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 master_netmask = opts.master_netmask
144 if master_netmask is not None:
145 master_netmask = int(master_netmask)
146 except (ValueError, TypeError), err:
147 ToStderr("Invalid master netmask value: %s" % str(err))
150 bootstrap.InitCluster(cluster_name=args[0],
151 secondary_ip=opts.secondary_ip,
153 mac_prefix=opts.mac_prefix,
154 master_netmask=master_netmask,
155 master_netdev=master_netdev,
156 file_storage_dir=opts.file_storage_dir,
157 shared_file_storage_dir=opts.shared_file_storage_dir,
158 enabled_hypervisors=hvlist,
163 candidate_pool_size=opts.candidate_pool_size,
164 modify_etc_hosts=opts.modify_etc_hosts,
165 modify_ssh_setup=opts.modify_ssh_setup,
166 maintain_node_health=opts.maintain_node_health,
167 drbd_helper=drbd_helper,
169 default_iallocator=opts.default_iallocator,
170 primary_ip_version=primary_ip_version,
171 prealloc_wipe_disks=opts.prealloc_wipe_disks,
173 op = opcodes.OpClusterPostInit()
174 SubmitOpCode(op, opts=opts)
179 def DestroyCluster(opts, args):
180 """Destroy the cluster.
182 @param opts: the command line options selected by the user
184 @param args: should be an empty list
186 @return: the desired exit code
189 if not opts.yes_do_it:
190 ToStderr("Destroying a cluster is irreversible. If you really want"
191 " destroy this cluster, supply the --yes-do-it option.")
194 op = opcodes.OpClusterDestroy()
195 master = SubmitOpCode(op, opts=opts)
196 # if we reached this, the opcode didn't fail; we can proceed to
197 # shutdown all the daemons
198 bootstrap.FinalizeClusterDestroy(master)
202 def RenameCluster(opts, args):
203 """Rename the cluster.
205 @param opts: the command line options selected by the user
207 @param args: should contain only one element, the new cluster name
209 @return: the desired exit code
214 (cluster_name, ) = cl.QueryConfigValues(["cluster_name"])
218 usertext = ("This will rename the cluster from '%s' to '%s'. If you are"
219 " connected over the network to the cluster name, the"
220 " operation is very dangerous as the IP address will be"
221 " removed from the node and the change may not go through."
222 " Continue?") % (cluster_name, new_name)
223 if not AskUser(usertext):
226 op = opcodes.OpClusterRename(name=new_name)
227 result = SubmitOpCode(op, opts=opts, cl=cl)
230 ToStdout("Cluster renamed from '%s' to '%s'", cluster_name, result)
235 def ActivateMasterIp(opts, args):
236 """Activates the master IP.
239 op = opcodes.OpClusterActivateMasterIp()
244 def DeactivateMasterIp(opts, args):
245 """Deactivates the master IP.
249 usertext = ("This will disable the master IP. All the open connections to"
250 " the master IP will be closed. To reach the master you will"
251 " need to use its node IP."
253 if not AskUser(usertext):
256 op = opcodes.OpClusterDeactivateMasterIp()
261 def RedistributeConfig(opts, args):
262 """Forces push of the cluster configuration.
264 @param opts: the command line options selected by the user
266 @param args: empty list
268 @return: the desired exit code
271 op = opcodes.OpClusterRedistConf()
272 SubmitOrSend(op, opts)
276 def ShowClusterVersion(opts, args):
277 """Write version of ganeti software to the standard output.
279 @param opts: the command line options selected by the user
281 @param args: should be an empty list
283 @return: the desired exit code
287 result = cl.QueryClusterInfo()
288 ToStdout("Software version: %s", result["software_version"])
289 ToStdout("Internode protocol: %s", result["protocol_version"])
290 ToStdout("Configuration format: %s", result["config_version"])
291 ToStdout("OS api version: %s", result["os_api_version"])
292 ToStdout("Export interface: %s", result["export_version"])
296 def ShowClusterMaster(opts, args):
297 """Write name of master node to the standard output.
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
306 master = bootstrap.GetMaster()
311 def _PrintGroupedParams(paramsdict, level=1, roman=False):
312 """Print Grouped parameters (be, nic, disk) by group.
314 @type paramsdict: dict of dicts
315 @param paramsdict: {group: {param: value, ...}, ...}
317 @param level: Level of indention
321 for item, val in sorted(paramsdict.items()):
322 if isinstance(val, dict):
323 ToStdout("%s- %s:", indent, item)
324 _PrintGroupedParams(val, level=level + 1, roman=roman)
325 elif roman and isinstance(val, int):
326 ToStdout("%s %s: %s", indent, item, compat.TryToRoman(val))
328 ToStdout("%s %s: %s", indent, item, val)
331 def ShowClusterConfig(opts, args):
332 """Shows cluster information.
334 @param opts: the command line options selected by the user
336 @param args: should be an empty list
338 @return: the desired exit code
342 result = cl.QueryClusterInfo()
344 ToStdout("Cluster name: %s", result["name"])
345 ToStdout("Cluster UUID: %s", result["uuid"])
347 ToStdout("Creation time: %s", utils.FormatTime(result["ctime"]))
348 ToStdout("Modification time: %s", utils.FormatTime(result["mtime"]))
350 ToStdout("Master node: %s", result["master"])
352 ToStdout("Architecture (this node): %s (%s)",
353 result["architecture"][0], result["architecture"][1])
356 tags = utils.CommaJoin(utils.NiceSort(result["tags"]))
360 ToStdout("Tags: %s", tags)
362 ToStdout("Default hypervisor: %s", result["default_hypervisor"])
363 ToStdout("Enabled hypervisors: %s",
364 utils.CommaJoin(result["enabled_hypervisors"]))
366 ToStdout("Hypervisor parameters:")
367 _PrintGroupedParams(result["hvparams"])
369 ToStdout("OS-specific hypervisor parameters:")
370 _PrintGroupedParams(result["os_hvp"])
372 ToStdout("OS parameters:")
373 _PrintGroupedParams(result["osparams"])
375 ToStdout("Hidden OSes: %s", utils.CommaJoin(result["hidden_os"]))
376 ToStdout("Blacklisted OSes: %s", utils.CommaJoin(result["blacklisted_os"]))
378 ToStdout("Cluster parameters:")
379 ToStdout(" - candidate pool size: %s",
380 compat.TryToRoman(result["candidate_pool_size"],
381 convert=opts.roman_integers))
382 ToStdout(" - master netdev: %s", result["master_netdev"])
383 ToStdout(" - master netmask: %s", result["master_netmask"])
384 ToStdout(" - lvm volume group: %s", result["volume_group_name"])
385 if result["reserved_lvs"]:
386 reserved_lvs = utils.CommaJoin(result["reserved_lvs"])
388 reserved_lvs = "(none)"
389 ToStdout(" - lvm reserved volumes: %s", reserved_lvs)
390 ToStdout(" - drbd usermode helper: %s", result["drbd_usermode_helper"])
391 ToStdout(" - file storage path: %s", result["file_storage_dir"])
392 ToStdout(" - shared file storage path: %s",
393 result["shared_file_storage_dir"])
394 ToStdout(" - maintenance of node health: %s",
395 result["maintain_node_health"])
396 ToStdout(" - uid pool: %s",
397 uidpool.FormatUidPool(result["uid_pool"],
398 roman=opts.roman_integers))
399 ToStdout(" - default instance allocator: %s", result["default_iallocator"])
400 ToStdout(" - primary ip version: %d", result["primary_ip_version"])
401 ToStdout(" - preallocation wipe disks: %s", result["prealloc_wipe_disks"])
402 ToStdout(" - OS search path: %s", utils.CommaJoin(constants.OS_SEARCH_PATH))
404 ToStdout("Default node parameters:")
405 _PrintGroupedParams(result["ndparams"], roman=opts.roman_integers)
407 ToStdout("Default instance parameters:")
408 _PrintGroupedParams(result["beparams"], roman=opts.roman_integers)
410 ToStdout("Default nic parameters:")
411 _PrintGroupedParams(result["nicparams"], roman=opts.roman_integers)
416 def ClusterCopyFile(opts, args):
417 """Copy a file from master to some nodes.
419 @param opts: the command line options selected by the user
421 @param args: should contain only one element, the path of
422 the file to be copied
424 @return: the desired exit code
428 if not os.path.exists(filename):
429 raise errors.OpPrereqError("No such filename '%s'" % filename,
434 cluster_name = cl.QueryConfigValues(["cluster_name"])[0]
436 results = GetOnlineNodes(nodes=opts.nodes, cl=cl, filter_master=True,
437 secondary_ips=opts.use_replication_network,
438 nodegroup=opts.nodegroup)
440 srun = ssh.SshRunner(cluster_name=cluster_name)
442 if not srun.CopyFileToNode(node, filename):
443 ToStderr("Copy of file %s to node %s failed", filename, node)
448 def RunClusterCommand(opts, args):
449 """Run a command on some nodes.
451 @param opts: the command line options selected by the user
453 @param args: should contain the command to be run and its arguments
455 @return: the desired exit code
460 command = " ".join(args)
462 nodes = GetOnlineNodes(nodes=opts.nodes, cl=cl, nodegroup=opts.nodegroup)
464 cluster_name, master_node = cl.QueryConfigValues(["cluster_name",
467 srun = ssh.SshRunner(cluster_name=cluster_name)
469 # Make sure master node is at list end
470 if master_node in nodes:
471 nodes.remove(master_node)
472 nodes.append(master_node)
475 result = srun.Run(name, "root", command)
476 ToStdout("------------------------------------------------")
477 ToStdout("node: %s", name)
478 ToStdout("%s", result.output)
479 ToStdout("return code = %s", result.exit_code)
484 def VerifyCluster(opts, args):
485 """Verify integrity of cluster, performing various test on nodes.
487 @param opts: the command line options selected by the user
489 @param args: should be an empty list
491 @return: the desired exit code
496 if opts.skip_nplusone_mem:
497 skip_checks.append(constants.VERIFY_NPLUSONE_MEM)
501 op = opcodes.OpClusterVerify(verbose=opts.verbose,
502 error_codes=opts.error_codes,
503 debug_simulate_errors=opts.simulate_errors,
504 skip_checks=skip_checks,
505 group_name=opts.nodegroup)
506 result = SubmitOpCode(op, cl=cl, opts=opts)
508 # Keep track of submitted jobs
509 jex = JobExecutor(cl=cl, opts=opts)
511 for (status, job_id) in result[constants.JOB_IDS_KEY]:
512 jex.AddJobId(None, status, job_id)
514 results = jex.GetResults()
516 (bad_jobs, bad_results) = \
518 # Convert iterators to lists
521 map(compat.partial(itertools.ifilterfalse, bool),
522 # Convert result to booleans in a tuple
523 zip(*((job_success, len(op_results) == 1 and op_results[0])
524 for (job_success, op_results) in results)))))
526 if bad_jobs == 0 and bad_results == 0:
527 rcode = constants.EXIT_SUCCESS
529 rcode = constants.EXIT_FAILURE
531 ToStdout("%s job(s) failed while verifying the cluster.", bad_jobs)
536 def VerifyDisks(opts, args):
537 """Verify integrity of cluster disks.
539 @param opts: the command line options selected by the user
541 @param args: should be an empty list
543 @return: the desired exit code
548 op = opcodes.OpClusterVerifyDisks()
550 result = SubmitOpCode(op, cl=cl, opts=opts)
552 # Keep track of submitted jobs
553 jex = JobExecutor(cl=cl, opts=opts)
555 for (status, job_id) in result[constants.JOB_IDS_KEY]:
556 jex.AddJobId(None, status, job_id)
558 retcode = constants.EXIT_SUCCESS
560 for (status, result) in jex.GetResults():
562 ToStdout("Job failed: %s", result)
565 ((bad_nodes, instances, missing), ) = result
567 for node, text in bad_nodes.items():
568 ToStdout("Error gathering data on node %s: %s",
569 node, utils.SafeEncode(text[-400:]))
570 retcode = constants.EXIT_FAILURE
571 ToStdout("You need to fix these nodes first before fixing instances")
573 for iname in instances:
576 op = opcodes.OpInstanceActivateDisks(instance_name=iname)
578 ToStdout("Activating disks for instance '%s'", iname)
579 SubmitOpCode(op, opts=opts, cl=cl)
580 except errors.GenericError, err:
581 nret, msg = FormatError(err)
583 ToStderr("Error activating disks for instance %s: %s", iname, msg)
586 for iname, ival in missing.iteritems():
587 all_missing = compat.all(x[0] in bad_nodes for x in ival)
589 ToStdout("Instance %s cannot be verified as it lives on"
590 " broken nodes", iname)
592 ToStdout("Instance %s has missing logical volumes:", iname)
594 for node, vol in ival:
595 if node in bad_nodes:
596 ToStdout("\tbroken node %s /dev/%s", node, vol)
598 ToStdout("\t%s /dev/%s", node, vol)
600 ToStdout("You need to replace or recreate disks for all the above"
601 " instances if this message persists after fixing broken nodes.")
602 retcode = constants.EXIT_FAILURE
607 def RepairDiskSizes(opts, args):
608 """Verify sizes of cluster disks.
610 @param opts: the command line options selected by the user
612 @param args: optional list of instances to restrict check to
614 @return: the desired exit code
617 op = opcodes.OpClusterRepairDiskSizes(instances=args)
618 SubmitOpCode(op, opts=opts)
622 def MasterFailover(opts, args):
623 """Failover the master node.
625 This command, when run on a non-master node, will cause the current
626 master to cease being master, and the non-master to become new
629 @param opts: the command line options selected by the user
631 @param args: should be an empty list
633 @return: the desired exit code
637 usertext = ("This will perform the failover even if most other nodes"
638 " are down, or if this node is outdated. This is dangerous"
639 " as it can lead to a non-consistent cluster. Check the"
640 " gnt-cluster(8) man page before proceeding. Continue?")
641 if not AskUser(usertext):
644 return bootstrap.MasterFailover(no_voting=opts.no_voting)
647 def MasterPing(opts, args):
648 """Checks if the master is alive.
650 @param opts: the command line options selected by the user
652 @param args: should be an empty list
654 @return: the desired exit code
659 cl.QueryClusterInfo()
661 except Exception: # pylint: disable=W0703
665 def SearchTags(opts, args):
666 """Searches the tags on all the cluster.
668 @param opts: the command line options selected by the user
670 @param args: should contain only one element, the tag pattern
672 @return: the desired exit code
675 op = opcodes.OpTagsSearch(pattern=args[0])
676 result = SubmitOpCode(op, opts=opts)
679 result = list(result)
681 for path, tag in result:
682 ToStdout("%s %s", path, tag)
685 def _ReadAndVerifyCert(cert_filename, verify_private_key=False):
686 """Reads and verifies an X509 certificate.
688 @type cert_filename: string
689 @param cert_filename: the path of the file containing the certificate to
690 verify encoded in PEM format
691 @type verify_private_key: bool
692 @param verify_private_key: whether to verify the private key in addition to
693 the public certificate
695 @return: a string containing the PEM-encoded certificate.
699 pem = utils.ReadFile(cert_filename)
701 raise errors.X509CertError(cert_filename,
702 "Unable to read certificate: %s" % str(err))
705 OpenSSL.crypto.load_certificate(OpenSSL.crypto.FILETYPE_PEM, pem)
706 except Exception, err:
707 raise errors.X509CertError(cert_filename,
708 "Unable to load certificate: %s" % str(err))
710 if verify_private_key:
712 OpenSSL.crypto.load_privatekey(OpenSSL.crypto.FILETYPE_PEM, pem)
713 except Exception, err:
714 raise errors.X509CertError(cert_filename,
715 "Unable to load private key: %s" % str(err))
720 def _RenewCrypto(new_cluster_cert, new_rapi_cert, #pylint: disable=R0911
721 rapi_cert_filename, new_spice_cert, spice_cert_filename,
722 spice_cacert_filename, new_confd_hmac_key, new_cds,
723 cds_filename, force):
724 """Renews cluster certificates, keys and secrets.
726 @type new_cluster_cert: bool
727 @param new_cluster_cert: Whether to generate a new cluster certificate
728 @type new_rapi_cert: bool
729 @param new_rapi_cert: Whether to generate a new RAPI certificate
730 @type rapi_cert_filename: string
731 @param rapi_cert_filename: Path to file containing new RAPI certificate
732 @type new_spice_cert: bool
733 @param new_spice_cert: Whether to generate a new SPICE certificate
734 @type spice_cert_filename: string
735 @param spice_cert_filename: Path to file containing new SPICE certificate
736 @type spice_cacert_filename: string
737 @param spice_cacert_filename: Path to file containing the certificate of the
738 CA that signed the SPICE certificate
739 @type new_confd_hmac_key: bool
740 @param new_confd_hmac_key: Whether to generate a new HMAC key
742 @param new_cds: Whether to generate a new cluster domain secret
743 @type cds_filename: string
744 @param cds_filename: Path to file containing new cluster domain secret
746 @param force: Whether to ask user for confirmation
749 if new_rapi_cert and rapi_cert_filename:
750 ToStderr("Only one of the --new-rapi-certificate and --rapi-certificate"
751 " options can be specified at the same time.")
754 if new_cds and cds_filename:
755 ToStderr("Only one of the --new-cluster-domain-secret and"
756 " --cluster-domain-secret options can be specified at"
760 if new_spice_cert and (spice_cert_filename or spice_cacert_filename):
761 ToStderr("When using --new-spice-certificate, the --spice-certificate"
762 " and --spice-ca-certificate must not be used.")
765 if bool(spice_cacert_filename) ^ bool(spice_cert_filename):
766 ToStderr("Both --spice-certificate and --spice-ca-certificate must be"
770 rapi_cert_pem, spice_cert_pem, spice_cacert_pem = (None, None, None)
772 if rapi_cert_filename:
773 rapi_cert_pem = _ReadAndVerifyCert(rapi_cert_filename, True)
774 if spice_cert_filename:
775 spice_cert_pem = _ReadAndVerifyCert(spice_cert_filename, True)
776 spice_cacert_pem = _ReadAndVerifyCert(spice_cacert_filename)
777 except errors.X509CertError, err:
778 ToStderr("Unable to load X509 certificate from %s: %s", err[0], err[1])
783 cds = utils.ReadFile(cds_filename)
784 except Exception, err: # pylint: disable=W0703
785 ToStderr("Can't load new cluster domain secret from %s: %s" %
786 (cds_filename, str(err)))
792 usertext = ("This requires all daemons on all nodes to be restarted and"
793 " may take some time. Continue?")
794 if not AskUser(usertext):
797 def _RenewCryptoInner(ctx):
798 ctx.feedback_fn("Updating certificates and keys")
799 bootstrap.GenerateClusterCrypto(new_cluster_cert,
804 rapi_cert_pem=rapi_cert_pem,
805 spice_cert_pem=spice_cert_pem,
806 spice_cacert_pem=spice_cacert_pem,
812 files_to_copy.append(constants.NODED_CERT_FILE)
814 if new_rapi_cert or rapi_cert_pem:
815 files_to_copy.append(constants.RAPI_CERT_FILE)
817 if new_spice_cert or spice_cert_pem:
818 files_to_copy.append(constants.SPICE_CERT_FILE)
819 files_to_copy.append(constants.SPICE_CACERT_FILE)
821 if new_confd_hmac_key:
822 files_to_copy.append(constants.CONFD_HMAC_KEY)
825 files_to_copy.append(constants.CLUSTER_DOMAIN_SECRET_FILE)
828 for node_name in ctx.nonmaster_nodes:
829 ctx.feedback_fn("Copying %s to %s" %
830 (", ".join(files_to_copy), node_name))
831 for file_name in files_to_copy:
832 ctx.ssh.CopyFileToNode(node_name, file_name)
834 RunWhileClusterStopped(ToStdout, _RenewCryptoInner)
836 ToStdout("All requested certificates and keys have been replaced."
837 " Running \"gnt-cluster verify\" now is recommended.")
842 def RenewCrypto(opts, args):
843 """Renews cluster certificates, keys and secrets.
846 return _RenewCrypto(opts.new_cluster_cert,
852 opts.new_confd_hmac_key,
853 opts.new_cluster_domain_secret,
854 opts.cluster_domain_secret,
858 def SetClusterParams(opts, args):
859 """Modify the cluster.
861 @param opts: the command line options selected by the user
863 @param args: should be an empty list
865 @return: the desired exit code
868 if not (not opts.lvm_storage or opts.vg_name or
869 not opts.drbd_storage or opts.drbd_helper or
870 opts.enabled_hypervisors or opts.hvparams or
871 opts.beparams or opts.nicparams or opts.ndparams or
872 opts.candidate_pool_size is not None or
873 opts.uid_pool is not None or
874 opts.maintain_node_health is not None or
875 opts.add_uids is not None or
876 opts.remove_uids is not None or
877 opts.default_iallocator is not None or
878 opts.reserved_lvs is not None or
879 opts.master_netdev is not None or
880 opts.master_netmask is not None or
881 opts.prealloc_wipe_disks is not None):
882 ToStderr("Please give at least one of the parameters.")
885 vg_name = opts.vg_name
886 if not opts.lvm_storage and opts.vg_name:
887 ToStderr("Options --no-lvm-storage and --vg-name conflict.")
890 if not opts.lvm_storage:
893 drbd_helper = opts.drbd_helper
894 if not opts.drbd_storage and opts.drbd_helper:
895 ToStderr("Options --no-drbd-storage and --drbd-usermode-helper conflict.")
898 if not opts.drbd_storage:
901 hvlist = opts.enabled_hypervisors
902 if hvlist is not None:
903 hvlist = hvlist.split(",")
905 # a list of (name, dict) we can pass directly to dict() (or [])
906 hvparams = dict(opts.hvparams)
907 for hv_params in hvparams.values():
908 utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES)
910 beparams = opts.beparams
911 utils.ForceDictType(beparams, constants.BES_PARAMETER_TYPES)
913 nicparams = opts.nicparams
914 utils.ForceDictType(nicparams, constants.NICS_PARAMETER_TYPES)
916 ndparams = opts.ndparams
917 if ndparams is not None:
918 utils.ForceDictType(ndparams, constants.NDS_PARAMETER_TYPES)
920 mnh = opts.maintain_node_health
922 uid_pool = opts.uid_pool
923 if uid_pool is not None:
924 uid_pool = uidpool.ParseUidPool(uid_pool)
926 add_uids = opts.add_uids
927 if add_uids is not None:
928 add_uids = uidpool.ParseUidPool(add_uids)
930 remove_uids = opts.remove_uids
931 if remove_uids is not None:
932 remove_uids = uidpool.ParseUidPool(remove_uids)
934 if opts.reserved_lvs is not None:
935 if opts.reserved_lvs == "":
936 opts.reserved_lvs = []
938 opts.reserved_lvs = utils.UnescapeAndSplit(opts.reserved_lvs, sep=",")
940 if opts.master_netmask is not None:
942 opts.master_netmask = int(opts.master_netmask)
944 ToStderr("The --master-netmask option expects an int parameter.")
947 op = opcodes.OpClusterSetParams(vg_name=vg_name,
948 drbd_helper=drbd_helper,
949 enabled_hypervisors=hvlist,
955 candidate_pool_size=opts.candidate_pool_size,
956 maintain_node_health=mnh,
959 remove_uids=remove_uids,
960 default_iallocator=opts.default_iallocator,
961 prealloc_wipe_disks=opts.prealloc_wipe_disks,
962 master_netdev=opts.master_netdev,
963 master_netmask=opts.master_netmask,
964 reserved_lvs=opts.reserved_lvs)
965 SubmitOpCode(op, opts=opts)
969 def QueueOps(opts, args):
972 @param opts: the command line options selected by the user
974 @param args: should contain only one element, the subcommand
976 @return: the desired exit code
981 if command in ("drain", "undrain"):
982 drain_flag = command == "drain"
983 client.SetQueueDrainFlag(drain_flag)
984 elif command == "info":
985 result = client.QueryConfigValues(["drain_flag"])
990 ToStdout("The drain flag is %s" % val)
992 raise errors.OpPrereqError("Command '%s' is not valid." % command,
998 def _ShowWatcherPause(until):
999 if until is None or until < time.time():
1000 ToStdout("The watcher is not paused.")
1002 ToStdout("The watcher is paused until %s.", time.ctime(until))
1005 def WatcherOps(opts, args):
1006 """Watcher operations.
1008 @param opts: the command line options selected by the user
1010 @param args: should contain only one element, the subcommand
1012 @return: the desired exit code
1016 client = GetClient()
1018 if command == "continue":
1019 client.SetWatcherPause(None)
1020 ToStdout("The watcher is no longer paused.")
1022 elif command == "pause":
1024 raise errors.OpPrereqError("Missing pause duration", errors.ECODE_INVAL)
1026 result = client.SetWatcherPause(time.time() + ParseTimespec(args[1]))
1027 _ShowWatcherPause(result)
1029 elif command == "info":
1030 result = client.QueryConfigValues(["watcher_pause"])
1031 _ShowWatcherPause(result[0])
1034 raise errors.OpPrereqError("Command '%s' is not valid." % command,
1040 def _OobPower(opts, node_list, power):
1041 """Puts the node in the list to desired power state.
1043 @param opts: The command line options selected by the user
1044 @param node_list: The list of nodes to operate on
1045 @param power: True if they should be powered on, False otherwise
1046 @return: The success of the operation (none failed)
1050 command = constants.OOB_POWER_ON
1052 command = constants.OOB_POWER_OFF
1054 op = opcodes.OpOobCommand(node_names=node_list,
1057 timeout=opts.oob_timeout,
1058 power_delay=opts.power_delay)
1059 result = SubmitOpCode(op, opts=opts)
1061 for node_result in result:
1062 (node_tuple, data_tuple) = node_result
1063 (_, node_name) = node_tuple
1064 (data_status, _) = data_tuple
1065 if data_status != constants.RS_NORMAL:
1066 assert data_status != constants.RS_UNAVAIL
1068 ToStderr("There was a problem changing power for %s, please investigate",
1077 def _InstanceStart(opts, inst_list, start):
1078 """Puts the instances in the list to desired state.
1080 @param opts: The command line options selected by the user
1081 @param inst_list: The list of instances to operate on
1082 @param start: True if they should be started, False for shutdown
1083 @return: The success of the operation (none failed)
1087 opcls = opcodes.OpInstanceStartup
1088 text_submit, text_success, text_failed = ("startup", "started", "starting")
1090 opcls = compat.partial(opcodes.OpInstanceShutdown,
1091 timeout=opts.shutdown_timeout)
1092 text_submit, text_success, text_failed = ("shutdown", "stopped", "stopping")
1094 jex = JobExecutor(opts=opts)
1096 for inst in inst_list:
1097 ToStdout("Submit %s of instance %s", text_submit, inst)
1098 op = opcls(instance_name=inst)
1099 jex.QueueJob(inst, op)
1101 results = jex.GetResults()
1102 bad_cnt = len([1 for (success, _) in results if not success])
1105 ToStdout("All instances have been %s successfully", text_success)
1107 ToStderr("There were errors while %s instances:\n"
1108 "%d error(s) out of %d instance(s)", text_failed, bad_cnt,
1115 class _RunWhenNodesReachableHelper:
1116 """Helper class to make shared internal state sharing easier.
1118 @ivar success: Indicates if all action_cb calls were successful
1121 def __init__(self, node_list, action_cb, node2ip, port, feedback_fn,
1122 _ping_fn=netutils.TcpPing, _sleep_fn=time.sleep):
1125 @param node_list: The list of nodes to be reachable
1126 @param action_cb: Callback called when a new host is reachable
1128 @param node2ip: Node to ip mapping
1129 @param port: The port to use for the TCP ping
1130 @param feedback_fn: The function used for feedback
1131 @param _ping_fn: Function to check reachabilty (for unittest use only)
1132 @param _sleep_fn: Function to sleep (for unittest use only)
1135 self.down = set(node_list)
1137 self.node2ip = node2ip
1139 self.action_cb = action_cb
1141 self.feedback_fn = feedback_fn
1142 self._ping_fn = _ping_fn
1143 self._sleep_fn = _sleep_fn
1146 """When called we run action_cb.
1148 @raises utils.RetryAgain: When there are still down nodes
1151 if not self.action_cb(self.up):
1152 self.success = False
1155 raise utils.RetryAgain()
1159 def Wait(self, secs):
1160 """Checks if a host is up or waits remaining seconds.
1162 @param secs: The secs remaining
1166 for node in self.down:
1167 if self._ping_fn(self.node2ip[node], self.port, timeout=_EPO_PING_TIMEOUT,
1168 live_port_needed=True):
1169 self.feedback_fn("Node %s became available" % node)
1171 self.down -= self.up
1172 # If we have a node available there is the possibility to run the
1173 # action callback successfully, therefore we don't wait and return
1176 self._sleep_fn(max(0.0, start + secs - time.time()))
1179 def _RunWhenNodesReachable(node_list, action_cb, interval):
1180 """Run action_cb when nodes become reachable.
1182 @param node_list: The list of nodes to be reachable
1183 @param action_cb: Callback called when a new host is reachable
1184 @param interval: The earliest time to retry
1187 client = GetClient()
1188 cluster_info = client.QueryClusterInfo()
1189 if cluster_info["primary_ip_version"] == constants.IP4_VERSION:
1190 family = netutils.IPAddress.family
1192 family = netutils.IP6Address.family
1194 node2ip = dict((node, netutils.GetHostname(node, family=family).ip)
1195 for node in node_list)
1197 port = netutils.GetDaemonPort(constants.NODED)
1198 helper = _RunWhenNodesReachableHelper(node_list, action_cb, node2ip, port,
1202 return utils.Retry(helper, interval, _EPO_REACHABLE_TIMEOUT,
1203 wait_fn=helper.Wait)
1204 except utils.RetryTimeout:
1205 ToStderr("Time exceeded while waiting for nodes to become reachable"
1206 " again:\n - %s", " - ".join(helper.down))
1210 def _MaybeInstanceStartup(opts, inst_map, nodes_online,
1211 _instance_start_fn=_InstanceStart):
1212 """Start the instances conditional based on node_states.
1214 @param opts: The command line options selected by the user
1215 @param inst_map: A dict of inst -> nodes mapping
1216 @param nodes_online: A list of nodes online
1217 @param _instance_start_fn: Callback to start instances (unittest use only)
1218 @return: Success of the operation on all instances
1221 start_inst_list = []
1222 for (inst, nodes) in inst_map.items():
1223 if not (nodes - nodes_online):
1224 # All nodes the instance lives on are back online
1225 start_inst_list.append(inst)
1227 for inst in start_inst_list:
1231 return _instance_start_fn(opts, start_inst_list, True)
1236 def _EpoOn(opts, full_node_list, node_list, inst_map):
1237 """Does the actual power on.
1239 @param opts: The command line options selected by the user
1240 @param full_node_list: All nodes to operate on (includes nodes not supporting
1242 @param node_list: The list of nodes to operate on (all need to support OOB)
1243 @param inst_map: A dict of inst -> nodes mapping
1244 @return: The desired exit status
1247 if node_list and not _OobPower(opts, node_list, False):
1248 ToStderr("Not all nodes seem to get back up, investigate and start"
1249 " manually if needed")
1251 # Wait for the nodes to be back up
1252 action_cb = compat.partial(_MaybeInstanceStartup, opts, dict(inst_map))
1254 ToStdout("Waiting until all nodes are available again")
1255 if not _RunWhenNodesReachable(full_node_list, action_cb, _EPO_PING_INTERVAL):
1256 ToStderr("Please investigate and start stopped instances manually")
1257 return constants.EXIT_FAILURE
1259 return constants.EXIT_SUCCESS
1262 def _EpoOff(opts, node_list, inst_map):
1263 """Does the actual power off.
1265 @param opts: The command line options selected by the user
1266 @param node_list: The list of nodes to operate on (all need to support OOB)
1267 @param inst_map: A dict of inst -> nodes mapping
1268 @return: The desired exit status
1271 if not _InstanceStart(opts, inst_map.keys(), False):
1272 ToStderr("Please investigate and stop instances manually before continuing")
1273 return constants.EXIT_FAILURE
1276 return constants.EXIT_SUCCESS
1278 if _OobPower(opts, node_list, False):
1279 return constants.EXIT_SUCCESS
1281 return constants.EXIT_FAILURE
1284 def Epo(opts, args):
1287 @param opts: the command line options selected by the user
1289 @param args: should contain only one element, the subcommand
1291 @return: the desired exit code
1294 if opts.groups and opts.show_all:
1295 ToStderr("Only one of --groups or --all are allowed")
1296 return constants.EXIT_FAILURE
1297 elif args and opts.show_all:
1298 ToStderr("Arguments in combination with --all are not allowed")
1299 return constants.EXIT_FAILURE
1301 client = GetClient()
1304 node_query_list = itertools.chain(*client.QueryGroups(names=args,
1305 fields=["node_list"],
1308 node_query_list = args
1310 result = client.QueryNodes(names=node_query_list,
1311 fields=["name", "master", "pinst_list",
1312 "sinst_list", "powered", "offline"],
1316 for (idx, (node, master, pinsts, sinsts, powered,
1317 offline)) in enumerate(result):
1318 # Normalize the node_query_list as well
1319 if not opts.show_all:
1320 node_query_list[idx] = node
1322 for inst in (pinsts + sinsts):
1323 if inst in inst_map:
1325 inst_map[inst].add(node)
1327 inst_map[inst] = set()
1329 inst_map[inst] = set([node])
1331 if master and opts.on:
1332 # We ignore the master for turning on the machines, in fact we are
1333 # already operating on the master at this point :)
1335 elif master and not opts.show_all:
1336 ToStderr("%s is the master node, please do a master-failover to another"
1337 " node not affected by the EPO or use --all if you intend to"
1338 " shutdown the whole cluster", node)
1339 return constants.EXIT_FAILURE
1340 elif powered is None:
1341 ToStdout("Node %s does not support out-of-band handling, it can not be"
1342 " handled in a fully automated manner", node)
1343 elif powered == opts.on:
1344 ToStdout("Node %s is already in desired power state, skipping", node)
1345 elif not offline or (offline and powered):
1346 node_list.append(node)
1348 if not opts.force and not ConfirmOperation(node_query_list, "nodes", "epo"):
1349 return constants.EXIT_FAILURE
1352 return _EpoOn(opts, node_query_list, node_list, inst_map)
1354 return _EpoOff(opts, node_list, inst_map)
1359 InitCluster, [ArgHost(min=1, max=1)],
1360 [BACKEND_OPT, CP_SIZE_OPT, ENABLED_HV_OPT, GLOBAL_FILEDIR_OPT,
1361 HVLIST_OPT, MAC_PREFIX_OPT, MASTER_NETDEV_OPT, MASTER_NETMASK_OPT,
1362 NIC_PARAMS_OPT, NOLVM_STORAGE_OPT, NOMODIFY_ETCHOSTS_OPT,
1363 NOMODIFY_SSH_SETUP_OPT, SECONDARY_IP_OPT, VG_NAME_OPT,
1364 MAINTAIN_NODE_HEALTH_OPT, UIDPOOL_OPT, DRBD_HELPER_OPT, NODRBD_STORAGE_OPT,
1365 DEFAULT_IALLOCATOR_OPT, PRIMARY_IP_VERSION_OPT, PREALLOC_WIPE_DISKS_OPT,
1366 NODE_PARAMS_OPT, GLOBAL_SHARED_FILEDIR_OPT],
1367 "[opts...] <cluster_name>", "Initialises a new cluster configuration"),
1369 DestroyCluster, ARGS_NONE, [YES_DOIT_OPT],
1370 "", "Destroy cluster"),
1372 RenameCluster, [ArgHost(min=1, max=1)],
1373 [FORCE_OPT, DRY_RUN_OPT],
1375 "Renames the cluster"),
1377 RedistributeConfig, ARGS_NONE, [SUBMIT_OPT, DRY_RUN_OPT, PRIORITY_OPT],
1378 "", "Forces a push of the configuration file and ssconf files"
1379 " to the nodes in the cluster"),
1381 VerifyCluster, ARGS_NONE,
1382 [VERBOSE_OPT, DEBUG_SIMERR_OPT, ERROR_CODES_OPT, NONPLUS1_OPT,
1383 DRY_RUN_OPT, PRIORITY_OPT, NODEGROUP_OPT],
1384 "", "Does a check on the cluster configuration"),
1386 VerifyDisks, ARGS_NONE, [PRIORITY_OPT],
1387 "", "Does a check on the cluster disk status"),
1388 "repair-disk-sizes": (
1389 RepairDiskSizes, ARGS_MANY_INSTANCES, [DRY_RUN_OPT, PRIORITY_OPT],
1390 "", "Updates mismatches in recorded disk sizes"),
1391 "master-failover": (
1392 MasterFailover, ARGS_NONE, [NOVOTING_OPT],
1393 "", "Makes the current node the master"),
1395 MasterPing, ARGS_NONE, [],
1396 "", "Checks if the master is alive"),
1398 ShowClusterVersion, ARGS_NONE, [],
1399 "", "Shows the cluster version"),
1401 ShowClusterMaster, ARGS_NONE, [],
1402 "", "Shows the cluster master"),
1404 ClusterCopyFile, [ArgFile(min=1, max=1)],
1405 [NODE_LIST_OPT, USE_REPL_NET_OPT, NODEGROUP_OPT],
1406 "[-n node...] <filename>", "Copies a file to all (or only some) nodes"),
1408 RunClusterCommand, [ArgCommand(min=1)],
1409 [NODE_LIST_OPT, NODEGROUP_OPT],
1410 "[-n node...] <command>", "Runs a command on all (or only some) nodes"),
1412 ShowClusterConfig, ARGS_NONE, [ROMAN_OPT],
1413 "[--roman]", "Show cluster configuration"),
1415 ListTags, ARGS_NONE, [], "", "List the tags of the cluster"),
1417 AddTags, [ArgUnknown()], [TAG_SRC_OPT, PRIORITY_OPT],
1418 "tag...", "Add tags to the cluster"),
1420 RemoveTags, [ArgUnknown()], [TAG_SRC_OPT, PRIORITY_OPT],
1421 "tag...", "Remove tags from the cluster"),
1423 SearchTags, [ArgUnknown(min=1, max=1)], [PRIORITY_OPT], "",
1424 "Searches the tags on all objects on"
1425 " the cluster for a given pattern (regex)"),
1428 [ArgChoice(min=1, max=1, choices=["drain", "undrain", "info"])],
1429 [], "drain|undrain|info", "Change queue properties"),
1432 [ArgChoice(min=1, max=1, choices=["pause", "continue", "info"]),
1433 ArgSuggest(min=0, max=1, choices=["30m", "1h", "4h"])],
1435 "{pause <timespec>|continue|info}", "Change watcher properties"),
1437 SetClusterParams, ARGS_NONE,
1438 [BACKEND_OPT, CP_SIZE_OPT, ENABLED_HV_OPT, HVLIST_OPT, MASTER_NETDEV_OPT,
1439 MASTER_NETMASK_OPT, NIC_PARAMS_OPT, NOLVM_STORAGE_OPT, VG_NAME_OPT,
1440 MAINTAIN_NODE_HEALTH_OPT, UIDPOOL_OPT, ADD_UIDS_OPT, REMOVE_UIDS_OPT,
1441 DRBD_HELPER_OPT, NODRBD_STORAGE_OPT, DEFAULT_IALLOCATOR_OPT,
1442 RESERVED_LVS_OPT, DRY_RUN_OPT, PRIORITY_OPT, PREALLOC_WIPE_DISKS_OPT,
1445 "Alters the parameters of the cluster"),
1447 RenewCrypto, ARGS_NONE,
1448 [NEW_CLUSTER_CERT_OPT, NEW_RAPI_CERT_OPT, RAPI_CERT_OPT,
1449 NEW_CONFD_HMAC_KEY_OPT, FORCE_OPT,
1450 NEW_CLUSTER_DOMAIN_SECRET_OPT, CLUSTER_DOMAIN_SECRET_OPT,
1451 NEW_SPICE_CERT_OPT, SPICE_CERT_OPT, SPICE_CACERT_OPT],
1453 "Renews cluster certificates, keys and secrets"),
1455 Epo, [ArgUnknown()],
1456 [FORCE_OPT, ON_OPT, GROUPS_OPT, ALL_OPT, OOB_TIMEOUT_OPT,
1457 SHUTDOWN_TIMEOUT_OPT, POWER_DELAY_OPT],
1459 "Performs an emergency power-off on given args"),
1460 "activate-master-ip": (
1461 ActivateMasterIp, ARGS_NONE, [], "", "Activates the master IP"),
1462 "deactivate-master-ip": (
1463 DeactivateMasterIp, ARGS_NONE, [CONFIRM_OPT], "",
1464 "Deactivates the master IP"),
1468 #: dictionary with aliases for commands
1470 "masterfailover": "master-failover",
1475 return GenericMain(commands, override={"tag_type": constants.TAG_CLUSTER},