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=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 ActivateMasterIp(opts, args):
227 """Activates the master IP.
230 op = opcodes.OpClusterActivateMasterIp()
235 def DeactivateMasterIp(opts, args):
236 """Deactivates the master IP.
240 usertext = ("This will disable the master IP. All the open connections to"
241 " the master IP will be closed. To reach the master you will"
242 " need to use its node IP."
244 if not AskUser(usertext):
247 op = opcodes.OpClusterDeactivateMasterIp()
252 def RedistributeConfig(opts, args):
253 """Forces push of the cluster configuration.
255 @param opts: the command line options selected by the user
257 @param args: empty list
259 @return: the desired exit code
262 op = opcodes.OpClusterRedistConf()
263 SubmitOrSend(op, opts)
267 def ShowClusterVersion(opts, args):
268 """Write version of ganeti software to the standard output.
270 @param opts: the command line options selected by the user
272 @param args: should be an empty list
274 @return: the desired exit code
278 result = cl.QueryClusterInfo()
279 ToStdout("Software version: %s", result["software_version"])
280 ToStdout("Internode protocol: %s", result["protocol_version"])
281 ToStdout("Configuration format: %s", result["config_version"])
282 ToStdout("OS api version: %s", result["os_api_version"])
283 ToStdout("Export interface: %s", result["export_version"])
287 def ShowClusterMaster(opts, args):
288 """Write name of master node to the standard output.
290 @param opts: the command line options selected by the user
292 @param args: should be an empty list
294 @return: the desired exit code
297 master = bootstrap.GetMaster()
302 def _PrintGroupedParams(paramsdict, level=1, roman=False):
303 """Print Grouped parameters (be, nic, disk) by group.
305 @type paramsdict: dict of dicts
306 @param paramsdict: {group: {param: value, ...}, ...}
308 @param level: Level of indention
312 for item, val in sorted(paramsdict.items()):
313 if isinstance(val, dict):
314 ToStdout("%s- %s:", indent, item)
315 _PrintGroupedParams(val, level=level + 1, roman=roman)
316 elif roman and isinstance(val, int):
317 ToStdout("%s %s: %s", indent, item, compat.TryToRoman(val))
319 ToStdout("%s %s: %s", indent, item, val)
322 def ShowClusterConfig(opts, args):
323 """Shows cluster information.
325 @param opts: the command line options selected by the user
327 @param args: should be an empty list
329 @return: the desired exit code
333 result = cl.QueryClusterInfo()
335 ToStdout("Cluster name: %s", result["name"])
336 ToStdout("Cluster UUID: %s", result["uuid"])
338 ToStdout("Creation time: %s", utils.FormatTime(result["ctime"]))
339 ToStdout("Modification time: %s", utils.FormatTime(result["mtime"]))
341 ToStdout("Master node: %s", result["master"])
343 ToStdout("Architecture (this node): %s (%s)",
344 result["architecture"][0], result["architecture"][1])
347 tags = utils.CommaJoin(utils.NiceSort(result["tags"]))
351 ToStdout("Tags: %s", tags)
353 ToStdout("Default hypervisor: %s", result["default_hypervisor"])
354 ToStdout("Enabled hypervisors: %s",
355 utils.CommaJoin(result["enabled_hypervisors"]))
357 ToStdout("Hypervisor parameters:")
358 _PrintGroupedParams(result["hvparams"])
360 ToStdout("OS-specific hypervisor parameters:")
361 _PrintGroupedParams(result["os_hvp"])
363 ToStdout("OS parameters:")
364 _PrintGroupedParams(result["osparams"])
366 ToStdout("Hidden OSes: %s", utils.CommaJoin(result["hidden_os"]))
367 ToStdout("Blacklisted OSes: %s", utils.CommaJoin(result["blacklisted_os"]))
369 ToStdout("Cluster parameters:")
370 ToStdout(" - candidate pool size: %s",
371 compat.TryToRoman(result["candidate_pool_size"],
372 convert=opts.roman_integers))
373 ToStdout(" - master netdev: %s", result["master_netdev"])
374 ToStdout(" - lvm volume group: %s", result["volume_group_name"])
375 if result["reserved_lvs"]:
376 reserved_lvs = utils.CommaJoin(result["reserved_lvs"])
378 reserved_lvs = "(none)"
379 ToStdout(" - lvm reserved volumes: %s", reserved_lvs)
380 ToStdout(" - drbd usermode helper: %s", result["drbd_usermode_helper"])
381 ToStdout(" - file storage path: %s", result["file_storage_dir"])
382 ToStdout(" - shared file storage path: %s",
383 result["shared_file_storage_dir"])
384 ToStdout(" - maintenance of node health: %s",
385 result["maintain_node_health"])
386 ToStdout(" - uid pool: %s",
387 uidpool.FormatUidPool(result["uid_pool"],
388 roman=opts.roman_integers))
389 ToStdout(" - default instance allocator: %s", result["default_iallocator"])
390 ToStdout(" - primary ip version: %d", result["primary_ip_version"])
391 ToStdout(" - preallocation wipe disks: %s", result["prealloc_wipe_disks"])
392 ToStdout(" - OS search path: %s", utils.CommaJoin(constants.OS_SEARCH_PATH))
394 ToStdout("Default node parameters:")
395 _PrintGroupedParams(result["ndparams"], roman=opts.roman_integers)
397 ToStdout("Default instance parameters:")
398 _PrintGroupedParams(result["beparams"], roman=opts.roman_integers)
400 ToStdout("Default nic parameters:")
401 _PrintGroupedParams(result["nicparams"], roman=opts.roman_integers)
406 def ClusterCopyFile(opts, args):
407 """Copy a file from master to some nodes.
409 @param opts: the command line options selected by the user
411 @param args: should contain only one element, the path of
412 the file to be copied
414 @return: the desired exit code
418 if not os.path.exists(filename):
419 raise errors.OpPrereqError("No such filename '%s'" % filename,
424 cluster_name = cl.QueryConfigValues(["cluster_name"])[0]
426 results = GetOnlineNodes(nodes=opts.nodes, cl=cl, filter_master=True,
427 secondary_ips=opts.use_replication_network,
428 nodegroup=opts.nodegroup)
430 srun = ssh.SshRunner(cluster_name=cluster_name)
432 if not srun.CopyFileToNode(node, filename):
433 ToStderr("Copy of file %s to node %s failed", filename, node)
438 def RunClusterCommand(opts, args):
439 """Run a command on some nodes.
441 @param opts: the command line options selected by the user
443 @param args: should contain the command to be run and its arguments
445 @return: the desired exit code
450 command = " ".join(args)
452 nodes = GetOnlineNodes(nodes=opts.nodes, cl=cl, nodegroup=opts.nodegroup)
454 cluster_name, master_node = cl.QueryConfigValues(["cluster_name",
457 srun = ssh.SshRunner(cluster_name=cluster_name)
459 # Make sure master node is at list end
460 if master_node in nodes:
461 nodes.remove(master_node)
462 nodes.append(master_node)
465 result = srun.Run(name, "root", command)
466 ToStdout("------------------------------------------------")
467 ToStdout("node: %s", name)
468 ToStdout("%s", result.output)
469 ToStdout("return code = %s", result.exit_code)
474 def VerifyCluster(opts, args):
475 """Verify integrity of cluster, performing various test on nodes.
477 @param opts: the command line options selected by the user
479 @param args: should be an empty list
481 @return: the desired exit code
486 if opts.skip_nplusone_mem:
487 skip_checks.append(constants.VERIFY_NPLUSONE_MEM)
491 op = opcodes.OpClusterVerify(verbose=opts.verbose,
492 error_codes=opts.error_codes,
493 debug_simulate_errors=opts.simulate_errors,
494 skip_checks=skip_checks,
495 group_name=opts.nodegroup)
496 result = SubmitOpCode(op, cl=cl, opts=opts)
498 # Keep track of submitted jobs
499 jex = JobExecutor(cl=cl, opts=opts)
501 for (status, job_id) in result[constants.JOB_IDS_KEY]:
502 jex.AddJobId(None, status, job_id)
504 results = jex.GetResults()
506 (bad_jobs, bad_results) = \
508 # Convert iterators to lists
511 map(compat.partial(itertools.ifilterfalse, bool),
512 # Convert result to booleans in a tuple
513 zip(*((job_success, len(op_results) == 1 and op_results[0])
514 for (job_success, op_results) in results)))))
516 if bad_jobs == 0 and bad_results == 0:
517 rcode = constants.EXIT_SUCCESS
519 rcode = constants.EXIT_FAILURE
521 ToStdout("%s job(s) failed while verifying the cluster.", bad_jobs)
526 def VerifyDisks(opts, args):
527 """Verify integrity of cluster disks.
529 @param opts: the command line options selected by the user
531 @param args: should be an empty list
533 @return: the desired exit code
538 op = opcodes.OpClusterVerifyDisks()
540 result = SubmitOpCode(op, cl=cl, opts=opts)
542 # Keep track of submitted jobs
543 jex = JobExecutor(cl=cl, opts=opts)
545 for (status, job_id) in result[constants.JOB_IDS_KEY]:
546 jex.AddJobId(None, status, job_id)
548 retcode = constants.EXIT_SUCCESS
550 for (status, result) in jex.GetResults():
552 ToStdout("Job failed: %s", result)
555 ((bad_nodes, instances, missing), ) = result
557 for node, text in bad_nodes.items():
558 ToStdout("Error gathering data on node %s: %s",
559 node, utils.SafeEncode(text[-400:]))
560 retcode = constants.EXIT_FAILURE
561 ToStdout("You need to fix these nodes first before fixing instances")
563 for iname in instances:
566 op = opcodes.OpInstanceActivateDisks(instance_name=iname)
568 ToStdout("Activating disks for instance '%s'", iname)
569 SubmitOpCode(op, opts=opts, cl=cl)
570 except errors.GenericError, err:
571 nret, msg = FormatError(err)
573 ToStderr("Error activating disks for instance %s: %s", iname, msg)
576 for iname, ival in missing.iteritems():
577 all_missing = compat.all(x[0] in bad_nodes for x in ival)
579 ToStdout("Instance %s cannot be verified as it lives on"
580 " broken nodes", iname)
582 ToStdout("Instance %s has missing logical volumes:", iname)
584 for node, vol in ival:
585 if node in bad_nodes:
586 ToStdout("\tbroken node %s /dev/%s", node, vol)
588 ToStdout("\t%s /dev/%s", node, vol)
590 ToStdout("You need to replace or recreate disks for all the above"
591 " instances if this message persists after fixing broken nodes.")
592 retcode = constants.EXIT_FAILURE
597 def RepairDiskSizes(opts, args):
598 """Verify sizes of cluster disks.
600 @param opts: the command line options selected by the user
602 @param args: optional list of instances to restrict check to
604 @return: the desired exit code
607 op = opcodes.OpClusterRepairDiskSizes(instances=args)
608 SubmitOpCode(op, opts=opts)
612 def MasterFailover(opts, args):
613 """Failover the master node.
615 This command, when run on a non-master node, will cause the current
616 master to cease being master, and the non-master to become new
619 @param opts: the command line options selected by the user
621 @param args: should be an empty list
623 @return: the desired exit code
627 usertext = ("This will perform the failover even if most other nodes"
628 " are down, or if this node is outdated. This is dangerous"
629 " as it can lead to a non-consistent cluster. Check the"
630 " gnt-cluster(8) man page before proceeding. Continue?")
631 if not AskUser(usertext):
634 return bootstrap.MasterFailover(no_voting=opts.no_voting)
637 def MasterPing(opts, args):
638 """Checks if the master is alive.
640 @param opts: the command line options selected by the user
642 @param args: should be an empty list
644 @return: the desired exit code
649 cl.QueryClusterInfo()
651 except Exception: # pylint: disable=W0703
655 def SearchTags(opts, args):
656 """Searches the tags on all the cluster.
658 @param opts: the command line options selected by the user
660 @param args: should contain only one element, the tag pattern
662 @return: the desired exit code
665 op = opcodes.OpTagsSearch(pattern=args[0])
666 result = SubmitOpCode(op, opts=opts)
669 result = list(result)
671 for path, tag in result:
672 ToStdout("%s %s", path, tag)
675 def _RenewCrypto(new_cluster_cert, new_rapi_cert, rapi_cert_filename,
676 new_confd_hmac_key, new_cds, cds_filename,
678 """Renews cluster certificates, keys and secrets.
680 @type new_cluster_cert: bool
681 @param new_cluster_cert: Whether to generate a new cluster certificate
682 @type new_rapi_cert: bool
683 @param new_rapi_cert: Whether to generate a new RAPI certificate
684 @type rapi_cert_filename: string
685 @param rapi_cert_filename: Path to file containing new RAPI certificate
686 @type new_confd_hmac_key: bool
687 @param new_confd_hmac_key: Whether to generate a new HMAC key
689 @param new_cds: Whether to generate a new cluster domain secret
690 @type cds_filename: string
691 @param cds_filename: Path to file containing new cluster domain secret
693 @param force: Whether to ask user for confirmation
696 if new_rapi_cert and rapi_cert_filename:
697 ToStderr("Only one of the --new-rapi-certificate and --rapi-certificate"
698 " options can be specified at the same time.")
701 if new_cds and cds_filename:
702 ToStderr("Only one of the --new-cluster-domain-secret and"
703 " --cluster-domain-secret options can be specified at"
707 if rapi_cert_filename:
708 # Read and verify new certificate
710 rapi_cert_pem = utils.ReadFile(rapi_cert_filename)
712 OpenSSL.crypto.load_certificate(OpenSSL.crypto.FILETYPE_PEM,
714 except Exception, err: # pylint: disable=W0703
715 ToStderr("Can't load new RAPI certificate from %s: %s" %
716 (rapi_cert_filename, str(err)))
720 OpenSSL.crypto.load_privatekey(OpenSSL.crypto.FILETYPE_PEM, rapi_cert_pem)
721 except Exception, err: # pylint: disable=W0703
722 ToStderr("Can't load new RAPI private key from %s: %s" %
723 (rapi_cert_filename, str(err)))
731 cds = utils.ReadFile(cds_filename)
732 except Exception, err: # pylint: disable=W0703
733 ToStderr("Can't load new cluster domain secret from %s: %s" %
734 (cds_filename, str(err)))
740 usertext = ("This requires all daemons on all nodes to be restarted and"
741 " may take some time. Continue?")
742 if not AskUser(usertext):
745 def _RenewCryptoInner(ctx):
746 ctx.feedback_fn("Updating certificates and keys")
747 bootstrap.GenerateClusterCrypto(new_cluster_cert, new_rapi_cert,
750 rapi_cert_pem=rapi_cert_pem,
756 files_to_copy.append(constants.NODED_CERT_FILE)
758 if new_rapi_cert or rapi_cert_pem:
759 files_to_copy.append(constants.RAPI_CERT_FILE)
761 if new_confd_hmac_key:
762 files_to_copy.append(constants.CONFD_HMAC_KEY)
765 files_to_copy.append(constants.CLUSTER_DOMAIN_SECRET_FILE)
768 for node_name in ctx.nonmaster_nodes:
769 ctx.feedback_fn("Copying %s to %s" %
770 (", ".join(files_to_copy), node_name))
771 for file_name in files_to_copy:
772 ctx.ssh.CopyFileToNode(node_name, file_name)
774 RunWhileClusterStopped(ToStdout, _RenewCryptoInner)
776 ToStdout("All requested certificates and keys have been replaced."
777 " Running \"gnt-cluster verify\" now is recommended.")
782 def RenewCrypto(opts, args):
783 """Renews cluster certificates, keys and secrets.
786 return _RenewCrypto(opts.new_cluster_cert,
789 opts.new_confd_hmac_key,
790 opts.new_cluster_domain_secret,
791 opts.cluster_domain_secret,
795 def SetClusterParams(opts, args):
796 """Modify the cluster.
798 @param opts: the command line options selected by the user
800 @param args: should be an empty list
802 @return: the desired exit code
805 if not (not opts.lvm_storage or opts.vg_name or
806 not opts.drbd_storage or opts.drbd_helper or
807 opts.enabled_hypervisors or opts.hvparams or
808 opts.beparams or opts.nicparams or opts.ndparams or
809 opts.candidate_pool_size is not None or
810 opts.uid_pool is not None or
811 opts.maintain_node_health is not None or
812 opts.add_uids is not None or
813 opts.remove_uids is not None or
814 opts.default_iallocator is not None or
815 opts.reserved_lvs is not None or
816 opts.master_netdev is not None or
817 opts.prealloc_wipe_disks is not None):
818 ToStderr("Please give at least one of the parameters.")
821 vg_name = opts.vg_name
822 if not opts.lvm_storage and opts.vg_name:
823 ToStderr("Options --no-lvm-storage and --vg-name conflict.")
826 if not opts.lvm_storage:
829 drbd_helper = opts.drbd_helper
830 if not opts.drbd_storage and opts.drbd_helper:
831 ToStderr("Options --no-drbd-storage and --drbd-usermode-helper conflict.")
834 if not opts.drbd_storage:
837 hvlist = opts.enabled_hypervisors
838 if hvlist is not None:
839 hvlist = hvlist.split(",")
841 # a list of (name, dict) we can pass directly to dict() (or [])
842 hvparams = dict(opts.hvparams)
843 for hv_params in hvparams.values():
844 utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES)
846 beparams = opts.beparams
847 utils.ForceDictType(beparams, constants.BES_PARAMETER_TYPES)
849 nicparams = opts.nicparams
850 utils.ForceDictType(nicparams, constants.NICS_PARAMETER_TYPES)
852 ndparams = opts.ndparams
853 if ndparams is not None:
854 utils.ForceDictType(ndparams, constants.NDS_PARAMETER_TYPES)
856 mnh = opts.maintain_node_health
858 uid_pool = opts.uid_pool
859 if uid_pool is not None:
860 uid_pool = uidpool.ParseUidPool(uid_pool)
862 add_uids = opts.add_uids
863 if add_uids is not None:
864 add_uids = uidpool.ParseUidPool(add_uids)
866 remove_uids = opts.remove_uids
867 if remove_uids is not None:
868 remove_uids = uidpool.ParseUidPool(remove_uids)
870 if opts.reserved_lvs is not None:
871 if opts.reserved_lvs == "":
872 opts.reserved_lvs = []
874 opts.reserved_lvs = utils.UnescapeAndSplit(opts.reserved_lvs, sep=",")
876 op = opcodes.OpClusterSetParams(vg_name=vg_name,
877 drbd_helper=drbd_helper,
878 enabled_hypervisors=hvlist,
884 candidate_pool_size=opts.candidate_pool_size,
885 maintain_node_health=mnh,
888 remove_uids=remove_uids,
889 default_iallocator=opts.default_iallocator,
890 prealloc_wipe_disks=opts.prealloc_wipe_disks,
891 master_netdev=opts.master_netdev,
892 reserved_lvs=opts.reserved_lvs)
893 SubmitOpCode(op, opts=opts)
897 def QueueOps(opts, args):
900 @param opts: the command line options selected by the user
902 @param args: should contain only one element, the subcommand
904 @return: the desired exit code
909 if command in ("drain", "undrain"):
910 drain_flag = command == "drain"
911 client.SetQueueDrainFlag(drain_flag)
912 elif command == "info":
913 result = client.QueryConfigValues(["drain_flag"])
918 ToStdout("The drain flag is %s" % val)
920 raise errors.OpPrereqError("Command '%s' is not valid." % command,
926 def _ShowWatcherPause(until):
927 if until is None or until < time.time():
928 ToStdout("The watcher is not paused.")
930 ToStdout("The watcher is paused until %s.", time.ctime(until))
933 def WatcherOps(opts, args):
934 """Watcher operations.
936 @param opts: the command line options selected by the user
938 @param args: should contain only one element, the subcommand
940 @return: the desired exit code
946 if command == "continue":
947 client.SetWatcherPause(None)
948 ToStdout("The watcher is no longer paused.")
950 elif command == "pause":
952 raise errors.OpPrereqError("Missing pause duration", errors.ECODE_INVAL)
954 result = client.SetWatcherPause(time.time() + ParseTimespec(args[1]))
955 _ShowWatcherPause(result)
957 elif command == "info":
958 result = client.QueryConfigValues(["watcher_pause"])
959 _ShowWatcherPause(result[0])
962 raise errors.OpPrereqError("Command '%s' is not valid." % command,
968 def _OobPower(opts, node_list, power):
969 """Puts the node in the list to desired power state.
971 @param opts: The command line options selected by the user
972 @param node_list: The list of nodes to operate on
973 @param power: True if they should be powered on, False otherwise
974 @return: The success of the operation (none failed)
978 command = constants.OOB_POWER_ON
980 command = constants.OOB_POWER_OFF
982 op = opcodes.OpOobCommand(node_names=node_list,
985 timeout=opts.oob_timeout,
986 power_delay=opts.power_delay)
987 result = SubmitOpCode(op, opts=opts)
989 for node_result in result:
990 (node_tuple, data_tuple) = node_result
991 (_, node_name) = node_tuple
992 (data_status, _) = data_tuple
993 if data_status != constants.RS_NORMAL:
994 assert data_status != constants.RS_UNAVAIL
996 ToStderr("There was a problem changing power for %s, please investigate",
1005 def _InstanceStart(opts, inst_list, start):
1006 """Puts the instances in the list to desired state.
1008 @param opts: The command line options selected by the user
1009 @param inst_list: The list of instances to operate on
1010 @param start: True if they should be started, False for shutdown
1011 @return: The success of the operation (none failed)
1015 opcls = opcodes.OpInstanceStartup
1016 text_submit, text_success, text_failed = ("startup", "started", "starting")
1018 opcls = compat.partial(opcodes.OpInstanceShutdown,
1019 timeout=opts.shutdown_timeout)
1020 text_submit, text_success, text_failed = ("shutdown", "stopped", "stopping")
1022 jex = JobExecutor(opts=opts)
1024 for inst in inst_list:
1025 ToStdout("Submit %s of instance %s", text_submit, inst)
1026 op = opcls(instance_name=inst)
1027 jex.QueueJob(inst, op)
1029 results = jex.GetResults()
1030 bad_cnt = len([1 for (success, _) in results if not success])
1033 ToStdout("All instances have been %s successfully", text_success)
1035 ToStderr("There were errors while %s instances:\n"
1036 "%d error(s) out of %d instance(s)", text_failed, bad_cnt,
1043 class _RunWhenNodesReachableHelper:
1044 """Helper class to make shared internal state sharing easier.
1046 @ivar success: Indicates if all action_cb calls were successful
1049 def __init__(self, node_list, action_cb, node2ip, port, feedback_fn,
1050 _ping_fn=netutils.TcpPing, _sleep_fn=time.sleep):
1053 @param node_list: The list of nodes to be reachable
1054 @param action_cb: Callback called when a new host is reachable
1056 @param node2ip: Node to ip mapping
1057 @param port: The port to use for the TCP ping
1058 @param feedback_fn: The function used for feedback
1059 @param _ping_fn: Function to check reachabilty (for unittest use only)
1060 @param _sleep_fn: Function to sleep (for unittest use only)
1063 self.down = set(node_list)
1065 self.node2ip = node2ip
1067 self.action_cb = action_cb
1069 self.feedback_fn = feedback_fn
1070 self._ping_fn = _ping_fn
1071 self._sleep_fn = _sleep_fn
1074 """When called we run action_cb.
1076 @raises utils.RetryAgain: When there are still down nodes
1079 if not self.action_cb(self.up):
1080 self.success = False
1083 raise utils.RetryAgain()
1087 def Wait(self, secs):
1088 """Checks if a host is up or waits remaining seconds.
1090 @param secs: The secs remaining
1094 for node in self.down:
1095 if self._ping_fn(self.node2ip[node], self.port, timeout=_EPO_PING_TIMEOUT,
1096 live_port_needed=True):
1097 self.feedback_fn("Node %s became available" % node)
1099 self.down -= self.up
1100 # If we have a node available there is the possibility to run the
1101 # action callback successfully, therefore we don't wait and return
1104 self._sleep_fn(max(0.0, start + secs - time.time()))
1107 def _RunWhenNodesReachable(node_list, action_cb, interval):
1108 """Run action_cb when nodes become reachable.
1110 @param node_list: The list of nodes to be reachable
1111 @param action_cb: Callback called when a new host is reachable
1112 @param interval: The earliest time to retry
1115 client = GetClient()
1116 cluster_info = client.QueryClusterInfo()
1117 if cluster_info["primary_ip_version"] == constants.IP4_VERSION:
1118 family = netutils.IPAddress.family
1120 family = netutils.IP6Address.family
1122 node2ip = dict((node, netutils.GetHostname(node, family=family).ip)
1123 for node in node_list)
1125 port = netutils.GetDaemonPort(constants.NODED)
1126 helper = _RunWhenNodesReachableHelper(node_list, action_cb, node2ip, port,
1130 return utils.Retry(helper, interval, _EPO_REACHABLE_TIMEOUT,
1131 wait_fn=helper.Wait)
1132 except utils.RetryTimeout:
1133 ToStderr("Time exceeded while waiting for nodes to become reachable"
1134 " again:\n - %s", " - ".join(helper.down))
1138 def _MaybeInstanceStartup(opts, inst_map, nodes_online,
1139 _instance_start_fn=_InstanceStart):
1140 """Start the instances conditional based on node_states.
1142 @param opts: The command line options selected by the user
1143 @param inst_map: A dict of inst -> nodes mapping
1144 @param nodes_online: A list of nodes online
1145 @param _instance_start_fn: Callback to start instances (unittest use only)
1146 @return: Success of the operation on all instances
1149 start_inst_list = []
1150 for (inst, nodes) in inst_map.items():
1151 if not (nodes - nodes_online):
1152 # All nodes the instance lives on are back online
1153 start_inst_list.append(inst)
1155 for inst in start_inst_list:
1159 return _instance_start_fn(opts, start_inst_list, True)
1164 def _EpoOn(opts, full_node_list, node_list, inst_map):
1165 """Does the actual power on.
1167 @param opts: The command line options selected by the user
1168 @param full_node_list: All nodes to operate on (includes nodes not supporting
1170 @param node_list: The list of nodes to operate on (all need to support OOB)
1171 @param inst_map: A dict of inst -> nodes mapping
1172 @return: The desired exit status
1175 if node_list and not _OobPower(opts, node_list, False):
1176 ToStderr("Not all nodes seem to get back up, investigate and start"
1177 " manually if needed")
1179 # Wait for the nodes to be back up
1180 action_cb = compat.partial(_MaybeInstanceStartup, opts, dict(inst_map))
1182 ToStdout("Waiting until all nodes are available again")
1183 if not _RunWhenNodesReachable(full_node_list, action_cb, _EPO_PING_INTERVAL):
1184 ToStderr("Please investigate and start stopped instances manually")
1185 return constants.EXIT_FAILURE
1187 return constants.EXIT_SUCCESS
1190 def _EpoOff(opts, node_list, inst_map):
1191 """Does the actual power off.
1193 @param opts: The command line options selected by the user
1194 @param node_list: The list of nodes to operate on (all need to support OOB)
1195 @param inst_map: A dict of inst -> nodes mapping
1196 @return: The desired exit status
1199 if not _InstanceStart(opts, inst_map.keys(), False):
1200 ToStderr("Please investigate and stop instances manually before continuing")
1201 return constants.EXIT_FAILURE
1204 return constants.EXIT_SUCCESS
1206 if _OobPower(opts, node_list, False):
1207 return constants.EXIT_SUCCESS
1209 return constants.EXIT_FAILURE
1212 def Epo(opts, args):
1215 @param opts: the command line options selected by the user
1217 @param args: should contain only one element, the subcommand
1219 @return: the desired exit code
1222 if opts.groups and opts.show_all:
1223 ToStderr("Only one of --groups or --all are allowed")
1224 return constants.EXIT_FAILURE
1225 elif args and opts.show_all:
1226 ToStderr("Arguments in combination with --all are not allowed")
1227 return constants.EXIT_FAILURE
1229 client = GetClient()
1232 node_query_list = itertools.chain(*client.QueryGroups(names=args,
1233 fields=["node_list"],
1236 node_query_list = args
1238 result = client.QueryNodes(names=node_query_list,
1239 fields=["name", "master", "pinst_list",
1240 "sinst_list", "powered", "offline"],
1244 for (idx, (node, master, pinsts, sinsts, powered,
1245 offline)) in enumerate(result):
1246 # Normalize the node_query_list as well
1247 if not opts.show_all:
1248 node_query_list[idx] = node
1250 for inst in (pinsts + sinsts):
1251 if inst in inst_map:
1253 inst_map[inst].add(node)
1255 inst_map[inst] = set()
1257 inst_map[inst] = set([node])
1259 if master and opts.on:
1260 # We ignore the master for turning on the machines, in fact we are
1261 # already operating on the master at this point :)
1263 elif master and not opts.show_all:
1264 ToStderr("%s is the master node, please do a master-failover to another"
1265 " node not affected by the EPO or use --all if you intend to"
1266 " shutdown the whole cluster", node)
1267 return constants.EXIT_FAILURE
1268 elif powered is None:
1269 ToStdout("Node %s does not support out-of-band handling, it can not be"
1270 " handled in a fully automated manner", node)
1271 elif powered == opts.on:
1272 ToStdout("Node %s is already in desired power state, skipping", node)
1273 elif not offline or (offline and powered):
1274 node_list.append(node)
1276 if not opts.force and not ConfirmOperation(node_query_list, "nodes", "epo"):
1277 return constants.EXIT_FAILURE
1280 return _EpoOn(opts, node_query_list, node_list, inst_map)
1282 return _EpoOff(opts, node_list, inst_map)
1287 InitCluster, [ArgHost(min=1, max=1)],
1288 [BACKEND_OPT, CP_SIZE_OPT, ENABLED_HV_OPT, GLOBAL_FILEDIR_OPT,
1289 HVLIST_OPT, MAC_PREFIX_OPT, MASTER_NETDEV_OPT, NIC_PARAMS_OPT,
1290 NOLVM_STORAGE_OPT, NOMODIFY_ETCHOSTS_OPT, NOMODIFY_SSH_SETUP_OPT,
1291 SECONDARY_IP_OPT, VG_NAME_OPT, MAINTAIN_NODE_HEALTH_OPT,
1292 UIDPOOL_OPT, DRBD_HELPER_OPT, NODRBD_STORAGE_OPT,
1293 DEFAULT_IALLOCATOR_OPT, PRIMARY_IP_VERSION_OPT, PREALLOC_WIPE_DISKS_OPT,
1294 NODE_PARAMS_OPT, GLOBAL_SHARED_FILEDIR_OPT],
1295 "[opts...] <cluster_name>", "Initialises a new cluster configuration"),
1297 DestroyCluster, ARGS_NONE, [YES_DOIT_OPT],
1298 "", "Destroy cluster"),
1300 RenameCluster, [ArgHost(min=1, max=1)],
1301 [FORCE_OPT, DRY_RUN_OPT],
1303 "Renames the cluster"),
1305 RedistributeConfig, ARGS_NONE, [SUBMIT_OPT, DRY_RUN_OPT, PRIORITY_OPT],
1306 "", "Forces a push of the configuration file and ssconf files"
1307 " to the nodes in the cluster"),
1309 VerifyCluster, ARGS_NONE,
1310 [VERBOSE_OPT, DEBUG_SIMERR_OPT, ERROR_CODES_OPT, NONPLUS1_OPT,
1311 DRY_RUN_OPT, PRIORITY_OPT, NODEGROUP_OPT],
1312 "", "Does a check on the cluster configuration"),
1314 VerifyDisks, ARGS_NONE, [PRIORITY_OPT],
1315 "", "Does a check on the cluster disk status"),
1316 "repair-disk-sizes": (
1317 RepairDiskSizes, ARGS_MANY_INSTANCES, [DRY_RUN_OPT, PRIORITY_OPT],
1318 "[instance...]", "Updates mismatches in recorded disk sizes"),
1319 "master-failover": (
1320 MasterFailover, ARGS_NONE, [NOVOTING_OPT],
1321 "", "Makes the current node the master"),
1323 MasterPing, ARGS_NONE, [],
1324 "", "Checks if the master is alive"),
1326 ShowClusterVersion, ARGS_NONE, [],
1327 "", "Shows the cluster version"),
1329 ShowClusterMaster, ARGS_NONE, [],
1330 "", "Shows the cluster master"),
1332 ClusterCopyFile, [ArgFile(min=1, max=1)],
1333 [NODE_LIST_OPT, USE_REPL_NET_OPT, NODEGROUP_OPT],
1334 "[-n node...] <filename>", "Copies a file to all (or only some) nodes"),
1336 RunClusterCommand, [ArgCommand(min=1)],
1337 [NODE_LIST_OPT, NODEGROUP_OPT],
1338 "[-n node...] <command>", "Runs a command on all (or only some) nodes"),
1340 ShowClusterConfig, ARGS_NONE, [ROMAN_OPT],
1341 "[--roman]", "Show cluster configuration"),
1343 ListTags, ARGS_NONE, [], "", "List the tags of the cluster"),
1345 AddTags, [ArgUnknown()], [TAG_SRC_OPT, PRIORITY_OPT],
1346 "tag...", "Add tags to the cluster"),
1348 RemoveTags, [ArgUnknown()], [TAG_SRC_OPT, PRIORITY_OPT],
1349 "tag...", "Remove tags from the cluster"),
1351 SearchTags, [ArgUnknown(min=1, max=1)], [PRIORITY_OPT], "",
1352 "Searches the tags on all objects on"
1353 " the cluster for a given pattern (regex)"),
1356 [ArgChoice(min=1, max=1, choices=["drain", "undrain", "info"])],
1357 [], "drain|undrain|info", "Change queue properties"),
1360 [ArgChoice(min=1, max=1, choices=["pause", "continue", "info"]),
1361 ArgSuggest(min=0, max=1, choices=["30m", "1h", "4h"])],
1363 "{pause <timespec>|continue|info}", "Change watcher properties"),
1365 SetClusterParams, ARGS_NONE,
1366 [BACKEND_OPT, CP_SIZE_OPT, ENABLED_HV_OPT, HVLIST_OPT, MASTER_NETDEV_OPT,
1367 NIC_PARAMS_OPT, NOLVM_STORAGE_OPT, VG_NAME_OPT, MAINTAIN_NODE_HEALTH_OPT,
1368 UIDPOOL_OPT, ADD_UIDS_OPT, REMOVE_UIDS_OPT, DRBD_HELPER_OPT,
1369 NODRBD_STORAGE_OPT, DEFAULT_IALLOCATOR_OPT, RESERVED_LVS_OPT,
1370 DRY_RUN_OPT, PRIORITY_OPT, PREALLOC_WIPE_DISKS_OPT, NODE_PARAMS_OPT],
1372 "Alters the parameters of the cluster"),
1374 RenewCrypto, ARGS_NONE,
1375 [NEW_CLUSTER_CERT_OPT, NEW_RAPI_CERT_OPT, RAPI_CERT_OPT,
1376 NEW_CONFD_HMAC_KEY_OPT, FORCE_OPT,
1377 NEW_CLUSTER_DOMAIN_SECRET_OPT, CLUSTER_DOMAIN_SECRET_OPT],
1379 "Renews cluster certificates, keys and secrets"),
1381 Epo, [ArgUnknown()],
1382 [FORCE_OPT, ON_OPT, GROUPS_OPT, ALL_OPT, OOB_TIMEOUT_OPT,
1383 SHUTDOWN_TIMEOUT_OPT, POWER_DELAY_OPT],
1385 "Performs an emergency power-off on given args"),
1386 "activate-master-ip": (
1387 ActivateMasterIp, ARGS_NONE, [], "", "Activates the master IP"),
1388 "deactivate-master-ip": (
1389 DeactivateMasterIp, ARGS_NONE, [CONFIRM_OPT], "",
1390 "Deactivates the master IP"),
1394 #: dictionary with aliases for commands
1396 "masterfailover": "master-failover",
1401 return GenericMain(commands, override={"tag_type": constants.TAG_CLUSTER},