir = os.path.dirname(os.path.dirname(disk.logical_id[1]))
749 disk.physical_id = disk.logical_id = (disk.logical_id[0],
750 os.path.join(file_storage_dir,
754 self._config_data.instances[inst.name] = inst
757 @locking.ssynchronized(_config_lock)
758 def MarkInstanceDown(self, instance_name):
759 """Mark the status of an instance to down in the configuration.
762 self._SetInstanceStatus(instance_name, False)
764 def _UnlockedGetInstanceList(self):
765 """Get the list of instances.
767 This function is for internal use, when the config lock is already held.
770 return self._config_data.instances.keys()
772 @locking.ssynchronized(_config_lock, shared=1)
773 def GetInstanceList(self):
774 """Get the list of instances.
776 @return: array of instances, ex. ['instance2.example.com',
777 'instance1.example.com']
780 return self._UnlockedGetInstanceList()
782 @locking.ssynchronized(_config_lock, shared=1)
783 def ExpandInstanceName(self, short_name):
784 """Attempt to expand an incomplete instance name.
787 return utils.MatchNameComponent(short_name,
788 self._config_data.instances.keys())
790 def _UnlockedGetInstanceInfo(self, instance_name):
791 """Returns informations about an instance.
793 This function is for internal use, when the config lock is already held.
796 if instance_name not in self._config_data.instances:
799 return self._config_data.instances[instance_name]
801 @locking.ssynchronized(_config_lock, shared=1)
802 def GetInstanceInfo(self, instance_name):
803 """Returns informations about an instance.
805 It takes the information from the configuration file. Other informations of
806 an instance are taken from the live systems.
808 @param instance_name: name of the instance, e.g.
809 I{instance1.example.com}
811 @rtype: L{objects.Instance}
812 @return: the instance object
815 return self._UnlockedGetInstanceInfo(instance_name)
817 @locking.ssynchronized(_config_lock, shared=1)
818 def GetAllInstancesInfo(self):
819 """Get the configuration of all instances.
822 @returns: dict of (instance, instance_info), where instance_info is what
823 would GetInstanceInfo return for the node
826 my_dict = dict([(instance, self._UnlockedGetInstanceInfo(instance))
827 for instance in self._UnlockedGetInstanceList()])
830 @locking.ssynchronized(_config_lock)
831 def AddNode(self, node):
832 """Add a node to the configuration.
834 @type node: L{objects.Node}
835 @param node: a Node instance
838 logging.info("Adding node %s to configuration" % node.name)
841 self._config_data.nodes[node.name] = node
842 self._config_data.cluster.serial_no += 1
845 @locking.ssynchronized(_config_lock)
846 def RemoveNode(self, node_name):
847 """Remove a node from the configuration.
850 logging.info("Removing node %s from configuration" % node_name)
852 if node_name not in self._config_data.nodes:
853 raise errors.ConfigurationError("Unknown node '%s'" % node_name)
855 del self._config_data.nodes[node_name]
856 self._config_data.cluster.serial_no += 1
859 @locking.ssynchronized(_config_lock, shared=1)
860 def ExpandNodeName(self, short_name):
861 """Attempt to expand an incomplete instance name.
864 return utils.MatchNameComponent(short_name,
865 self._config_data.nodes.keys())
867 def _UnlockedGetNodeInfo(self, node_name):
868 """Get the configuration of a node, as stored in the config.
870 This function is for internal use, when the config lock is already
873 @param node_name: the node name, e.g. I{node1.example.com}
875 @rtype: L{objects.Node}
876 @return: the node object
879 if node_name not in self._config_data.nodes:
882 return self._config_data.nodes[node_name]
885 @locking.ssynchronized(_config_lock, shared=1)
886 def GetNodeInfo(self, node_name):
887 """Get the configuration of a node, as stored in the config.
889 This is just a locked wrapper over L{_UnlockedGetNodeInfo}.
891 @param node_name: the node name, e.g. I{node1.example.com}
893 @rtype: L{objects.Node}
894 @return: the node object
897 return self._UnlockedGetNodeInfo(node_name)
899 def _UnlockedGetNodeList(self):
900 """Return the list of nodes which are in the configuration.
902 This function is for internal use, when the config lock is already
908 return self._config_data.nodes.keys()
911 @locking.ssynchronized(_config_lock, shared=1)
912 def GetNodeList(self):
913 """Return the list of nodes which are in the configuration.
916 return self._UnlockedGetNodeList()
918 @locking.ssynchronized(_config_lock, shared=1)
919 def GetOnlineNodeList(self):
920 """Return the list of nodes which are online.
923 all_nodes = [self._UnlockedGetNodeInfo(node)
924 for node in self._UnlockedGetNodeList()]
925 return [node.name for node in all_nodes if not node.offline]
927 @locking.ssynchronized(_config_lock, shared=1)
928 def GetAllNodesInfo(self):
929 """Get the configuration of all nodes.
932 @return: dict of (node, node_info), where node_info is what
933 would GetNodeInfo return for the node
936 my_dict = dict([(node, self._UnlockedGetNodeInfo(node))
937 for node in self._UnlockedGetNodeList()])
940 def _UnlockedGetMasterCandidateStats(self):
941 """Get the number of current and maximum desired and possible candidates.
944 @return: tuple of (current, desired and possible)
948 for node in self._config_data.nodes.itervalues():
949 if not (node.offline or node.drained):
951 if node.master_candidate:
953 mc_max = min(mc_max, self._config_data.cluster.candidate_pool_size)
954 return (mc_now, mc_max)
956 @locking.ssynchronized(_config_lock, shared=1)
957 def GetMasterCandidateStats(self):
958 """Get the number of current and maximum possible candidates.
960 This is just a wrapper over L{_UnlockedGetMasterCandidateStats}.
963 @return: tuple of (current, max)
966 return self._UnlockedGetMasterCandidateStats()
968 @locking.ssynchronized(_config_lock)
969 def MaintainCandidatePool(self):
970 """Try to grow the candidate pool to the desired size.
973 @return: list with the adjusted nodes (L{objects.Node} instances)
976 mc_now, mc_max = self._UnlockedGetMasterCandidateStats()
979 node_list = self._config_data.nodes.keys()
980 random.shuffle(node_list)
981 for name in node_list:
984 node = self._config_data.nodes[name]
985 if node.master_candidate or node.offline or node.drained:
987 mod_list.append(node)
988 node.master_candidate = True
992 # this should not happen
993 logging.warning("Warning: MaintainCandidatePool didn't manage to"
994 " fill the candidate pool (%d/%d)", mc_now, mc_max)
996 self._config_data.cluster.serial_no += 1
1001 def _BumpSerialNo(self):
1002 """Bump up the serial number of the config.
1005 self._config_data.serial_no += 1
1007 def _OpenConfig(self):
1008 """Read the config data from disk.
1011 f = open(self._cfg_file, 'r')
1014 data = objects.ConfigData.FromDict(serializer.Load(f.read()))
1015 except Exception, err:
1016 raise errors.ConfigurationError(err)
1020 # Make sure the configuration has the right version
1021 _ValidateConfig(data)
1023 if (not hasattr(data, 'cluster') or
1024 not hasattr(data.cluster, 'rsahostkeypub')):
1025 raise errors.ConfigurationError("Incomplete configuration"
1026 " (missing cluster.rsahostkeypub)")
1027 self._config_data = data
1028 # reset the last serial as -1 so that the next write will cause
1030 self._last_cluster_serial = -1
1032 def _DistributeConfig(self):
1033 """Distribute the configuration to the other nodes.
1035 Currently, this only copies the configuration file. In the future,
1036 it could be used to encapsulate the 2/3-phase update mechanism.
1045 myhostname = self._my_hostname
1046 # we can skip checking whether _UnlockedGetNodeInfo returns None
1047 # since the node list comes from _UnlocketGetNodeList, and we are
1048 # called with the lock held, so no modifications should take place
1050 for node_name in self._UnlockedGetNodeList():
1051 if node_name == myhostname:
1053 node_info = self._UnlockedGetNodeInfo(node_name)
1054 if not node_info.master_candidate:
1056 node_list.append(node_info.name)
1057 addr_list.append(node_info.primary_ip)
1059 result = rpc.RpcRunner.call_upload_file(node_list, self._cfg_file,
1060 address_list=addr_list)
1061 for node in node_list:
1062 if not result[node]:
1063 logging.error("copy of file %s to node %s failed",
1064 self._cfg_file, node)
1068 def _WriteConfig(self, destination=None):
1069 """Write the configuration data to persistent storage.
1072 config_errors = self._UnlockedVerifyConfig()
1074 raise errors.ConfigurationError("Configuration data is not"
1076 (", ".join(config_errors)))
1077 if destination is None:
1078 destination = self._cfg_file
1079 self._BumpSerialNo()
1080 txt = serializer.Dump(self._config_data.ToDict())
1081 dir_name, file_name = os.path.split(destination)
1082 fd, name = tempfile.mkstemp('.newconfig', file_name, dir_name)
1083 f = os.fdopen(fd, 'w')
1086 os.fsync(f.fileno())
1089 # we don't need to do os.close(fd) as f.close() did it
1090 os.rename(name, destination)
1091 self.write_count += 1
1093 # and redistribute the config file to master candidates
1094 self._DistributeConfig()
1096 # Write ssconf files on all nodes (including locally)
1097 if self._last_cluster_serial < self._config_data.cluster.serial_no:
1098 if not self._offline:
1099 rpc.RpcRunner.call_write_ssconf_files(self._UnlockedGetNodeList(),
1100 self._UnlockedGetSsconfValues())
1101 self._last_cluster_serial = self._config_data.cluster.serial_no
1103 def _UnlockedGetSsconfValues(self):
1104 """Return the values needed by ssconf.
1107 @return: a dictionary with keys the ssconf names and values their
1112 instance_names = utils.NiceSort(self._UnlockedGetInstanceList())
1113 node_names = utils.NiceSort(self._UnlockedGetNodeList())
1114 node_info = [self._UnlockedGetNodeInfo(name) for name in node_names]
1116 instance_data = fn(instance_names)
1117 off_data = fn(node.name for node in node_info if node.offline)
1118 on_data = fn(node.name for node in node_info if not node.offline)
1119 mc_data = fn(node.name for node in node_info if node.master_candidate)
1120 node_data = fn(node_names)
1122 cluster = self._config_data.cluster
1124 constants.SS_CLUSTER_NAME: cluster.cluster_name,
1125 constants.SS_FILE_STORAGE_DIR: cluster.file_storage_dir,
1126 constants.SS_MASTER_CANDIDATES: mc_data,
1127 constants.SS_MASTER_IP: cluster.master_ip,
1128 constants.SS_MASTER_NETDEV: cluster.master_netdev,
1129 constants.SS_MASTER_NODE: cluster.master_node,
1130 constants.SS_NODE_LIST: node_data,
1131 constants.SS_OFFLINE_NODES: off_data,
1132 constants.SS_ONLINE_NODES: on_data,
1133 constants.SS_INSTANCE_LIST: instance_data,
1134 constants.SS_RELEASE_VERSION: constants.RELEASE_VERSION,
1137 @locking.ssynchronized(_config_lock)
1138 def InitConfig(self, version, cluster_config, master_node_config):
1139 """Create the initial cluster configuration.
1141 It will contain the current node, which will also be the master
1142 node, and no instances.
1145 @param version: Configuration version
1146 @type cluster_config: objects.Cluster
1147 @param cluster_config: Cluster configuration
1148 @type master_node_config: objects.Node
1149 @param master_node_config: Master node configuration
1153 master_node_config.name: master_node_config,
1156 self._config_data = objects.ConfigData(version=version,
1157 cluster=cluster_config,
1163 @locking.ssynchronized(_config_lock, shared=1)
1164 def GetVGName(self):
1165 """Return the volume group name.
1168 return self._config_data.cluster.volume_group_name
1170 @locking.ssynchronized(_config_lock)
1171 def SetVGName(self, vg_name):
1172 """Set the volume group name.
1175 self._config_data.cluster.volume_group_name = vg_name
1176 self._config_data.cluster.serial_no += 1
1179 @locking.ssynchronized(_config_lock, shared=1)
1180 def GetDefBridge(self):
1181 """Return the default bridge.
1184 return self._config_data.cluster.default_bridge
1186 @locking.ssynchronized(_config_lock, shared=1)
1187 def GetMACPrefix(self):
1188 """Return the mac prefix.
1191 return self._config_data.cluster.mac_prefix
1193 @locking.ssynchronized(_config_lock, shared=1)
1194 def GetClusterInfo(self):
1195 """Returns informations about the cluster
1197 @rtype: L{objects.Cluster}
1198 @return: the cluster object
1201 return self._config_data.cluster
1203 @locking.ssynchronized(_config_lock)
1204 def Update(self, target):
1205 """Notify function to be called after updates.
1207 This function must be called when an object (as returned by
1208 GetInstanceInfo, GetNodeInfo, GetCluster) has been updated and the
1209 caller wants the modifications saved to the backing store. Note
1210 that all modified objects will be saved, but the target argument
1211 is the one the caller wants to ensure that it's saved.
1213 @param target: an instance of either L{objects.Cluster},
1214 L{objects.Node} or L{objects.Instance} which is existing in
1218 if self._config_data is None:
1219 raise errors.ProgrammerError("Configuration file not read,"
1221 update_serial = False
1222 if isinstance(target, objects.Cluster):
1223 test = target == self._config_data.cluster
1224 elif isinstance(target, objects.Node):
1225 test = target in self._config_data.nodes.values()
1226 update_serial = True
1227 elif isinstance(target, objects.Instance):
1228 test = target in self._config_data.instances.values()
1230 raise errors.ProgrammerError("Invalid object type (%s) passed to"
1231 " ConfigWriter.Update" % type(target))
1233 raise errors.ConfigurationError("Configuration updated since object"
1234 " has been read or unknown object")
1235 target.serial_no += 1
1238 # for node updates, we need to increase the cluster serial too
1239 self._config_data.cluster.serial_no += 1
1241 if isinstance(target, objects.Instance):
1242 self._UnlockedReleaseDRBDMinors(target.name)
1243 for nic in target.nics:
1244 self._temporary_macs.discard(nic.mac)