4 # Copyright (C) 2006, 2007, 2010, 2011 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
21 """Cluster related commands"""
23 # pylint: disable-msg=W0401,W0613,W0614,C0103
24 # W0401: Wildcard import ganeti.cli
25 # W0613: Unused argument, since all functions follow the same API
26 # W0614: Unused import %s from wildcard import (since we need cli)
27 # C0103: Invalid name gnt-cluster
34 from ganeti.cli import *
35 from ganeti import opcodes
36 from ganeti import constants
37 from ganeti import errors
38 from ganeti import utils
39 from ganeti import bootstrap
40 from ganeti import ssh
41 from ganeti import objects
42 from ganeti import uidpool
43 from ganeti import compat
44 from ganeti import netutils
47 ON_OPT = cli_option("--on", default=False,
48 action="store_true", dest="on",
49 help="Recover from an EPO")
51 GROUPS_OPT = cli_option("--groups", default=False,
52 action="store_true", dest="groups",
53 help="Arguments are node groups instead of nodes")
55 _EPO_PING_INTERVAL = 30 # 30 seconds between pings
56 _EPO_PING_TIMEOUT = 1 # 1 second
57 _EPO_REACHABLE_TIMEOUT = 15 * 60 # 15 minutes
61 def InitCluster(opts, args):
62 """Initialize the cluster.
64 @param opts: the command line options selected by the user
66 @param args: should contain only one element, the desired
69 @return: the desired exit code
72 if not opts.lvm_storage and opts.vg_name:
73 ToStderr("Options --no-lvm-storage and --vg-name conflict.")
76 vg_name = opts.vg_name
77 if opts.lvm_storage and not opts.vg_name:
78 vg_name = constants.DEFAULT_VG
80 if not opts.drbd_storage and opts.drbd_helper:
81 ToStderr("Options --no-drbd-storage and --drbd-usermode-helper conflict.")
84 drbd_helper = opts.drbd_helper
85 if opts.drbd_storage and not opts.drbd_helper:
86 drbd_helper = constants.DEFAULT_DRBD_HELPER
88 master_netdev = opts.master_netdev
89 if master_netdev is None:
90 master_netdev = constants.DEFAULT_BRIDGE
92 hvlist = opts.enabled_hypervisors
94 hvlist = constants.DEFAULT_ENABLED_HYPERVISOR
95 hvlist = hvlist.split(",")
97 hvparams = dict(opts.hvparams)
98 beparams = opts.beparams
99 nicparams = opts.nicparams
101 # prepare beparams dict
102 beparams = objects.FillDict(constants.BEC_DEFAULTS, beparams)
103 utils.ForceDictType(beparams, constants.BES_PARAMETER_TYPES)
105 # prepare nicparams dict
106 nicparams = objects.FillDict(constants.NICC_DEFAULTS, nicparams)
107 utils.ForceDictType(nicparams, constants.NICS_PARAMETER_TYPES)
109 # prepare ndparams dict
110 if opts.ndparams is None:
111 ndparams = dict(constants.NDC_DEFAULTS)
113 ndparams = objects.FillDict(constants.NDC_DEFAULTS, opts.ndparams)
114 utils.ForceDictType(ndparams, constants.NDS_PARAMETER_TYPES)
116 # prepare hvparams dict
117 for hv in constants.HYPER_TYPES:
118 if hv not in hvparams:
120 hvparams[hv] = objects.FillDict(constants.HVC_DEFAULTS[hv], hvparams[hv])
121 utils.ForceDictType(hvparams[hv], constants.HVS_PARAMETER_TYPES)
123 if opts.candidate_pool_size is None:
124 opts.candidate_pool_size = constants.MASTER_POOL_SIZE_DEFAULT
126 if opts.mac_prefix is None:
127 opts.mac_prefix = constants.DEFAULT_MAC_PREFIX
129 uid_pool = opts.uid_pool
130 if uid_pool is not None:
131 uid_pool = uidpool.ParseUidPool(uid_pool)
133 if opts.prealloc_wipe_disks is None:
134 opts.prealloc_wipe_disks = False
137 primary_ip_version = int(opts.primary_ip_version)
138 except (ValueError, TypeError), err:
139 ToStderr("Invalid primary ip version value: %s" % str(err))
142 bootstrap.InitCluster(cluster_name=args[0],
143 secondary_ip=opts.secondary_ip,
145 mac_prefix=opts.mac_prefix,
146 master_netdev=master_netdev,
147 file_storage_dir=opts.file_storage_dir,
148 shared_file_storage_dir=opts.shared_file_storage_dir,
149 enabled_hypervisors=hvlist,
154 candidate_pool_size=opts.candidate_pool_size,
155 modify_etc_hosts=opts.modify_etc_hosts,
156 modify_ssh_setup=opts.modify_ssh_setup,
157 maintain_node_health=opts.maintain_node_health,
158 drbd_helper=drbd_helper,
160 default_iallocator=opts.default_iallocator,
161 primary_ip_version=primary_ip_version,
162 prealloc_wipe_disks=opts.prealloc_wipe_disks,
164 op = opcodes.OpClusterPostInit()
165 SubmitOpCode(op, opts=opts)
170 def DestroyCluster(opts, args):
171 """Destroy the cluster.
173 @param opts: the command line options selected by the user
175 @param args: should be an empty list
177 @return: the desired exit code
180 if not opts.yes_do_it:
181 ToStderr("Destroying a cluster is irreversible. If you really want"
182 " destroy this cluster, supply the --yes-do-it option.")
185 op = opcodes.OpClusterDestroy()
186 master = SubmitOpCode(op, opts=opts)
187 # if we reached this, the opcode didn't fail; we can proceed to
188 # shutdown all the daemons
189 bootstrap.FinalizeClusterDestroy(master)
193 def RenameCluster(opts, args):
194 """Rename the cluster.
196 @param opts: the command line options selected by the user
198 @param args: should contain only one element, the new cluster name
200 @return: the desired exit code
205 (cluster_name, ) = cl.QueryConfigValues(["cluster_name"])
209 usertext = ("This will rename the cluster from '%s' to '%s'. If you are"
210 " connected over the network to the cluster name, the"
211 " operation is very dangerous as the IP address will be"
212 " removed from the node and the change may not go through."
213 " Continue?") % (cluster_name, new_name)
214 if not AskUser(usertext):
217 op = opcodes.OpClusterRename(name=new_name)
218 result = SubmitOpCode(op, opts=opts, cl=cl)
221 ToStdout("Cluster renamed from '%s' to '%s'", cluster_name, result)
226 def RedistributeConfig(opts, args):
227 """Forces push of the cluster configuration.
229 @param opts: the command line options selected by the user
231 @param args: empty list
233 @return: the desired exit code
236 op = opcodes.OpClusterRedistConf()
237 SubmitOrSend(op, opts)
241 def ShowClusterVersion(opts, args):
242 """Write version of ganeti software to the standard output.
244 @param opts: the command line options selected by the user
246 @param args: should be an empty list
248 @return: the desired exit code
252 result = cl.QueryClusterInfo()
253 ToStdout("Software version: %s", result["software_version"])
254 ToStdout("Internode protocol: %s", result["protocol_version"])
255 ToStdout("Configuration format: %s", result["config_version"])
256 ToStdout("OS api version: %s", result["os_api_version"])
257 ToStdout("Export interface: %s", result["export_version"])
261 def ShowClusterMaster(opts, args):
262 """Write name of master node to the standard output.
264 @param opts: the command line options selected by the user
266 @param args: should be an empty list
268 @return: the desired exit code
271 master = bootstrap.GetMaster()
276 def _PrintGroupedParams(paramsdict, level=1, roman=False):
277 """Print Grouped parameters (be, nic, disk) by group.
279 @type paramsdict: dict of dicts
280 @param paramsdict: {group: {param: value, ...}, ...}
282 @param level: Level of indention
286 for item, val in sorted(paramsdict.items()):
287 if isinstance(val, dict):
288 ToStdout("%s- %s:", indent, item)
289 _PrintGroupedParams(val, level=level + 1, roman=roman)
290 elif roman and isinstance(val, int):
291 ToStdout("%s %s: %s", indent, item, compat.TryToRoman(val))
293 ToStdout("%s %s: %s", indent, item, val)
296 def ShowClusterConfig(opts, args):
297 """Shows cluster information.
299 @param opts: the command line options selected by the user
301 @param args: should be an empty list
303 @return: the desired exit code
307 result = cl.QueryClusterInfo()
309 ToStdout("Cluster name: %s", result["name"])
310 ToStdout("Cluster UUID: %s", result["uuid"])
312 ToStdout("Creation time: %s", utils.FormatTime(result["ctime"]))
313 ToStdout("Modification time: %s", utils.FormatTime(result["mtime"]))
315 ToStdout("Master node: %s", result["master"])
317 ToStdout("Architecture (this node): %s (%s)",
318 result["architecture"][0], result["architecture"][1])
321 tags = utils.CommaJoin(utils.NiceSort(result["tags"]))
325 ToStdout("Tags: %s", tags)
327 ToStdout("Default hypervisor: %s", result["default_hypervisor"])
328 ToStdout("Enabled hypervisors: %s",
329 utils.CommaJoin(result["enabled_hypervisors"]))
331 ToStdout("Hypervisor parameters:")
332 _PrintGroupedParams(result["hvparams"])
334 ToStdout("OS-specific hypervisor parameters:")
335 _PrintGroupedParams(result["os_hvp"])
337 ToStdout("OS parameters:")
338 _PrintGroupedParams(result["osparams"])
340 ToStdout("Hidden OSes: %s", utils.CommaJoin(result["hidden_os"]))
341 ToStdout("Blacklisted OSes: %s", utils.CommaJoin(result["blacklisted_os"]))
343 ToStdout("Cluster parameters:")
344 ToStdout(" - candidate pool size: %s",
345 compat.TryToRoman(result["candidate_pool_size"],
346 convert=opts.roman_integers))
347 ToStdout(" - master netdev: %s", result["master_netdev"])
348 ToStdout(" - lvm volume group: %s", result["volume_group_name"])
349 if result["reserved_lvs"]:
350 reserved_lvs = utils.CommaJoin(result["reserved_lvs"])
352 reserved_lvs = "(none)"
353 ToStdout(" - lvm reserved volumes: %s", reserved_lvs)
354 ToStdout(" - drbd usermode helper: %s", result["drbd_usermode_helper"])
355 ToStdout(" - file storage path: %s", result["file_storage_dir"])
356 ToStdout(" - shared file storage path: %s",
357 result["shared_file_storage_dir"])
358 ToStdout(" - maintenance of node health: %s",
359 result["maintain_node_health"])
360 ToStdout(" - uid pool: %s",
361 uidpool.FormatUidPool(result["uid_pool"],
362 roman=opts.roman_integers))
363 ToStdout(" - default instance allocator: %s", result["default_iallocator"])
364 ToStdout(" - primary ip version: %d", result["primary_ip_version"])
365 ToStdout(" - preallocation wipe disks: %s", result["prealloc_wipe_disks"])
367 ToStdout("Default node parameters:")
368 _PrintGroupedParams(result["ndparams"], roman=opts.roman_integers)
370 ToStdout("Default instance parameters:")
371 _PrintGroupedParams(result["beparams"], roman=opts.roman_integers)
373 ToStdout("Default nic parameters:")
374 _PrintGroupedParams(result["nicparams"], roman=opts.roman_integers)
379 def ClusterCopyFile(opts, args):
380 """Copy a file from master to some nodes.
382 @param opts: the command line options selected by the user
384 @param args: should contain only one element, the path of
385 the file to be copied
387 @return: the desired exit code
391 if not os.path.exists(filename):
392 raise errors.OpPrereqError("No such filename '%s'" % filename,
397 cluster_name = cl.QueryConfigValues(["cluster_name"])[0]
399 results = GetOnlineNodes(nodes=opts.nodes, cl=cl, filter_master=True,
400 secondary_ips=opts.use_replication_network)
402 srun = ssh.SshRunner(cluster_name=cluster_name)
404 if not srun.CopyFileToNode(node, filename):
405 ToStderr("Copy of file %s to node %s failed", filename, node)
410 def RunClusterCommand(opts, args):
411 """Run a command on some nodes.
413 @param opts: the command line options selected by the user
415 @param args: should contain the command to be run and its arguments
417 @return: the desired exit code
422 command = " ".join(args)
424 nodes = GetOnlineNodes(nodes=opts.nodes, cl=cl)
426 cluster_name, master_node = cl.QueryConfigValues(["cluster_name",
429 srun = ssh.SshRunner(cluster_name=cluster_name)
431 # Make sure master node is at list end
432 if master_node in nodes:
433 nodes.remove(master_node)
434 nodes.append(master_node)
437 result = srun.Run(name, "root", command)
438 ToStdout("------------------------------------------------")
439 ToStdout("node: %s", name)
440 ToStdout("%s", result.output)
441 ToStdout("return code = %s", result.exit_code)
446 def VerifyCluster(opts, args):
447 """Verify integrity of cluster, performing various test on nodes.
449 @param opts: the command line options selected by the user
451 @param args: should be an empty list
453 @return: the desired exit code
457 if opts.skip_nplusone_mem:
458 skip_checks.append(constants.VERIFY_NPLUSONE_MEM)
459 op = opcodes.OpClusterVerify(skip_checks=skip_checks,
460 verbose=opts.verbose,
461 error_codes=opts.error_codes,
462 debug_simulate_errors=opts.simulate_errors)
463 if SubmitOpCode(op, opts=opts):
469 def VerifyDisks(opts, args):
470 """Verify integrity of cluster disks.
472 @param opts: the command line options selected by the user
474 @param args: should be an empty list
476 @return: the desired exit code
481 op = opcodes.OpClusterVerifyDisks()
482 result = SubmitOpCode(op, opts=opts, cl=cl)
483 if not isinstance(result, (list, tuple)) or len(result) != 3:
484 raise errors.ProgrammerError("Unknown result type for OpClusterVerifyDisks")
486 bad_nodes, instances, missing = result
488 retcode = constants.EXIT_SUCCESS
491 for node, text in bad_nodes.items():
492 ToStdout("Error gathering data on node %s: %s",
493 node, utils.SafeEncode(text[-400:]))
495 ToStdout("You need to fix these nodes first before fixing instances")
498 for iname in instances:
501 op = opcodes.OpInstanceActivateDisks(instance_name=iname)
503 ToStdout("Activating disks for instance '%s'", iname)
504 SubmitOpCode(op, opts=opts, cl=cl)
505 except errors.GenericError, err:
506 nret, msg = FormatError(err)
508 ToStderr("Error activating disks for instance %s: %s", iname, msg)
511 for iname, ival in missing.iteritems():
512 all_missing = compat.all(x[0] in bad_nodes for x in ival)
514 ToStdout("Instance %s cannot be verified as it lives on"
515 " broken nodes", iname)
517 ToStdout("Instance %s has missing logical volumes:", iname)
519 for node, vol in ival:
520 if node in bad_nodes:
521 ToStdout("\tbroken node %s /dev/%s", node, vol)
523 ToStdout("\t%s /dev/%s", node, vol)
525 ToStdout("You need to run replace or recreate disks for all the above"
526 " instances, if this message persist after fixing nodes.")
532 def RepairDiskSizes(opts, args):
533 """Verify sizes of cluster disks.
535 @param opts: the command line options selected by the user
537 @param args: optional list of instances to restrict check to
539 @return: the desired exit code
542 op = opcodes.OpClusterRepairDiskSizes(instances=args)
543 SubmitOpCode(op, opts=opts)
547 def MasterFailover(opts, args):
548 """Failover the master node.
550 This command, when run on a non-master node, will cause the current
551 master to cease being master, and the non-master to become new
554 @param opts: the command line options selected by the user
556 @param args: should be an empty list
558 @return: the desired exit code
562 usertext = ("This will perform the failover even if most other nodes"
563 " are down, or if this node is outdated. This is dangerous"
564 " as it can lead to a non-consistent cluster. Check the"
565 " gnt-cluster(8) man page before proceeding. Continue?")
566 if not AskUser(usertext):
569 return bootstrap.MasterFailover(no_voting=opts.no_voting)
572 def MasterPing(opts, args):
573 """Checks if the master is alive.
575 @param opts: the command line options selected by the user
577 @param args: should be an empty list
579 @return: the desired exit code
584 cl.QueryClusterInfo()
586 except Exception: # pylint: disable-msg=W0703
590 def SearchTags(opts, args):
591 """Searches the tags on all the cluster.
593 @param opts: the command line options selected by the user
595 @param args: should contain only one element, the tag pattern
597 @return: the desired exit code
600 op = opcodes.OpTagsSearch(pattern=args[0])
601 result = SubmitOpCode(op, opts=opts)
604 result = list(result)
606 for path, tag in result:
607 ToStdout("%s %s", path, tag)
610 def _RenewCrypto(new_cluster_cert, new_rapi_cert, rapi_cert_filename,
611 new_confd_hmac_key, new_cds, cds_filename,
613 """Renews cluster certificates, keys and secrets.
615 @type new_cluster_cert: bool
616 @param new_cluster_cert: Whether to generate a new cluster certificate
617 @type new_rapi_cert: bool
618 @param new_rapi_cert: Whether to generate a new RAPI certificate
619 @type rapi_cert_filename: string
620 @param rapi_cert_filename: Path to file containing new RAPI certificate
621 @type new_confd_hmac_key: bool
622 @param new_confd_hmac_key: Whether to generate a new HMAC key
624 @param new_cds: Whether to generate a new cluster domain secret
625 @type cds_filename: string
626 @param cds_filename: Path to file containing new cluster domain secret
628 @param force: Whether to ask user for confirmation
631 if new_rapi_cert and rapi_cert_filename:
632 ToStderr("Only one of the --new-rapi-certficate and --rapi-certificate"
633 " options can be specified at the same time.")
636 if new_cds and cds_filename:
637 ToStderr("Only one of the --new-cluster-domain-secret and"
638 " --cluster-domain-secret options can be specified at"
642 if rapi_cert_filename:
643 # Read and verify new certificate
645 rapi_cert_pem = utils.ReadFile(rapi_cert_filename)
647 OpenSSL.crypto.load_certificate(OpenSSL.crypto.FILETYPE_PEM,
649 except Exception, err: # pylint: disable-msg=W0703
650 ToStderr("Can't load new RAPI certificate from %s: %s" %
651 (rapi_cert_filename, str(err)))
655 OpenSSL.crypto.load_privatekey(OpenSSL.crypto.FILETYPE_PEM, rapi_cert_pem)
656 except Exception, err: # pylint: disable-msg=W0703
657 ToStderr("Can't load new RAPI private key from %s: %s" %
658 (rapi_cert_filename, str(err)))
666 cds = utils.ReadFile(cds_filename)
667 except Exception, err: # pylint: disable-msg=W0703
668 ToStderr("Can't load new cluster domain secret from %s: %s" %
669 (cds_filename, str(err)))
675 usertext = ("This requires all daemons on all nodes to be restarted and"
676 " may take some time. Continue?")
677 if not AskUser(usertext):
680 def _RenewCryptoInner(ctx):
681 ctx.feedback_fn("Updating certificates and keys")
682 bootstrap.GenerateClusterCrypto(new_cluster_cert, new_rapi_cert,
685 rapi_cert_pem=rapi_cert_pem,
691 files_to_copy.append(constants.NODED_CERT_FILE)
693 if new_rapi_cert or rapi_cert_pem:
694 files_to_copy.append(constants.RAPI_CERT_FILE)
696 if new_confd_hmac_key:
697 files_to_copy.append(constants.CONFD_HMAC_KEY)
700 files_to_copy.append(constants.CLUSTER_DOMAIN_SECRET_FILE)
703 for node_name in ctx.nonmaster_nodes:
704 ctx.feedback_fn("Copying %s to %s" %
705 (", ".join(files_to_copy), node_name))
706 for file_name in files_to_copy:
707 ctx.ssh.CopyFileToNode(node_name, file_name)
709 RunWhileClusterStopped(ToStdout, _RenewCryptoInner)
711 ToStdout("All requested certificates and keys have been replaced."
712 " Running \"gnt-cluster verify\" now is recommended.")
717 def RenewCrypto(opts, args):
718 """Renews cluster certificates, keys and secrets.
721 return _RenewCrypto(opts.new_cluster_cert,
724 opts.new_confd_hmac_key,
725 opts.new_cluster_domain_secret,
726 opts.cluster_domain_secret,
730 def SetClusterParams(opts, args):
731 """Modify the cluster.
733 @param opts: the command line options selected by the user
735 @param args: should be an empty list
737 @return: the desired exit code
740 if not (not opts.lvm_storage or opts.vg_name or
741 not opts.drbd_storage or opts.drbd_helper or
742 opts.enabled_hypervisors or opts.hvparams or
743 opts.beparams or opts.nicparams or opts.ndparams or
744 opts.candidate_pool_size is not None or
745 opts.uid_pool is not None or
746 opts.maintain_node_health is not None or
747 opts.add_uids is not None or
748 opts.remove_uids is not None or
749 opts.default_iallocator is not None or
750 opts.reserved_lvs is not None or
751 opts.master_netdev is not None or
752 opts.prealloc_wipe_disks is not None):
753 ToStderr("Please give at least one of the parameters.")
756 vg_name = opts.vg_name
757 if not opts.lvm_storage and opts.vg_name:
758 ToStderr("Options --no-lvm-storage and --vg-name conflict.")
761 if not opts.lvm_storage:
764 drbd_helper = opts.drbd_helper
765 if not opts.drbd_storage and opts.drbd_helper:
766 ToStderr("Options --no-drbd-storage and --drbd-usermode-helper conflict.")
769 if not opts.drbd_storage:
772 hvlist = opts.enabled_hypervisors
773 if hvlist is not None:
774 hvlist = hvlist.split(",")
776 # a list of (name, dict) we can pass directly to dict() (or [])
777 hvparams = dict(opts.hvparams)
778 for hv_params in hvparams.values():
779 utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES)
781 beparams = opts.beparams
782 utils.ForceDictType(beparams, constants.BES_PARAMETER_TYPES)
784 nicparams = opts.nicparams
785 utils.ForceDictType(nicparams, constants.NICS_PARAMETER_TYPES)
787 ndparams = opts.ndparams
788 if ndparams is not None:
789 utils.ForceDictType(ndparams, constants.NDS_PARAMETER_TYPES)
791 mnh = opts.maintain_node_health
793 uid_pool = opts.uid_pool
794 if uid_pool is not None:
795 uid_pool = uidpool.ParseUidPool(uid_pool)
797 add_uids = opts.add_uids
798 if add_uids is not None:
799 add_uids = uidpool.ParseUidPool(add_uids)
801 remove_uids = opts.remove_uids
802 if remove_uids is not None:
803 remove_uids = uidpool.ParseUidPool(remove_uids)
805 if opts.reserved_lvs is not None:
806 if opts.reserved_lvs == "":
807 opts.reserved_lvs = []
809 opts.reserved_lvs = utils.UnescapeAndSplit(opts.reserved_lvs, sep=",")
811 op = opcodes.OpClusterSetParams(vg_name=vg_name,
812 drbd_helper=drbd_helper,
813 enabled_hypervisors=hvlist,
819 candidate_pool_size=opts.candidate_pool_size,
820 maintain_node_health=mnh,
823 remove_uids=remove_uids,
824 default_iallocator=opts.default_iallocator,
825 prealloc_wipe_disks=opts.prealloc_wipe_disks,
826 master_netdev=opts.master_netdev,
827 reserved_lvs=opts.reserved_lvs)
828 SubmitOpCode(op, opts=opts)
832 def QueueOps(opts, args):
835 @param opts: the command line options selected by the user
837 @param args: should contain only one element, the subcommand
839 @return: the desired exit code
844 if command in ("drain", "undrain"):
845 drain_flag = command == "drain"
846 client.SetQueueDrainFlag(drain_flag)
847 elif command == "info":
848 result = client.QueryConfigValues(["drain_flag"])
853 ToStdout("The drain flag is %s" % val)
855 raise errors.OpPrereqError("Command '%s' is not valid." % command,
861 def _ShowWatcherPause(until):
862 if until is None or until < time.time():
863 ToStdout("The watcher is not paused.")
865 ToStdout("The watcher is paused until %s.", time.ctime(until))
868 def WatcherOps(opts, args):
869 """Watcher operations.
871 @param opts: the command line options selected by the user
873 @param args: should contain only one element, the subcommand
875 @return: the desired exit code
881 if command == "continue":
882 client.SetWatcherPause(None)
883 ToStdout("The watcher is no longer paused.")
885 elif command == "pause":
887 raise errors.OpPrereqError("Missing pause duration", errors.ECODE_INVAL)
889 result = client.SetWatcherPause(time.time() + ParseTimespec(args[1]))
890 _ShowWatcherPause(result)
892 elif command == "info":
893 result = client.QueryConfigValues(["watcher_pause"])
894 _ShowWatcherPause(result[0])
897 raise errors.OpPrereqError("Command '%s' is not valid." % command,
903 def _OobPower(opts, node_list, power):
904 """Puts the node in the list to desired power state.
906 @param opts: The command line options selected by the user
907 @param node_list: The list of nodes to operate on
908 @param power: True if they should be powered on, False otherwise
909 @return: The success of the operation (none failed)
913 command = constants.OOB_POWER_ON
915 command = constants.OOB_POWER_OFF
917 op = opcodes.OpOobCommand(node_names=node_list,
920 timeout=opts.oob_timeout,
921 power_delay=opts.power_delay)
922 result = SubmitOpCode(op, opts=opts)
924 for node_result in result:
925 (node_tuple, data_tuple) = node_result
926 (_, node_name) = node_tuple
927 (data_status, _) = data_tuple
928 if data_status != constants.RS_NORMAL:
929 assert data_status != constants.RS_UNAVAIL
931 ToStderr("There was a problem changing power for %s, please investigate",
940 def _InstanceStart(opts, inst_list, start):
941 """Puts the instances in the list to desired state.
943 @param opts: The command line options selected by the user
944 @param inst_list: The list of instances to operate on
945 @param start: True if they should be started, False for shutdown
946 @return: The success of the operation (none failed)
950 opcls = opcodes.OpInstanceStartup
951 text_submit, text_success, text_failed = ("startup", "started", "starting")
953 opcls = compat.partial(opcodes.OpInstanceShutdown,
954 timeout=opts.shutdown_timeout)
955 text_submit, text_success, text_failed = ("shutdown", "stopped", "stopping")
957 jex = JobExecutor(opts=opts)
959 for inst in inst_list:
960 ToStdout("Submit %s of instance %s", text_submit, inst)
961 op = opcls(instance_name=inst)
962 jex.QueueJob(inst, op)
964 results = jex.GetResults()
965 bad_cnt = len([1 for (success, _) in results if not success])
968 ToStdout("All instances have been %s successfully", text_success)
970 ToStderr("There were errors while %s instances:\n"
971 "%d error(s) out of %d instance(s)", text_failed, bad_cnt,
978 class _RunWhenNodesReachableHelper:
979 """Helper class to make shared internal state sharing easier.
981 @ivar success: Indicates if all action_cb calls were successful
984 def __init__(self, node_list, action_cb, node2ip, port, feedback_fn,
985 _ping_fn=netutils.TcpPing, _sleep_fn=time.sleep):
988 @param node_list: The list of nodes to be reachable
989 @param action_cb: Callback called when a new host is reachable
991 @param node2ip: Node to ip mapping
992 @param port: The port to use for the TCP ping
993 @param feedback_fn: The function used for feedback
994 @param _ping_fn: Function to check reachabilty (for unittest use only)
995 @param _sleep_fn: Function to sleep (for unittest use only)
998 self.down = set(node_list)
1000 self.node2ip = node2ip
1002 self.action_cb = action_cb
1004 self.feedback_fn = feedback_fn
1005 self._ping_fn = _ping_fn
1006 self._sleep_fn = _sleep_fn
1009 """When called we run action_cb.
1011 @raises utils.RetryAgain: When there are still down nodes
1014 if not self.action_cb(self.up):
1015 self.success = False
1018 raise utils.RetryAgain()
1022 def Wait(self, secs):
1023 """Checks if a host is up or waits remaining seconds.
1025 @param secs: The secs remaining
1029 for node in self.down:
1030 if self._ping_fn(self.node2ip[node], self.port, timeout=_EPO_PING_TIMEOUT,
1031 live_port_needed=True):
1032 self.feedback_fn("Node %s became available" % node)
1034 self.down -= self.up
1035 # If we have a node available there is the possibility to run the
1036 # action callback successfully, therefore we don't wait and return
1039 self._sleep_fn(max(0.0, start + secs - time.time()))
1042 def _RunWhenNodesReachable(node_list, action_cb, interval):
1043 """Run action_cb when nodes become reachable.
1045 @param node_list: The list of nodes to be reachable
1046 @param action_cb: Callback called when a new host is reachable
1047 @param interval: The earliest time to retry
1050 client = GetClient()
1051 cluster_info = client.QueryClusterInfo()
1052 if cluster_info["primary_ip_version"] == constants.IP4_VERSION:
1053 family = netutils.IPAddress.family
1055 family = netutils.IP6Address.family
1057 node2ip = dict((node, netutils.GetHostname(node, family=family).ip)
1058 for node in node_list)
1060 port = netutils.GetDaemonPort(constants.NODED)
1061 helper = _RunWhenNodesReachableHelper(node_list, action_cb, node2ip, port,
1065 return utils.Retry(helper, interval, _EPO_REACHABLE_TIMEOUT,
1066 wait_fn=helper.Wait)
1067 except utils.RetryTimeout:
1068 ToStderr("Time exceeded while waiting for nodes to become reachable"
1069 " again:\n - %s", " - ".join(helper.down))
1073 def _MaybeInstanceStartup(opts, inst_map, nodes_online,
1074 _instance_start_fn=_InstanceStart):
1075 """Start the instances conditional based on node_states.
1077 @param opts: The command line options selected by the user
1078 @param inst_map: A dict of inst -> nodes mapping
1079 @param nodes_online: A list of nodes online
1080 @param _instance_start_fn: Callback to start instances (unittest use only)
1081 @return: Success of the operation on all instances
1084 start_inst_list = []
1085 for (inst, nodes) in inst_map.items():
1086 if not (nodes - nodes_online):
1087 # All nodes the instance lives on are back online
1088 start_inst_list.append(inst)
1090 for inst in start_inst_list:
1094 return _instance_start_fn(opts, start_inst_list, True)
1099 def _EpoOn(opts, full_node_list, node_list, inst_map):
1100 """Does the actual power on.
1102 @param opts: The command line options selected by the user
1103 @param full_node_list: All nodes to operate on (includes nodes not supporting
1105 @param node_list: The list of nodes to operate on (all need to support OOB)
1106 @param inst_map: A dict of inst -> nodes mapping
1107 @return: The desired exit status
1110 if node_list and not _OobPower(opts, node_list, False):
1111 ToStderr("Not all nodes seem to get back up, investigate and start"
1112 " manually if needed")
1114 # Wait for the nodes to be back up
1115 action_cb = compat.partial(_MaybeInstanceStartup, opts, dict(inst_map))
1117 ToStdout("Waiting until all nodes are available again")
1118 if not _RunWhenNodesReachable(full_node_list, action_cb, _EPO_PING_INTERVAL):
1119 ToStderr("Please investigate and start stopped instances manually")
1120 return constants.EXIT_FAILURE
1122 return constants.EXIT_SUCCESS
1125 def _EpoOff(opts, node_list, inst_map):
1126 """Does the actual power off.
1128 @param opts: The command line options selected by the user
1129 @param node_list: The list of nodes to operate on (all need to support OOB)
1130 @param inst_map: A dict of inst -> nodes mapping
1131 @return: The desired exit status
1134 if not _InstanceStart(opts, inst_map.keys(), False):
1135 ToStderr("Please investigate and stop instances manually before continuing")
1136 return constants.EXIT_FAILURE
1139 return constants.EXIT_SUCCESS
1141 if _OobPower(opts, node_list, False):
1142 return constants.EXIT_SUCCESS
1144 return constants.EXIT_FAILURE
1147 def Epo(opts, args):
1150 @param opts: the command line options selected by the user
1152 @param args: should contain only one element, the subcommand
1154 @return: the desired exit code
1157 if opts.groups and opts.show_all:
1158 ToStderr("Only one of --groups or --all are allowed")
1159 return constants.EXIT_FAILURE
1160 elif args and opts.show_all:
1161 ToStderr("Arguments in combination with --all are not allowed")
1162 return constants.EXIT_FAILURE
1164 client = GetClient()
1167 node_query_list = itertools.chain(*client.QueryGroups(names=args,
1168 fields=["node_list"],
1171 node_query_list = args
1173 result = client.QueryNodes(names=node_query_list,
1174 fields=["name", "master", "pinst_list",
1175 "sinst_list", "powered", "offline"],
1179 for (idx, (node, master, pinsts, sinsts, powered,
1180 offline)) in enumerate(result):
1181 # Normalize the node_query_list as well
1182 if not opts.show_all:
1183 node_query_list[idx] = node
1185 for inst in (pinsts + sinsts):
1186 if inst in inst_map:
1188 inst_map[inst].add(node)
1190 inst_map[inst] = set()
1192 inst_map[inst] = set([node])
1194 if master and opts.on:
1195 # We ignore the master for turning on the machines, in fact we are
1196 # already operating on the master at this point :)
1198 elif master and not opts.show_all:
1199 ToStderr("%s is the master node, please do a master-failover to another"
1200 " node not affected by the EPO or use --all if you intend to"
1201 " shutdown the whole cluster", node)
1202 return constants.EXIT_FAILURE
1203 elif powered is None:
1204 ToStdout("Node %s does not support out-of-band handling, it can not be"
1205 " handled in a fully automated manner", node)
1206 elif powered == opts.on:
1207 ToStdout("Node %s is already in desired power state, skipping", node)
1208 elif not offline or (offline and powered):
1209 node_list.append(node)
1211 if not opts.force and not ConfirmOperation(node_query_list, "nodes", "epo"):
1212 return constants.EXIT_FAILURE
1215 return _EpoOn(opts, node_query_list, node_list, inst_map)
1217 return _EpoOff(opts, node_list, inst_map)
1222 InitCluster, [ArgHost(min=1, max=1)],
1223 [BACKEND_OPT, CP_SIZE_OPT, ENABLED_HV_OPT, GLOBAL_FILEDIR_OPT,
1224 HVLIST_OPT, MAC_PREFIX_OPT, MASTER_NETDEV_OPT, NIC_PARAMS_OPT,
1225 NOLVM_STORAGE_OPT, NOMODIFY_ETCHOSTS_OPT, NOMODIFY_SSH_SETUP_OPT,
1226 SECONDARY_IP_OPT, VG_NAME_OPT, MAINTAIN_NODE_HEALTH_OPT,
1227 UIDPOOL_OPT, DRBD_HELPER_OPT, NODRBD_STORAGE_OPT,
1228 DEFAULT_IALLOCATOR_OPT, PRIMARY_IP_VERSION_OPT, PREALLOC_WIPE_DISKS_OPT,
1229 NODE_PARAMS_OPT, GLOBAL_SHARED_FILEDIR_OPT],
1230 "[opts...] <cluster_name>", "Initialises a new cluster configuration"),
1232 DestroyCluster, ARGS_NONE, [YES_DOIT_OPT],
1233 "", "Destroy cluster"),
1235 RenameCluster, [ArgHost(min=1, max=1)],
1236 [FORCE_OPT, DRY_RUN_OPT],
1238 "Renames the cluster"),
1240 RedistributeConfig, ARGS_NONE, [SUBMIT_OPT, DRY_RUN_OPT, PRIORITY_OPT],
1241 "", "Forces a push of the configuration file and ssconf files"
1242 " to the nodes in the cluster"),
1244 VerifyCluster, ARGS_NONE,
1245 [VERBOSE_OPT, DEBUG_SIMERR_OPT, ERROR_CODES_OPT, NONPLUS1_OPT,
1246 DRY_RUN_OPT, PRIORITY_OPT],
1247 "", "Does a check on the cluster configuration"),
1249 VerifyDisks, ARGS_NONE, [PRIORITY_OPT],
1250 "", "Does a check on the cluster disk status"),
1251 'repair-disk-sizes': (
1252 RepairDiskSizes, ARGS_MANY_INSTANCES, [DRY_RUN_OPT, PRIORITY_OPT],
1253 "", "Updates mismatches in recorded disk sizes"),
1254 'master-failover': (
1255 MasterFailover, ARGS_NONE, [NOVOTING_OPT],
1256 "", "Makes the current node the master"),
1258 MasterPing, ARGS_NONE, [],
1259 "", "Checks if the master is alive"),
1261 ShowClusterVersion, ARGS_NONE, [],
1262 "", "Shows the cluster version"),
1264 ShowClusterMaster, ARGS_NONE, [],
1265 "", "Shows the cluster master"),
1267 ClusterCopyFile, [ArgFile(min=1, max=1)],
1268 [NODE_LIST_OPT, USE_REPL_NET_OPT],
1269 "[-n node...] <filename>", "Copies a file to all (or only some) nodes"),
1271 RunClusterCommand, [ArgCommand(min=1)],
1273 "[-n node...] <command>", "Runs a command on all (or only some) nodes"),
1275 ShowClusterConfig, ARGS_NONE, [ROMAN_OPT],
1276 "[--roman]", "Show cluster configuration"),
1278 ListTags, ARGS_NONE, [], "", "List the tags of the cluster"),
1280 AddTags, [ArgUnknown()], [TAG_SRC_OPT, PRIORITY_OPT],
1281 "tag...", "Add tags to the cluster"),
1283 RemoveTags, [ArgUnknown()], [TAG_SRC_OPT, PRIORITY_OPT],
1284 "tag...", "Remove tags from the cluster"),
1286 SearchTags, [ArgUnknown(min=1, max=1)], [PRIORITY_OPT], "",
1287 "Searches the tags on all objects on"
1288 " the cluster for a given pattern (regex)"),
1291 [ArgChoice(min=1, max=1, choices=["drain", "undrain", "info"])],
1292 [], "drain|undrain|info", "Change queue properties"),
1295 [ArgChoice(min=1, max=1, choices=["pause", "continue", "info"]),
1296 ArgSuggest(min=0, max=1, choices=["30m", "1h", "4h"])],
1298 "{pause <timespec>|continue|info}", "Change watcher properties"),
1300 SetClusterParams, ARGS_NONE,
1301 [BACKEND_OPT, CP_SIZE_OPT, ENABLED_HV_OPT, HVLIST_OPT, MASTER_NETDEV_OPT,
1302 NIC_PARAMS_OPT, NOLVM_STORAGE_OPT, VG_NAME_OPT, MAINTAIN_NODE_HEALTH_OPT,
1303 UIDPOOL_OPT, ADD_UIDS_OPT, REMOVE_UIDS_OPT, DRBD_HELPER_OPT,
1304 NODRBD_STORAGE_OPT, DEFAULT_IALLOCATOR_OPT, RESERVED_LVS_OPT,
1305 DRY_RUN_OPT, PRIORITY_OPT, PREALLOC_WIPE_DISKS_OPT, NODE_PARAMS_OPT],
1307 "Alters the parameters of the cluster"),
1309 RenewCrypto, ARGS_NONE,
1310 [NEW_CLUSTER_CERT_OPT, NEW_RAPI_CERT_OPT, RAPI_CERT_OPT,
1311 NEW_CONFD_HMAC_KEY_OPT, FORCE_OPT,
1312 NEW_CLUSTER_DOMAIN_SECRET_OPT, CLUSTER_DOMAIN_SECRET_OPT],
1314 "Renews cluster certificates, keys and secrets"),
1316 Epo, [ArgUnknown()],
1317 [FORCE_OPT, ON_OPT, GROUPS_OPT, ALL_OPT, OOB_TIMEOUT_OPT,
1318 SHUTDOWN_TIMEOUT_OPT, POWER_DELAY_OPT],
1320 "Performs an emergency power-off on given args"),
1324 #: dictionary with aliases for commands
1326 'masterfailover': 'master-failover',
1331 return GenericMain(commands, override={"tag_type": constants.TAG_CLUSTER},