root / lib / cmdlib / cluster.py @ b6dd32db
History | View | Annotate | Download (135.6 kB)
| 1 |
#
|
|---|---|
| 2 |
#
|
| 3 |
|
| 4 |
# Copyright (C) 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014 Google Inc.
|
| 5 |
#
|
| 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.
|
| 10 |
#
|
| 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.
|
| 15 |
#
|
| 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
|
| 19 |
# 02110-1301, USA.
|
| 20 |
|
| 21 |
|
| 22 |
"""Logical units dealing with the cluster."""
|
| 23 |
|
| 24 |
import copy |
| 25 |
import itertools |
| 26 |
import logging |
| 27 |
import operator |
| 28 |
import os |
| 29 |
import re |
| 30 |
import time |
| 31 |
|
| 32 |
from ganeti import compat |
| 33 |
from ganeti import constants |
| 34 |
from ganeti import errors |
| 35 |
from ganeti import hypervisor |
| 36 |
from ganeti import locking |
| 37 |
from ganeti import masterd |
| 38 |
from ganeti import netutils |
| 39 |
from ganeti import objects |
| 40 |
from ganeti import opcodes |
| 41 |
from ganeti import pathutils |
| 42 |
from ganeti import query |
| 43 |
import ganeti.rpc.node as rpc |
| 44 |
from ganeti import runtime |
| 45 |
from ganeti import ssh |
| 46 |
from ganeti import uidpool |
| 47 |
from ganeti import utils |
| 48 |
from ganeti import vcluster |
| 49 |
|
| 50 |
from ganeti.cmdlib.base import NoHooksLU, QueryBase, LogicalUnit, \ |
| 51 |
ResultWithJobs
|
| 52 |
from ganeti.cmdlib.common import ShareAll, RunPostHook, \ |
| 53 |
ComputeAncillaryFiles, RedistributeAncillaryFiles, UploadHelper, \ |
| 54 |
GetWantedInstances, MergeAndVerifyHvState, MergeAndVerifyDiskState, \ |
| 55 |
GetUpdatedIPolicy, ComputeNewInstanceViolations, GetUpdatedParams, \ |
| 56 |
CheckOSParams, CheckHVParams, AdjustCandidatePool, CheckNodePVs, \ |
| 57 |
ComputeIPolicyInstanceViolation, AnnotateDiskParams, SupportsOob, \ |
| 58 |
CheckIpolicyVsDiskTemplates, CheckDiskAccessModeValidity, \ |
| 59 |
CheckDiskAccessModeConsistency, CreateNewClientCert, \ |
| 60 |
AddInstanceCommunicationNetworkOp, ConnectInstanceCommunicationNetworkOp |
| 61 |
|
| 62 |
import ganeti.masterd.instance |
| 63 |
|
| 64 |
|
| 65 |
def _UpdateMasterClientCert( |
| 66 |
lu, master_uuid, cluster, feedback_fn, |
| 67 |
client_cert=pathutils.NODED_CLIENT_CERT_FILE, |
| 68 |
client_cert_tmp=pathutils.NODED_CLIENT_CERT_FILE_TMP): |
| 69 |
"""Renews the master's client certificate and propagates the config.
|
| 70 |
|
| 71 |
@type lu: C{LogicalUnit}
|
| 72 |
@param lu: the logical unit holding the config
|
| 73 |
@type master_uuid: string
|
| 74 |
@param master_uuid: the master node's UUID
|
| 75 |
@type cluster: C{objects.Cluster}
|
| 76 |
@param cluster: the cluster's configuration
|
| 77 |
@type feedback_fn: function
|
| 78 |
@param feedback_fn: feedback functions for config updates
|
| 79 |
@type client_cert: string
|
| 80 |
@param client_cert: the path of the client certificate
|
| 81 |
@type client_cert_tmp: string
|
| 82 |
@param client_cert_tmp: the temporary path of the client certificate
|
| 83 |
@rtype: string
|
| 84 |
@return: the digest of the newly created client certificate
|
| 85 |
|
| 86 |
"""
|
| 87 |
client_digest = CreateNewClientCert(lu, master_uuid, filename=client_cert_tmp) |
| 88 |
utils.AddNodeToCandidateCerts(master_uuid, client_digest, |
| 89 |
cluster.candidate_certs) |
| 90 |
# This triggers an update of the config and distribution of it with the old
|
| 91 |
# SSL certificate
|
| 92 |
lu.cfg.Update(cluster, feedback_fn) |
| 93 |
|
| 94 |
utils.RemoveFile(client_cert) |
| 95 |
utils.RenameFile(client_cert_tmp, client_cert) |
| 96 |
return client_digest
|
| 97 |
|
| 98 |
|
| 99 |
class LUClusterRenewCrypto(NoHooksLU): |
| 100 |
"""Renew the cluster's crypto tokens.
|
| 101 |
|
| 102 |
Note that most of this operation is done in gnt_cluster.py, this LU only
|
| 103 |
takes care of the renewal of the client SSL certificates.
|
| 104 |
|
| 105 |
"""
|
| 106 |
def Exec(self, feedback_fn): |
| 107 |
master_uuid = self.cfg.GetMasterNode()
|
| 108 |
cluster = self.cfg.GetClusterInfo()
|
| 109 |
|
| 110 |
server_digest = utils.GetCertificateDigest( |
| 111 |
cert_filename=pathutils.NODED_CERT_FILE) |
| 112 |
utils.AddNodeToCandidateCerts("%s-SERVER" % master_uuid,
|
| 113 |
server_digest, |
| 114 |
cluster.candidate_certs) |
| 115 |
new_master_digest = _UpdateMasterClientCert(self, master_uuid, cluster,
|
| 116 |
feedback_fn) |
| 117 |
|
| 118 |
cluster.candidate_certs = {master_uuid: new_master_digest}
|
| 119 |
nodes = self.cfg.GetAllNodesInfo()
|
| 120 |
for (node_uuid, node_info) in nodes.items(): |
| 121 |
if node_uuid != master_uuid:
|
| 122 |
new_digest = CreateNewClientCert(self, node_uuid)
|
| 123 |
if node_info.master_candidate:
|
| 124 |
cluster.candidate_certs[node_uuid] = new_digest |
| 125 |
# Trigger another update of the config now with the new master cert
|
| 126 |
self.cfg.Update(cluster, feedback_fn)
|
| 127 |
|
| 128 |
|
| 129 |
class LUClusterActivateMasterIp(NoHooksLU): |
| 130 |
"""Activate the master IP on the master node.
|
| 131 |
|
| 132 |
"""
|
| 133 |
def Exec(self, feedback_fn): |
| 134 |
"""Activate the master IP.
|
| 135 |
|
| 136 |
"""
|
| 137 |
master_params = self.cfg.GetMasterNetworkParameters()
|
| 138 |
ems = self.cfg.GetUseExternalMipScript()
|
| 139 |
result = self.rpc.call_node_activate_master_ip(master_params.uuid,
|
| 140 |
master_params, ems) |
| 141 |
result.Raise("Could not activate the master IP")
|
| 142 |
|
| 143 |
|
| 144 |
class LUClusterDeactivateMasterIp(NoHooksLU): |
| 145 |
"""Deactivate the master IP on the master node.
|
| 146 |
|
| 147 |
"""
|
| 148 |
def Exec(self, feedback_fn): |
| 149 |
"""Deactivate the master IP.
|
| 150 |
|
| 151 |
"""
|
| 152 |
master_params = self.cfg.GetMasterNetworkParameters()
|
| 153 |
ems = self.cfg.GetUseExternalMipScript()
|
| 154 |
result = self.rpc.call_node_deactivate_master_ip(master_params.uuid,
|
| 155 |
master_params, ems) |
| 156 |
result.Raise("Could not deactivate the master IP")
|
| 157 |
|
| 158 |
|
| 159 |
class LUClusterConfigQuery(NoHooksLU): |
| 160 |
"""Return configuration values.
|
| 161 |
|
| 162 |
"""
|
| 163 |
REQ_BGL = False
|
| 164 |
|
| 165 |
def CheckArguments(self): |
| 166 |
self.cq = ClusterQuery(None, self.op.output_fields, False) |
| 167 |
|
| 168 |
def ExpandNames(self): |
| 169 |
self.cq.ExpandNames(self) |
| 170 |
|
| 171 |
def DeclareLocks(self, level): |
| 172 |
self.cq.DeclareLocks(self, level) |
| 173 |
|
| 174 |
def Exec(self, feedback_fn): |
| 175 |
result = self.cq.OldStyleQuery(self) |
| 176 |
|
| 177 |
assert len(result) == 1 |
| 178 |
|
| 179 |
return result[0] |
| 180 |
|
| 181 |
|
| 182 |
class LUClusterDestroy(LogicalUnit): |
| 183 |
"""Logical unit for destroying the cluster.
|
| 184 |
|
| 185 |
"""
|
| 186 |
HPATH = "cluster-destroy"
|
| 187 |
HTYPE = constants.HTYPE_CLUSTER |
| 188 |
|
| 189 |
def BuildHooksEnv(self): |
| 190 |
"""Build hooks env.
|
| 191 |
|
| 192 |
"""
|
| 193 |
return {
|
| 194 |
"OP_TARGET": self.cfg.GetClusterName(), |
| 195 |
} |
| 196 |
|
| 197 |
def BuildHooksNodes(self): |
| 198 |
"""Build hooks nodes.
|
| 199 |
|
| 200 |
"""
|
| 201 |
return ([], [])
|
| 202 |
|
| 203 |
def CheckPrereq(self): |
| 204 |
"""Check prerequisites.
|
| 205 |
|
| 206 |
This checks whether the cluster is empty.
|
| 207 |
|
| 208 |
Any errors are signaled by raising errors.OpPrereqError.
|
| 209 |
|
| 210 |
"""
|
| 211 |
master = self.cfg.GetMasterNode()
|
| 212 |
|
| 213 |
nodelist = self.cfg.GetNodeList()
|
| 214 |
if len(nodelist) != 1 or nodelist[0] != master: |
| 215 |
raise errors.OpPrereqError("There are still %d node(s) in" |
| 216 |
" this cluster." % (len(nodelist) - 1), |
| 217 |
errors.ECODE_INVAL) |
| 218 |
instancelist = self.cfg.GetInstanceList()
|
| 219 |
if instancelist:
|
| 220 |
raise errors.OpPrereqError("There are still %d instance(s) in" |
| 221 |
" this cluster." % len(instancelist), |
| 222 |
errors.ECODE_INVAL) |
| 223 |
|
| 224 |
def Exec(self, feedback_fn): |
| 225 |
"""Destroys the cluster.
|
| 226 |
|
| 227 |
"""
|
| 228 |
master_params = self.cfg.GetMasterNetworkParameters()
|
| 229 |
|
| 230 |
# Run post hooks on master node before it's removed
|
| 231 |
RunPostHook(self, self.cfg.GetNodeName(master_params.uuid)) |
| 232 |
|
| 233 |
ems = self.cfg.GetUseExternalMipScript()
|
| 234 |
result = self.rpc.call_node_deactivate_master_ip(master_params.uuid,
|
| 235 |
master_params, ems) |
| 236 |
result.Warn("Error disabling the master IP address", self.LogWarning) |
| 237 |
return master_params.uuid
|
| 238 |
|
| 239 |
|
| 240 |
class LUClusterPostInit(LogicalUnit): |
| 241 |
"""Logical unit for running hooks after cluster initialization.
|
| 242 |
|
| 243 |
"""
|
| 244 |
HPATH = "cluster-init"
|
| 245 |
HTYPE = constants.HTYPE_CLUSTER |
| 246 |
|
| 247 |
def CheckArguments(self): |
| 248 |
self.master_uuid = self.cfg.GetMasterNode() |
| 249 |
self.master_ndparams = self.cfg.GetNdParams(self.cfg.GetMasterNodeInfo()) |
| 250 |
|
| 251 |
# TODO: When Issue 584 is solved, and None is properly parsed when used
|
| 252 |
# as a default value, ndparams.get(.., None) can be changed to
|
| 253 |
# ndparams[..] to access the values directly
|
| 254 |
|
| 255 |
# OpenvSwitch: Warn user if link is missing
|
| 256 |
if (self.master_ndparams[constants.ND_OVS] and not |
| 257 |
self.master_ndparams.get(constants.ND_OVS_LINK, None)): |
| 258 |
self.LogInfo("No physical interface for OpenvSwitch was given." |
| 259 |
" OpenvSwitch will not have an outside connection. This"
|
| 260 |
" might not be what you want.")
|
| 261 |
|
| 262 |
def BuildHooksEnv(self): |
| 263 |
"""Build hooks env.
|
| 264 |
|
| 265 |
"""
|
| 266 |
return {
|
| 267 |
"OP_TARGET": self.cfg.GetClusterName(), |
| 268 |
} |
| 269 |
|
| 270 |
def BuildHooksNodes(self): |
| 271 |
"""Build hooks nodes.
|
| 272 |
|
| 273 |
"""
|
| 274 |
return ([], [self.cfg.GetMasterNode()]) |
| 275 |
|
| 276 |
def Exec(self, feedback_fn): |
| 277 |
"""Create and configure Open vSwitch
|
| 278 |
|
| 279 |
"""
|
| 280 |
if self.master_ndparams[constants.ND_OVS]: |
| 281 |
result = self.rpc.call_node_configure_ovs(
|
| 282 |
self.master_uuid,
|
| 283 |
self.master_ndparams[constants.ND_OVS_NAME],
|
| 284 |
self.master_ndparams.get(constants.ND_OVS_LINK, None)) |
| 285 |
result.Raise("Could not successully configure Open vSwitch")
|
| 286 |
|
| 287 |
cluster = self.cfg.GetClusterInfo()
|
| 288 |
_UpdateMasterClientCert(self, self.master_uuid, cluster, feedback_fn) |
| 289 |
|
| 290 |
return True |
| 291 |
|
| 292 |
|
| 293 |
class ClusterQuery(QueryBase): |
| 294 |
FIELDS = query.CLUSTER_FIELDS |
| 295 |
|
| 296 |
#: Do not sort (there is only one item)
|
| 297 |
SORT_FIELD = None
|
| 298 |
|
| 299 |
def ExpandNames(self, lu): |
| 300 |
lu.needed_locks = {}
|
| 301 |
|
| 302 |
# The following variables interact with _QueryBase._GetNames
|
| 303 |
self.wanted = locking.ALL_SET
|
| 304 |
self.do_locking = self.use_locking |
| 305 |
|
| 306 |
if self.do_locking: |
| 307 |
raise errors.OpPrereqError("Can not use locking for cluster queries", |
| 308 |
errors.ECODE_INVAL) |
| 309 |
|
| 310 |
def DeclareLocks(self, lu, level): |
| 311 |
pass
|
| 312 |
|
| 313 |
def _GetQueryData(self, lu): |
| 314 |
"""Computes the list of nodes and their attributes.
|
| 315 |
|
| 316 |
"""
|
| 317 |
# Locking is not used
|
| 318 |
assert not (compat.any(lu.glm.is_owned(level) |
| 319 |
for level in locking.LEVELS |
| 320 |
if level != locking.LEVEL_CLUSTER) or |
| 321 |
self.do_locking or self.use_locking) |
| 322 |
|
| 323 |
if query.CQ_CONFIG in self.requested_data: |
| 324 |
cluster = lu.cfg.GetClusterInfo() |
| 325 |
nodes = lu.cfg.GetAllNodesInfo() |
| 326 |
else:
|
| 327 |
cluster = NotImplemented
|
| 328 |
nodes = NotImplemented
|
| 329 |
|
| 330 |
if query.CQ_QUEUE_DRAINED in self.requested_data: |
| 331 |
drain_flag = os.path.exists(pathutils.JOB_QUEUE_DRAIN_FILE) |
| 332 |
else:
|
| 333 |
drain_flag = NotImplemented
|
| 334 |
|
| 335 |
if query.CQ_WATCHER_PAUSE in self.requested_data: |
| 336 |
master_node_uuid = lu.cfg.GetMasterNode() |
| 337 |
|
| 338 |
result = lu.rpc.call_get_watcher_pause(master_node_uuid) |
| 339 |
result.Raise("Can't retrieve watcher pause from master node '%s'" %
|
| 340 |
lu.cfg.GetMasterNodeName()) |
| 341 |
|
| 342 |
watcher_pause = result.payload |
| 343 |
else:
|
| 344 |
watcher_pause = NotImplemented
|
| 345 |
|
| 346 |
return query.ClusterQueryData(cluster, nodes, drain_flag, watcher_pause)
|
| 347 |
|
| 348 |
|
| 349 |
class LUClusterQuery(NoHooksLU): |
| 350 |
"""Query cluster configuration.
|
| 351 |
|
| 352 |
"""
|
| 353 |
REQ_BGL = False
|
| 354 |
|
| 355 |
def ExpandNames(self): |
| 356 |
self.needed_locks = {}
|
| 357 |
|
| 358 |
def Exec(self, feedback_fn): |
| 359 |
"""Return cluster config.
|
| 360 |
|
| 361 |
"""
|
| 362 |
cluster = self.cfg.GetClusterInfo()
|
| 363 |
os_hvp = {}
|
| 364 |
|
| 365 |
# Filter just for enabled hypervisors
|
| 366 |
for os_name, hv_dict in cluster.os_hvp.items(): |
| 367 |
os_hvp[os_name] = {}
|
| 368 |
for hv_name, hv_params in hv_dict.items(): |
| 369 |
if hv_name in cluster.enabled_hypervisors: |
| 370 |
os_hvp[os_name][hv_name] = hv_params |
| 371 |
|
| 372 |
# Convert ip_family to ip_version
|
| 373 |
primary_ip_version = constants.IP4_VERSION |
| 374 |
if cluster.primary_ip_family == netutils.IP6Address.family:
|
| 375 |
primary_ip_version = constants.IP6_VERSION |
| 376 |
|
| 377 |
result = {
|
| 378 |
"software_version": constants.RELEASE_VERSION,
|
| 379 |
"protocol_version": constants.PROTOCOL_VERSION,
|
| 380 |
"config_version": constants.CONFIG_VERSION,
|
| 381 |
"os_api_version": max(constants.OS_API_VERSIONS), |
| 382 |
"export_version": constants.EXPORT_VERSION,
|
| 383 |
"vcs_version": constants.VCS_VERSION,
|
| 384 |
"architecture": runtime.GetArchInfo(),
|
| 385 |
"name": cluster.cluster_name,
|
| 386 |
"master": self.cfg.GetMasterNodeName(), |
| 387 |
"default_hypervisor": cluster.primary_hypervisor,
|
| 388 |
"enabled_hypervisors": cluster.enabled_hypervisors,
|
| 389 |
"hvparams": dict([(hypervisor_name, cluster.hvparams[hypervisor_name]) |
| 390 |
for hypervisor_name in cluster.enabled_hypervisors]), |
| 391 |
"os_hvp": os_hvp,
|
| 392 |
"beparams": cluster.beparams,
|
| 393 |
"osparams": cluster.osparams,
|
| 394 |
"ipolicy": cluster.ipolicy,
|
| 395 |
"nicparams": cluster.nicparams,
|
| 396 |
"ndparams": cluster.ndparams,
|
| 397 |
"diskparams": cluster.diskparams,
|
| 398 |
"candidate_pool_size": cluster.candidate_pool_size,
|
| 399 |
"max_running_jobs": cluster.max_running_jobs,
|
| 400 |
"master_netdev": cluster.master_netdev,
|
| 401 |
"master_netmask": cluster.master_netmask,
|
| 402 |
"use_external_mip_script": cluster.use_external_mip_script,
|
| 403 |
"volume_group_name": cluster.volume_group_name,
|
| 404 |
"drbd_usermode_helper": cluster.drbd_usermode_helper,
|
| 405 |
"file_storage_dir": cluster.file_storage_dir,
|
| 406 |
"shared_file_storage_dir": cluster.shared_file_storage_dir,
|
| 407 |
"maintain_node_health": cluster.maintain_node_health,
|
| 408 |
"ctime": cluster.ctime,
|
| 409 |
"mtime": cluster.mtime,
|
| 410 |
"uuid": cluster.uuid,
|
| 411 |
"tags": list(cluster.GetTags()), |
| 412 |
"uid_pool": cluster.uid_pool,
|
| 413 |
"default_iallocator": cluster.default_iallocator,
|
| 414 |
"default_iallocator_params": cluster.default_iallocator_params,
|
| 415 |
"reserved_lvs": cluster.reserved_lvs,
|
| 416 |
"primary_ip_version": primary_ip_version,
|
| 417 |
"prealloc_wipe_disks": cluster.prealloc_wipe_disks,
|
| 418 |
"hidden_os": cluster.hidden_os,
|
| 419 |
"blacklisted_os": cluster.blacklisted_os,
|
| 420 |
"enabled_disk_templates": cluster.enabled_disk_templates,
|
| 421 |
"instance_communication_network": cluster.instance_communication_network,
|
| 422 |
} |
| 423 |
|
| 424 |
return result
|
| 425 |
|
| 426 |
|
| 427 |
class LUClusterRedistConf(NoHooksLU): |
| 428 |
"""Force the redistribution of cluster configuration.
|
| 429 |
|
| 430 |
This is a very simple LU.
|
| 431 |
|
| 432 |
"""
|
| 433 |
REQ_BGL = False
|
| 434 |
|
| 435 |
def ExpandNames(self): |
| 436 |
self.needed_locks = {
|
| 437 |
locking.LEVEL_NODE: locking.ALL_SET, |
| 438 |
locking.LEVEL_NODE_ALLOC: locking.ALL_SET, |
| 439 |
} |
| 440 |
self.share_locks = ShareAll()
|
| 441 |
|
| 442 |
def Exec(self, feedback_fn): |
| 443 |
"""Redistribute the configuration.
|
| 444 |
|
| 445 |
"""
|
| 446 |
self.cfg.Update(self.cfg.GetClusterInfo(), feedback_fn) |
| 447 |
RedistributeAncillaryFiles(self)
|
| 448 |
|
| 449 |
|
| 450 |
class LUClusterRename(LogicalUnit): |
| 451 |
"""Rename the cluster.
|
| 452 |
|
| 453 |
"""
|
| 454 |
HPATH = "cluster-rename"
|
| 455 |
HTYPE = constants.HTYPE_CLUSTER |
| 456 |
|
| 457 |
def BuildHooksEnv(self): |
| 458 |
"""Build hooks env.
|
| 459 |
|
| 460 |
"""
|
| 461 |
return {
|
| 462 |
"OP_TARGET": self.cfg.GetClusterName(), |
| 463 |
"NEW_NAME": self.op.name, |
| 464 |
} |
| 465 |
|
| 466 |
def BuildHooksNodes(self): |
| 467 |
"""Build hooks nodes.
|
| 468 |
|
| 469 |
"""
|
| 470 |
return ([self.cfg.GetMasterNode()], self.cfg.GetNodeList()) |
| 471 |
|
| 472 |
def CheckPrereq(self): |
| 473 |
"""Verify that the passed name is a valid one.
|
| 474 |
|
| 475 |
"""
|
| 476 |
hostname = netutils.GetHostname(name=self.op.name,
|
| 477 |
family=self.cfg.GetPrimaryIPFamily())
|
| 478 |
|
| 479 |
new_name = hostname.name |
| 480 |
self.ip = new_ip = hostname.ip
|
| 481 |
old_name = self.cfg.GetClusterName()
|
| 482 |
old_ip = self.cfg.GetMasterIP()
|
| 483 |
if new_name == old_name and new_ip == old_ip: |
| 484 |
raise errors.OpPrereqError("Neither the name nor the IP address of the" |
| 485 |
" cluster has changed",
|
| 486 |
errors.ECODE_INVAL) |
| 487 |
if new_ip != old_ip:
|
| 488 |
if netutils.TcpPing(new_ip, constants.DEFAULT_NODED_PORT):
|
| 489 |
raise errors.OpPrereqError("The given cluster IP address (%s) is" |
| 490 |
" reachable on the network" %
|
| 491 |
new_ip, errors.ECODE_NOTUNIQUE) |
| 492 |
|
| 493 |
self.op.name = new_name
|
| 494 |
|
| 495 |
def Exec(self, feedback_fn): |
| 496 |
"""Rename the cluster.
|
| 497 |
|
| 498 |
"""
|
| 499 |
clustername = self.op.name
|
| 500 |
new_ip = self.ip
|
| 501 |
|
| 502 |
# shutdown the master IP
|
| 503 |
master_params = self.cfg.GetMasterNetworkParameters()
|
| 504 |
ems = self.cfg.GetUseExternalMipScript()
|
| 505 |
result = self.rpc.call_node_deactivate_master_ip(master_params.uuid,
|
| 506 |
master_params, ems) |
| 507 |
result.Raise("Could not disable the master role")
|
| 508 |
|
| 509 |
try:
|
| 510 |
cluster = self.cfg.GetClusterInfo()
|
| 511 |
cluster.cluster_name = clustername |
| 512 |
cluster.master_ip = new_ip |
| 513 |
self.cfg.Update(cluster, feedback_fn)
|
| 514 |
|
| 515 |
# update the known hosts file
|
| 516 |
ssh.WriteKnownHostsFile(self.cfg, pathutils.SSH_KNOWN_HOSTS_FILE)
|
| 517 |
node_list = self.cfg.GetOnlineNodeList()
|
| 518 |
try:
|
| 519 |
node_list.remove(master_params.uuid) |
| 520 |
except ValueError: |
| 521 |
pass
|
| 522 |
UploadHelper(self, node_list, pathutils.SSH_KNOWN_HOSTS_FILE)
|
| 523 |
finally:
|
| 524 |
master_params.ip = new_ip |
| 525 |
result = self.rpc.call_node_activate_master_ip(master_params.uuid,
|
| 526 |
master_params, ems) |
| 527 |
result.Warn("Could not re-enable the master role on the master,"
|
| 528 |
" please restart manually", self.LogWarning) |
| 529 |
|
| 530 |
return clustername
|
| 531 |
|
| 532 |
|
| 533 |
class LUClusterRepairDiskSizes(NoHooksLU): |
| 534 |
"""Verifies the cluster disks sizes.
|
| 535 |
|
| 536 |
"""
|
| 537 |
REQ_BGL = False
|
| 538 |
|
| 539 |
def ExpandNames(self): |
| 540 |
if self.op.instances: |
| 541 |
(_, self.wanted_names) = GetWantedInstances(self, self.op.instances) |
| 542 |
# Not getting the node allocation lock as only a specific set of
|
| 543 |
# instances (and their nodes) is going to be acquired
|
| 544 |
self.needed_locks = {
|
| 545 |
locking.LEVEL_NODE_RES: [], |
| 546 |
locking.LEVEL_INSTANCE: self.wanted_names,
|
| 547 |
} |
| 548 |
self.recalculate_locks[locking.LEVEL_NODE_RES] = constants.LOCKS_REPLACE
|
| 549 |
else:
|
| 550 |
self.wanted_names = None |
| 551 |
self.needed_locks = {
|
| 552 |
locking.LEVEL_NODE_RES: locking.ALL_SET, |
| 553 |
locking.LEVEL_INSTANCE: locking.ALL_SET, |
| 554 |
|
| 555 |
# This opcode is acquires the node locks for all instances
|
| 556 |
locking.LEVEL_NODE_ALLOC: locking.ALL_SET, |
| 557 |
} |
| 558 |
|
| 559 |
self.share_locks = {
|
| 560 |
locking.LEVEL_NODE_RES: 1,
|
| 561 |
locking.LEVEL_INSTANCE: 0,
|
| 562 |
locking.LEVEL_NODE_ALLOC: 1,
|
| 563 |
} |
| 564 |
|
| 565 |
def DeclareLocks(self, level): |
| 566 |
if level == locking.LEVEL_NODE_RES and self.wanted_names is not None: |
| 567 |
self._LockInstancesNodes(primary_only=True, level=level) |
| 568 |
|
| 569 |
def CheckPrereq(self): |
| 570 |
"""Check prerequisites.
|
| 571 |
|
| 572 |
This only checks the optional instance list against the existing names.
|
| 573 |
|
| 574 |
"""
|
| 575 |
if self.wanted_names is None: |
| 576 |
self.wanted_names = self.owned_locks(locking.LEVEL_INSTANCE) |
| 577 |
|
| 578 |
self.wanted_instances = \
|
| 579 |
map(compat.snd, self.cfg.GetMultiInstanceInfoByName(self.wanted_names)) |
| 580 |
|
| 581 |
def _EnsureChildSizes(self, disk): |
| 582 |
"""Ensure children of the disk have the needed disk size.
|
| 583 |
|
| 584 |
This is valid mainly for DRBD8 and fixes an issue where the
|
| 585 |
children have smaller disk size.
|
| 586 |
|
| 587 |
@param disk: an L{ganeti.objects.Disk} object
|
| 588 |
|
| 589 |
"""
|
| 590 |
if disk.dev_type == constants.DT_DRBD8:
|
| 591 |
assert disk.children, "Empty children for DRBD8?" |
| 592 |
fchild = disk.children[0]
|
| 593 |
mismatch = fchild.size < disk.size |
| 594 |
if mismatch:
|
| 595 |
self.LogInfo("Child disk has size %d, parent %d, fixing", |
| 596 |
fchild.size, disk.size) |
| 597 |
fchild.size = disk.size |
| 598 |
|
| 599 |
# and we recurse on this child only, not on the metadev
|
| 600 |
return self._EnsureChildSizes(fchild) or mismatch |
| 601 |
else:
|
| 602 |
return False |
| 603 |
|
| 604 |
def Exec(self, feedback_fn): |
| 605 |
"""Verify the size of cluster disks.
|
| 606 |
|
| 607 |
"""
|
| 608 |
# TODO: check child disks too
|
| 609 |
# TODO: check differences in size between primary/secondary nodes
|
| 610 |
per_node_disks = {}
|
| 611 |
for instance in self.wanted_instances: |
| 612 |
pnode = instance.primary_node |
| 613 |
if pnode not in per_node_disks: |
| 614 |
per_node_disks[pnode] = [] |
| 615 |
for idx, disk in enumerate(instance.disks): |
| 616 |
per_node_disks[pnode].append((instance, idx, disk)) |
| 617 |
|
| 618 |
assert not (frozenset(per_node_disks.keys()) - |
| 619 |
self.owned_locks(locking.LEVEL_NODE_RES)), \
|
| 620 |
"Not owning correct locks"
|
| 621 |
assert not self.owned_locks(locking.LEVEL_NODE) |
| 622 |
|
| 623 |
es_flags = rpc.GetExclusiveStorageForNodes(self.cfg,
|
| 624 |
per_node_disks.keys()) |
| 625 |
|
| 626 |
changed = [] |
| 627 |
for node_uuid, dskl in per_node_disks.items(): |
| 628 |
if not dskl: |
| 629 |
# no disks on the node
|
| 630 |
continue
|
| 631 |
|
| 632 |
newl = [([v[2].Copy()], v[0]) for v in dskl] |
| 633 |
node_name = self.cfg.GetNodeName(node_uuid)
|
| 634 |
result = self.rpc.call_blockdev_getdimensions(node_uuid, newl)
|
| 635 |
if result.fail_msg:
|
| 636 |
self.LogWarning("Failure in blockdev_getdimensions call to node" |
| 637 |
" %s, ignoring", node_name)
|
| 638 |
continue
|
| 639 |
if len(result.payload) != len(dskl): |
| 640 |
logging.warning("Invalid result from node %s: len(dksl)=%d,"
|
| 641 |
" result.payload=%s", node_name, len(dskl), |
| 642 |
result.payload) |
| 643 |
self.LogWarning("Invalid result from node %s, ignoring node results", |
| 644 |
node_name) |
| 645 |
continue
|
| 646 |
for ((instance, idx, disk), dimensions) in zip(dskl, result.payload): |
| 647 |
if dimensions is None: |
| 648 |
self.LogWarning("Disk %d of instance %s did not return size" |
| 649 |
" information, ignoring", idx, instance.name)
|
| 650 |
continue
|
| 651 |
if not isinstance(dimensions, (tuple, list)): |
| 652 |
self.LogWarning("Disk %d of instance %s did not return valid" |
| 653 |
" dimension information, ignoring", idx,
|
| 654 |
instance.name) |
| 655 |
continue
|
| 656 |
(size, spindles) = dimensions |
| 657 |
if not isinstance(size, (int, long)): |
| 658 |
self.LogWarning("Disk %d of instance %s did not return valid" |
| 659 |
" size information, ignoring", idx, instance.name)
|
| 660 |
continue
|
| 661 |
size = size >> 20
|
| 662 |
if size != disk.size:
|
| 663 |
self.LogInfo("Disk %d of instance %s has mismatched size," |
| 664 |
" correcting: recorded %d, actual %d", idx,
|
| 665 |
instance.name, disk.size, size) |
| 666 |
disk.size = size |
| 667 |
self.cfg.Update(instance, feedback_fn)
|
| 668 |
changed.append((instance.name, idx, "size", size))
|
| 669 |
if es_flags[node_uuid]:
|
| 670 |
if spindles is None: |
| 671 |
self.LogWarning("Disk %d of instance %s did not return valid" |
| 672 |
" spindles information, ignoring", idx,
|
| 673 |
instance.name) |
| 674 |
elif disk.spindles is None or disk.spindles != spindles: |
| 675 |
self.LogInfo("Disk %d of instance %s has mismatched spindles," |
| 676 |
" correcting: recorded %s, actual %s",
|
| 677 |
idx, instance.name, disk.spindles, spindles) |
| 678 |
disk.spindles = spindles |
| 679 |
self.cfg.Update(instance, feedback_fn)
|
| 680 |
changed.append((instance.name, idx, "spindles", disk.spindles))
|
| 681 |
if self._EnsureChildSizes(disk): |
| 682 |
self.cfg.Update(instance, feedback_fn)
|
| 683 |
changed.append((instance.name, idx, "size", disk.size))
|
| 684 |
return changed
|
| 685 |
|
| 686 |
|
| 687 |
def _ValidateNetmask(cfg, netmask): |
| 688 |
"""Checks if a netmask is valid.
|
| 689 |
|
| 690 |
@type cfg: L{config.ConfigWriter}
|
| 691 |
@param cfg: cluster configuration
|
| 692 |
@type netmask: int
|
| 693 |
@param netmask: netmask to be verified
|
| 694 |
@raise errors.OpPrereqError: if the validation fails
|
| 695 |
|
| 696 |
"""
|
| 697 |
ip_family = cfg.GetPrimaryIPFamily() |
| 698 |
try:
|
| 699 |
ipcls = netutils.IPAddress.GetClassFromIpFamily(ip_family) |
| 700 |
except errors.ProgrammerError:
|
| 701 |
raise errors.OpPrereqError("Invalid primary ip family: %s." % |
| 702 |
ip_family, errors.ECODE_INVAL) |
| 703 |
if not ipcls.ValidateNetmask(netmask): |
| 704 |
raise errors.OpPrereqError("CIDR netmask (%s) not valid" % |
| 705 |
(netmask), errors.ECODE_INVAL) |
| 706 |
|
| 707 |
|
| 708 |
def CheckFileBasedStoragePathVsEnabledDiskTemplates( |
| 709 |
logging_warn_fn, file_storage_dir, enabled_disk_templates, |
| 710 |
file_disk_template): |
| 711 |
"""Checks whether the given file-based storage directory is acceptable.
|
| 712 |
|
| 713 |
Note: This function is public, because it is also used in bootstrap.py.
|
| 714 |
|
| 715 |
@type logging_warn_fn: function
|
| 716 |
@param logging_warn_fn: function which accepts a string and logs it
|
| 717 |
@type file_storage_dir: string
|
| 718 |
@param file_storage_dir: the directory to be used for file-based instances
|
| 719 |
@type enabled_disk_templates: list of string
|
| 720 |
@param enabled_disk_templates: the list of enabled disk templates
|
| 721 |
@type file_disk_template: string
|
| 722 |
@param file_disk_template: the file-based disk template for which the
|
| 723 |
path should be checked
|
| 724 |
|
| 725 |
"""
|
| 726 |
assert (file_disk_template in utils.storage.GetDiskTemplatesOfStorageTypes( |
| 727 |
constants.ST_FILE, constants.ST_SHARED_FILE |
| 728 |
)) |
| 729 |
file_storage_enabled = file_disk_template in enabled_disk_templates
|
| 730 |
if file_storage_dir is not None: |
| 731 |
if file_storage_dir == "": |
| 732 |
if file_storage_enabled:
|
| 733 |
raise errors.OpPrereqError(
|
| 734 |
"Unsetting the '%s' storage directory while having '%s' storage"
|
| 735 |
" enabled is not permitted." %
|
| 736 |
(file_disk_template, file_disk_template)) |
| 737 |
else:
|
| 738 |
if not file_storage_enabled: |
| 739 |
logging_warn_fn( |
| 740 |
"Specified a %s storage directory, although %s storage is not"
|
| 741 |
" enabled." % (file_disk_template, file_disk_template))
|
| 742 |
else:
|
| 743 |
raise errors.ProgrammerError("Received %s storage dir with value" |
| 744 |
" 'None'." % file_disk_template)
|
| 745 |
|
| 746 |
|
| 747 |
def CheckFileStoragePathVsEnabledDiskTemplates( |
| 748 |
logging_warn_fn, file_storage_dir, enabled_disk_templates): |
| 749 |
"""Checks whether the given file storage directory is acceptable.
|
| 750 |
|
| 751 |
@see: C{CheckFileBasedStoragePathVsEnabledDiskTemplates}
|
| 752 |
|
| 753 |
"""
|
| 754 |
CheckFileBasedStoragePathVsEnabledDiskTemplates( |
| 755 |
logging_warn_fn, file_storage_dir, enabled_disk_templates, |
| 756 |
constants.DT_FILE) |
| 757 |
|
| 758 |
|
| 759 |
def CheckSharedFileStoragePathVsEnabledDiskTemplates( |
| 760 |
logging_warn_fn, file_storage_dir, enabled_disk_templates): |
| 761 |
"""Checks whether the given shared file storage directory is acceptable.
|
| 762 |
|
| 763 |
@see: C{CheckFileBasedStoragePathVsEnabledDiskTemplates}
|
| 764 |
|
| 765 |
"""
|
| 766 |
CheckFileBasedStoragePathVsEnabledDiskTemplates( |
| 767 |
logging_warn_fn, file_storage_dir, enabled_disk_templates, |
| 768 |
constants.DT_SHARED_FILE) |
| 769 |
|
| 770 |
|
| 771 |
class LUClusterSetParams(LogicalUnit): |
| 772 |
"""Change the parameters of the cluster.
|
| 773 |
|
| 774 |
"""
|
| 775 |
HPATH = "cluster-modify"
|
| 776 |
HTYPE = constants.HTYPE_CLUSTER |
| 777 |
REQ_BGL = False
|
| 778 |
|
| 779 |
def CheckArguments(self): |
| 780 |
"""Check parameters
|
| 781 |
|
| 782 |
"""
|
| 783 |
if self.op.uid_pool: |
| 784 |
uidpool.CheckUidPool(self.op.uid_pool)
|
| 785 |
|
| 786 |
if self.op.add_uids: |
| 787 |
uidpool.CheckUidPool(self.op.add_uids)
|
| 788 |
|
| 789 |
if self.op.remove_uids: |
| 790 |
uidpool.CheckUidPool(self.op.remove_uids)
|
| 791 |
|
| 792 |
if self.op.master_netmask is not None: |
| 793 |
_ValidateNetmask(self.cfg, self.op.master_netmask) |
| 794 |
|
| 795 |
if self.op.diskparams: |
| 796 |
for dt_params in self.op.diskparams.values(): |
| 797 |
utils.ForceDictType(dt_params, constants.DISK_DT_TYPES) |
| 798 |
try:
|
| 799 |
utils.VerifyDictOptions(self.op.diskparams, constants.DISK_DT_DEFAULTS)
|
| 800 |
CheckDiskAccessModeValidity(self.op.diskparams)
|
| 801 |
except errors.OpPrereqError, err:
|
| 802 |
raise errors.OpPrereqError("While verify diskparams options: %s" % err, |
| 803 |
errors.ECODE_INVAL) |
| 804 |
|
| 805 |
def ExpandNames(self): |
| 806 |
# FIXME: in the future maybe other cluster params won't require checking on
|
| 807 |
# all nodes to be modified.
|
| 808 |
# FIXME: This opcode changes cluster-wide settings. Is acquiring all
|
| 809 |
# resource locks the right thing, shouldn't it be the BGL instead?
|
| 810 |
self.needed_locks = {
|
| 811 |
locking.LEVEL_NODE: locking.ALL_SET, |
| 812 |
locking.LEVEL_INSTANCE: locking.ALL_SET, |
| 813 |
locking.LEVEL_NODEGROUP: locking.ALL_SET, |
| 814 |
locking.LEVEL_NODE_ALLOC: locking.ALL_SET, |
| 815 |
} |
| 816 |
self.share_locks = ShareAll()
|
| 817 |
|
| 818 |
def BuildHooksEnv(self): |
| 819 |
"""Build hooks env.
|
| 820 |
|
| 821 |
"""
|
| 822 |
return {
|
| 823 |
"OP_TARGET": self.cfg.GetClusterName(), |
| 824 |
"NEW_VG_NAME": self.op.vg_name, |
| 825 |
} |
| 826 |
|
| 827 |
def BuildHooksNodes(self): |
| 828 |
"""Build hooks nodes.
|
| 829 |
|
| 830 |
"""
|
| 831 |
mn = self.cfg.GetMasterNode()
|
| 832 |
return ([mn], [mn])
|
| 833 |
|
| 834 |
def _CheckVgName(self, node_uuids, enabled_disk_templates, |
| 835 |
new_enabled_disk_templates): |
| 836 |
"""Check the consistency of the vg name on all nodes and in case it gets
|
| 837 |
unset whether there are instances still using it.
|
| 838 |
|
| 839 |
"""
|
| 840 |
lvm_is_enabled = utils.IsLvmEnabled(enabled_disk_templates) |
| 841 |
lvm_gets_enabled = utils.LvmGetsEnabled(enabled_disk_templates, |
| 842 |
new_enabled_disk_templates) |
| 843 |
current_vg_name = self.cfg.GetVGName()
|
| 844 |
|
| 845 |
if self.op.vg_name == '': |
| 846 |
if lvm_is_enabled:
|
| 847 |
raise errors.OpPrereqError("Cannot unset volume group if lvm-based" |
| 848 |
" disk templates are or get enabled.")
|
| 849 |
|
| 850 |
if self.op.vg_name is None: |
| 851 |
if current_vg_name is None and lvm_is_enabled: |
| 852 |
raise errors.OpPrereqError("Please specify a volume group when" |
| 853 |
" enabling lvm-based disk-templates.")
|
| 854 |
|
| 855 |
if self.op.vg_name is not None and not self.op.vg_name: |
| 856 |
if self.cfg.HasAnyDiskOfType(constants.DT_PLAIN): |
| 857 |
raise errors.OpPrereqError("Cannot disable lvm storage while lvm-based" |
| 858 |
" instances exist", errors.ECODE_INVAL)
|
| 859 |
|
| 860 |
if (self.op.vg_name is not None and lvm_is_enabled) or \ |
| 861 |
(self.cfg.GetVGName() is not None and lvm_gets_enabled): |
| 862 |
self._CheckVgNameOnNodes(node_uuids)
|
| 863 |
|
| 864 |
def _CheckVgNameOnNodes(self, node_uuids): |
| 865 |
"""Check the status of the volume group on each node.
|
| 866 |
|
| 867 |
"""
|
| 868 |
vglist = self.rpc.call_vg_list(node_uuids)
|
| 869 |
for node_uuid in node_uuids: |
| 870 |
msg = vglist[node_uuid].fail_msg |
| 871 |
if msg:
|
| 872 |
# ignoring down node
|
| 873 |
self.LogWarning("Error while gathering data on node %s" |
| 874 |
" (ignoring node): %s",
|
| 875 |
self.cfg.GetNodeName(node_uuid), msg)
|
| 876 |
continue
|
| 877 |
vgstatus = utils.CheckVolumeGroupSize(vglist[node_uuid].payload, |
| 878 |
self.op.vg_name,
|
| 879 |
constants.MIN_VG_SIZE) |
| 880 |
if vgstatus:
|
| 881 |
raise errors.OpPrereqError("Error on node '%s': %s" % |
| 882 |
(self.cfg.GetNodeName(node_uuid), vgstatus),
|
| 883 |
errors.ECODE_ENVIRON) |
| 884 |
|
| 885 |
@staticmethod
|
| 886 |
def _GetDiskTemplateSetsInner(op_enabled_disk_templates, |
| 887 |
old_enabled_disk_templates): |
| 888 |
"""Computes three sets of disk templates.
|
| 889 |
|
| 890 |
@see: C{_GetDiskTemplateSets} for more details.
|
| 891 |
|
| 892 |
"""
|
| 893 |
enabled_disk_templates = None
|
| 894 |
new_enabled_disk_templates = [] |
| 895 |
disabled_disk_templates = [] |
| 896 |
if op_enabled_disk_templates:
|
| 897 |
enabled_disk_templates = op_enabled_disk_templates |
| 898 |
new_enabled_disk_templates = \ |
| 899 |
list(set(enabled_disk_templates) |
| 900 |
- set(old_enabled_disk_templates))
|
| 901 |
disabled_disk_templates = \ |
| 902 |
list(set(old_enabled_disk_templates) |
| 903 |
- set(enabled_disk_templates))
|
| 904 |
else:
|
| 905 |
enabled_disk_templates = old_enabled_disk_templates |
| 906 |
return (enabled_disk_templates, new_enabled_disk_templates,
|
| 907 |
disabled_disk_templates) |
| 908 |
|
| 909 |
def _GetDiskTemplateSets(self, cluster): |
| 910 |
"""Computes three sets of disk templates.
|
| 911 |
|
| 912 |
The three sets are:
|
| 913 |
- disk templates that will be enabled after this operation (no matter if
|
| 914 |
they were enabled before or not)
|
| 915 |
- disk templates that get enabled by this operation (thus haven't been
|
| 916 |
enabled before.)
|
| 917 |
- disk templates that get disabled by this operation
|
| 918 |
|
| 919 |
"""
|
| 920 |
return self._GetDiskTemplateSetsInner(self.op.enabled_disk_templates, |
| 921 |
cluster.enabled_disk_templates) |
| 922 |
|
| 923 |
def _CheckIpolicy(self, cluster, enabled_disk_templates): |
| 924 |
"""Checks the ipolicy.
|
| 925 |
|
| 926 |
@type cluster: C{objects.Cluster}
|
| 927 |
@param cluster: the cluster's configuration
|
| 928 |
@type enabled_disk_templates: list of string
|
| 929 |
@param enabled_disk_templates: list of (possibly newly) enabled disk
|
| 930 |
templates
|
| 931 |
|
| 932 |
"""
|
| 933 |
# FIXME: write unit tests for this
|
| 934 |
if self.op.ipolicy: |
| 935 |
self.new_ipolicy = GetUpdatedIPolicy(cluster.ipolicy, self.op.ipolicy, |
| 936 |
group_policy=False)
|
| 937 |
|
| 938 |
CheckIpolicyVsDiskTemplates(self.new_ipolicy,
|
| 939 |
enabled_disk_templates) |
| 940 |
|
| 941 |
all_instances = self.cfg.GetAllInstancesInfo().values()
|
| 942 |
violations = set()
|
| 943 |
for group in self.cfg.GetAllNodeGroupsInfo().values(): |
| 944 |
instances = frozenset(
|
| 945 |
[inst for inst in all_instances |
| 946 |
if compat.any(nuuid in group.members |
| 947 |
for nuuid in self.cfg.GetInstanceNodes(inst))]) |
| 948 |
new_ipolicy = objects.FillIPolicy(self.new_ipolicy, group.ipolicy)
|
| 949 |
ipol = masterd.instance.CalculateGroupIPolicy(cluster, group) |
| 950 |
new = ComputeNewInstanceViolations(ipol, new_ipolicy, instances, |
| 951 |
self.cfg)
|
| 952 |
if new:
|
| 953 |
violations.update(new) |
| 954 |
|
| 955 |
if violations:
|
| 956 |
self.LogWarning("After the ipolicy change the following instances" |
| 957 |
" violate them: %s",
|
| 958 |
utils.CommaJoin(utils.NiceSort(violations))) |
| 959 |
else:
|
| 960 |
CheckIpolicyVsDiskTemplates(cluster.ipolicy, |
| 961 |
enabled_disk_templates) |
| 962 |
|
| 963 |
def _CheckDrbdHelperOnNodes(self, drbd_helper, node_uuids): |
| 964 |
"""Checks whether the set DRBD helper actually exists on the nodes.
|
| 965 |
|
| 966 |
@type drbd_helper: string
|
| 967 |
@param drbd_helper: path of the drbd usermode helper binary
|
| 968 |
@type node_uuids: list of strings
|
| 969 |
@param node_uuids: list of node UUIDs to check for the helper
|
| 970 |
|
| 971 |
"""
|
| 972 |
# checks given drbd helper on all nodes
|
| 973 |
helpers = self.rpc.call_drbd_helper(node_uuids)
|
| 974 |
for (_, ninfo) in self.cfg.GetMultiNodeInfo(node_uuids): |
| 975 |
if ninfo.offline:
|
| 976 |
self.LogInfo("Not checking drbd helper on offline node %s", |
| 977 |
ninfo.name) |
| 978 |
continue
|
| 979 |
msg = helpers[ninfo.uuid].fail_msg |
| 980 |
if msg:
|
| 981 |
raise errors.OpPrereqError("Error checking drbd helper on node" |
| 982 |
" '%s': %s" % (ninfo.name, msg),
|
| 983 |
errors.ECODE_ENVIRON) |
| 984 |
node_helper = helpers[ninfo.uuid].payload |
| 985 |
if node_helper != drbd_helper:
|
| 986 |
raise errors.OpPrereqError("Error on node '%s': drbd helper is %s" % |
| 987 |
(ninfo.name, node_helper), |
| 988 |
errors.ECODE_ENVIRON) |
| 989 |
|
| 990 |
def _CheckDrbdHelper(self, node_uuids, drbd_enabled, drbd_gets_enabled): |
| 991 |
"""Check the DRBD usermode helper.
|
| 992 |
|
| 993 |
@type node_uuids: list of strings
|
| 994 |
@param node_uuids: a list of nodes' UUIDs
|
| 995 |
@type drbd_enabled: boolean
|
| 996 |
@param drbd_enabled: whether DRBD will be enabled after this operation
|
| 997 |
(no matter if it was disabled before or not)
|
| 998 |
@type drbd_gets_enabled: boolen
|
| 999 |
@param drbd_gets_enabled: true if DRBD was disabled before this
|
| 1000 |
operation, but will be enabled afterwards
|
| 1001 |
|
| 1002 |
"""
|
| 1003 |
if self.op.drbd_helper == '': |
| 1004 |
if drbd_enabled:
|
| 1005 |
raise errors.OpPrereqError("Cannot disable drbd helper while" |
| 1006 |
" DRBD is enabled.")
|
| 1007 |
if self.cfg.HasAnyDiskOfType(constants.DT_DRBD8): |
| 1008 |
raise errors.OpPrereqError("Cannot disable drbd helper while" |
| 1009 |
" drbd-based instances exist",
|
| 1010 |
errors.ECODE_INVAL) |
| 1011 |
|
| 1012 |
else:
|
| 1013 |
if self.op.drbd_helper is not None and drbd_enabled: |
| 1014 |
self._CheckDrbdHelperOnNodes(self.op.drbd_helper, node_uuids) |
| 1015 |
else:
|
| 1016 |
if drbd_gets_enabled:
|
| 1017 |
current_drbd_helper = self.cfg.GetClusterInfo().drbd_usermode_helper
|
| 1018 |
if current_drbd_helper is not None: |
| 1019 |
self._CheckDrbdHelperOnNodes(current_drbd_helper, node_uuids)
|
| 1020 |
else:
|
| 1021 |
raise errors.OpPrereqError("Cannot enable DRBD without a" |
| 1022 |
" DRBD usermode helper set.")
|
| 1023 |
|
| 1024 |
def _CheckInstancesOfDisabledDiskTemplates( |
| 1025 |
self, disabled_disk_templates):
|
| 1026 |
"""Check whether we try to disable a disk template that is in use.
|
| 1027 |
|
| 1028 |
@type disabled_disk_templates: list of string
|
| 1029 |
@param disabled_disk_templates: list of disk templates that are going to
|
| 1030 |
be disabled by this operation
|
| 1031 |
|
| 1032 |
"""
|
| 1033 |
for disk_template in disabled_disk_templates: |
| 1034 |
if self.cfg.HasAnyDiskOfType(disk_template): |
| 1035 |
raise errors.OpPrereqError(
|
| 1036 |
"Cannot disable disk template '%s', because there is at least one"
|
| 1037 |
" instance using it." % disk_template)
|
| 1038 |
|
| 1039 |
@staticmethod
|
| 1040 |
def _CheckInstanceCommunicationNetwork(network, warning_fn): |
| 1041 |
"""Check whether an existing network is configured for instance
|
| 1042 |
communication.
|
| 1043 |
|
| 1044 |
Checks whether an existing network is configured with the
|
| 1045 |
parameters that are advisable for instance communication, and
|
| 1046 |
otherwise issue security warnings.
|
| 1047 |
|
| 1048 |
@type network: L{ganeti.objects.Network}
|
| 1049 |
@param network: L{ganeti.objects.Network} object whose
|
| 1050 |
configuration is being checked
|
| 1051 |
@type warning_fn: function
|
| 1052 |
@param warning_fn: function used to print warnings
|
| 1053 |
@rtype: None
|
| 1054 |
@return: None
|
| 1055 |
|
| 1056 |
"""
|
| 1057 |
def _MaybeWarn(err, val, default): |
| 1058 |
if val != default:
|
| 1059 |
warning_fn("Supplied instance communication network '%s' %s '%s',"
|
| 1060 |
" this might pose a security risk (default is '%s').",
|
| 1061 |
network.name, err, val, default) |
| 1062 |
|
| 1063 |
if network.network is None: |
| 1064 |
raise errors.OpPrereqError("Supplied instance communication network '%s'" |
| 1065 |
" must have an IPv4 network address.",
|
| 1066 |
network.name) |
| 1067 |
|
| 1068 |
_MaybeWarn("has an IPv4 gateway", network.gateway, None) |
| 1069 |
_MaybeWarn("has a non-standard IPv4 network address", network.network,
|
| 1070 |
constants.INSTANCE_COMMUNICATION_NETWORK4) |
| 1071 |
_MaybeWarn("has an IPv6 gateway", network.gateway6, None) |
| 1072 |
_MaybeWarn("has a non-standard IPv6 network address", network.network6,
|
| 1073 |
constants.INSTANCE_COMMUNICATION_NETWORK6) |
| 1074 |
_MaybeWarn("has a non-standard MAC prefix", network.mac_prefix,
|
| 1075 |
constants.INSTANCE_COMMUNICATION_MAC_PREFIX) |
| 1076 |
|
| 1077 |
def CheckPrereq(self): |
| 1078 |
"""Check prerequisites.
|
| 1079 |
|
| 1080 |
This checks whether the given params don't conflict and
|
| 1081 |
if the given volume group is valid.
|
| 1082 |
|
| 1083 |
"""
|
| 1084 |
node_uuids = self.owned_locks(locking.LEVEL_NODE)
|
| 1085 |
self.cluster = cluster = self.cfg.GetClusterInfo() |
| 1086 |
|
| 1087 |
vm_capable_node_uuids = [node.uuid |
| 1088 |
for node in self.cfg.GetAllNodesInfo().values() |
| 1089 |
if node.uuid in node_uuids and node.vm_capable] |
| 1090 |
|
| 1091 |
(enabled_disk_templates, new_enabled_disk_templates, |
| 1092 |
disabled_disk_templates) = self._GetDiskTemplateSets(cluster)
|
| 1093 |
self._CheckInstancesOfDisabledDiskTemplates(disabled_disk_templates)
|
| 1094 |
|
| 1095 |
self._CheckVgName(vm_capable_node_uuids, enabled_disk_templates,
|
| 1096 |
new_enabled_disk_templates) |
| 1097 |
|
| 1098 |
if self.op.file_storage_dir is not None: |
| 1099 |
CheckFileStoragePathVsEnabledDiskTemplates( |
| 1100 |
self.LogWarning, self.op.file_storage_dir, enabled_disk_templates) |
| 1101 |
|
| 1102 |
if self.op.shared_file_storage_dir is not None: |
| 1103 |
CheckSharedFileStoragePathVsEnabledDiskTemplates( |
| 1104 |
self.LogWarning, self.op.shared_file_storage_dir, |
| 1105 |
enabled_disk_templates) |
| 1106 |
|
| 1107 |
drbd_enabled = constants.DT_DRBD8 in enabled_disk_templates
|
| 1108 |
drbd_gets_enabled = constants.DT_DRBD8 in new_enabled_disk_templates
|
| 1109 |
self._CheckDrbdHelper(node_uuids, drbd_enabled, drbd_gets_enabled)
|
| 1110 |
|
| 1111 |
# validate params changes
|
| 1112 |
if self.op.beparams: |
| 1113 |
objects.UpgradeBeParams(self.op.beparams)
|
| 1114 |
utils.ForceDictType(self.op.beparams, constants.BES_PARAMETER_TYPES)
|
| 1115 |
self.new_beparams = cluster.SimpleFillBE(self.op.beparams) |
| 1116 |
|
| 1117 |
if self.op.ndparams: |
| 1118 |
utils.ForceDictType(self.op.ndparams, constants.NDS_PARAMETER_TYPES)
|
| 1119 |
self.new_ndparams = cluster.SimpleFillND(self.op.ndparams) |
| 1120 |
|
| 1121 |
# TODO: we need a more general way to handle resetting
|
| 1122 |
# cluster-level parameters to default values
|
| 1123 |
if self.new_ndparams["oob_program"] == "": |
| 1124 |
self.new_ndparams["oob_program"] = \ |
| 1125 |
constants.NDC_DEFAULTS[constants.ND_OOB_PROGRAM] |
| 1126 |
|
| 1127 |
if self.op.hv_state: |
| 1128 |
new_hv_state = MergeAndVerifyHvState(self.op.hv_state,
|
| 1129 |
self.cluster.hv_state_static)
|
| 1130 |
self.new_hv_state = dict((hv, cluster.SimpleFillHvState(values)) |
| 1131 |
for hv, values in new_hv_state.items()) |
| 1132 |
|
| 1133 |
if self.op.disk_state: |
| 1134 |
new_disk_state = MergeAndVerifyDiskState(self.op.disk_state,
|
| 1135 |
self.cluster.disk_state_static)
|
| 1136 |
self.new_disk_state = \
|
| 1137 |
dict((storage, dict((name, cluster.SimpleFillDiskState(values)) |
| 1138 |
for name, values in svalues.items())) |
| 1139 |
for storage, svalues in new_disk_state.items()) |
| 1140 |
|
| 1141 |
self._CheckIpolicy(cluster, enabled_disk_templates)
|
| 1142 |
|
| 1143 |
if self.op.nicparams: |
| 1144 |
utils.ForceDictType(self.op.nicparams, constants.NICS_PARAMETER_TYPES)
|
| 1145 |
self.new_nicparams = cluster.SimpleFillNIC(self.op.nicparams) |
| 1146 |
objects.NIC.CheckParameterSyntax(self.new_nicparams)
|
| 1147 |
nic_errors = [] |
| 1148 |
|
| 1149 |
# check all instances for consistency
|
| 1150 |
for instance in self.cfg.GetAllInstancesInfo().values(): |
| 1151 |
for nic_idx, nic in enumerate(instance.nics): |
| 1152 |
params_copy = copy.deepcopy(nic.nicparams) |
| 1153 |
params_filled = objects.FillDict(self.new_nicparams, params_copy)
|
| 1154 |
|
| 1155 |
# check parameter syntax
|
| 1156 |
try:
|
| 1157 |
objects.NIC.CheckParameterSyntax(params_filled) |
| 1158 |
except errors.ConfigurationError, err:
|
| 1159 |
nic_errors.append("Instance %s, nic/%d: %s" %
|
| 1160 |
(instance.name, nic_idx, err)) |
| 1161 |
|
| 1162 |
# if we're moving instances to routed, check that they have an ip
|
| 1163 |
target_mode = params_filled[constants.NIC_MODE] |
| 1164 |
if target_mode == constants.NIC_MODE_ROUTED and not nic.ip: |
| 1165 |
nic_errors.append("Instance %s, nic/%d: routed NIC with no ip"
|
| 1166 |
" address" % (instance.name, nic_idx))
|
| 1167 |
if nic_errors:
|
| 1168 |
raise errors.OpPrereqError("Cannot apply the change, errors:\n%s" % |
| 1169 |
"\n".join(nic_errors), errors.ECODE_INVAL)
|
| 1170 |
|
| 1171 |
# hypervisor list/parameters
|
| 1172 |
self.new_hvparams = new_hvp = objects.FillDict(cluster.hvparams, {})
|
| 1173 |
if self.op.hvparams: |
| 1174 |
for hv_name, hv_dict in self.op.hvparams.items(): |
| 1175 |
if hv_name not in self.new_hvparams: |
| 1176 |
self.new_hvparams[hv_name] = hv_dict
|
| 1177 |
else:
|
| 1178 |
self.new_hvparams[hv_name].update(hv_dict)
|
| 1179 |
|
| 1180 |
# disk template parameters
|
| 1181 |
self.new_diskparams = objects.FillDict(cluster.diskparams, {})
|
| 1182 |
if self.op.diskparams: |
| 1183 |
for dt_name, dt_params in self.op.diskparams.items(): |
| 1184 |
if dt_name not in self.new_diskparams: |
| 1185 |
self.new_diskparams[dt_name] = dt_params
|
| 1186 |
else:
|
| 1187 |
self.new_diskparams[dt_name].update(dt_params)
|
| 1188 |
CheckDiskAccessModeConsistency(self.op.diskparams, self.cfg) |
| 1189 |
|
| 1190 |
# os hypervisor parameters
|
| 1191 |
self.new_os_hvp = objects.FillDict(cluster.os_hvp, {})
|
| 1192 |
if self.op.os_hvp: |
| 1193 |
for os_name, hvs in self.op.os_hvp.items(): |
| 1194 |
if os_name not in self.new_os_hvp: |
| 1195 |
self.new_os_hvp[os_name] = hvs
|
| 1196 |
else:
|
| 1197 |
for hv_name, hv_dict in hvs.items(): |
| 1198 |
if hv_dict is None: |
| 1199 |
# Delete if it exists
|
| 1200 |
self.new_os_hvp[os_name].pop(hv_name, None) |
| 1201 |
elif hv_name not in self.new_os_hvp[os_name]: |
| 1202 |
self.new_os_hvp[os_name][hv_name] = hv_dict
|
| 1203 |
else:
|
| 1204 |
self.new_os_hvp[os_name][hv_name].update(hv_dict)
|
| 1205 |
|
| 1206 |
# os parameters
|
| 1207 |
self._BuildOSParams(cluster)
|
| 1208 |
|
| 1209 |
# changes to the hypervisor list
|
| 1210 |
if self.op.enabled_hypervisors is not None: |
| 1211 |
self.hv_list = self.op.enabled_hypervisors |
| 1212 |
for hv in self.hv_list: |
| 1213 |
# if the hypervisor doesn't already exist in the cluster
|
| 1214 |
# hvparams, we initialize it to empty, and then (in both
|
| 1215 |
# cases) we make sure to fill the defaults, as we might not
|
| 1216 |
# have a complete defaults list if the hypervisor wasn't
|
| 1217 |
# enabled before
|
| 1218 |
if hv not in new_hvp: |
| 1219 |
new_hvp[hv] = {}
|
| 1220 |
new_hvp[hv] = objects.FillDict(constants.HVC_DEFAULTS[hv], new_hvp[hv]) |
| 1221 |
utils.ForceDictType(new_hvp[hv], constants.HVS_PARAMETER_TYPES) |
| 1222 |
else:
|
| 1223 |
self.hv_list = cluster.enabled_hypervisors
|
| 1224 |
|
| 1225 |
if self.op.hvparams or self.op.enabled_hypervisors is not None: |
| 1226 |
# either the enabled list has changed, or the parameters have, validate
|
| 1227 |
for hv_name, hv_params in self.new_hvparams.items(): |
| 1228 |
if ((self.op.hvparams and hv_name in self.op.hvparams) or |
| 1229 |
(self.op.enabled_hypervisors and |
| 1230 |
hv_name in self.op.enabled_hypervisors)): |
| 1231 |
# either this is a new hypervisor, or its parameters have changed
|
| 1232 |
hv_class = hypervisor.GetHypervisorClass(hv_name) |
| 1233 |
utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES) |
| 1234 |
hv_class.CheckParameterSyntax(hv_params) |
| 1235 |
CheckHVParams(self, node_uuids, hv_name, hv_params)
|
| 1236 |
|
| 1237 |
self._CheckDiskTemplateConsistency()
|
| 1238 |
|
| 1239 |
if self.op.os_hvp: |
| 1240 |
# no need to check any newly-enabled hypervisors, since the
|
| 1241 |
# defaults have already been checked in the above code-block
|
| 1242 |
for os_name, os_hvp in self.new_os_hvp.items(): |
| 1243 |
for hv_name, hv_params in os_hvp.items(): |
| 1244 |
utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES) |
| 1245 |
# we need to fill in the new os_hvp on top of the actual hv_p
|
| 1246 |
cluster_defaults = self.new_hvparams.get(hv_name, {})
|
| 1247 |
new_osp = objects.FillDict(cluster_defaults, hv_params) |
| 1248 |
hv_class = hypervisor.GetHypervisorClass(hv_name) |
| 1249 |
hv_class.CheckParameterSyntax(new_osp) |
| 1250 |
CheckHVParams(self, node_uuids, hv_name, new_osp)
|
| 1251 |
|
| 1252 |
if self.op.default_iallocator: |
| 1253 |
alloc_script = utils.FindFile(self.op.default_iallocator,
|
| 1254 |
constants.IALLOCATOR_SEARCH_PATH, |
| 1255 |
os.path.isfile) |
| 1256 |
if alloc_script is None: |
| 1257 |
raise errors.OpPrereqError("Invalid default iallocator script '%s'" |
| 1258 |
" specified" % self.op.default_iallocator, |
| 1259 |
errors.ECODE_INVAL) |
| 1260 |
|
| 1261 |
if self.op.instance_communication_network: |
| 1262 |
network_name = self.op.instance_communication_network
|
| 1263 |
|
| 1264 |
try:
|
| 1265 |
network_uuid = self.cfg.LookupNetwork(network_name)
|
| 1266 |
except errors.OpPrereqError:
|
| 1267 |
network_uuid = None
|
| 1268 |
|
| 1269 |
if network_uuid is not None: |
| 1270 |
network = self.cfg.GetNetwork(network_uuid)
|
| 1271 |
self._CheckInstanceCommunicationNetwork(network, self.LogWarning) |
| 1272 |
|
| 1273 |
def _BuildOSParams(self, cluster): |
| 1274 |
"Calculate the new OS parameters for this operation."
|
| 1275 |
|
| 1276 |
def _GetNewParams(source, new_params): |
| 1277 |
"Wrapper around GetUpdatedParams."
|
| 1278 |
if new_params is None: |
| 1279 |
return source
|
| 1280 |
result = objects.FillDict(source, {}) # deep copy of source
|
| 1281 |
for os_name in new_params: |
| 1282 |
result[os_name] = GetUpdatedParams(result.get(os_name, {}),
|
| 1283 |
new_params[os_name], |
| 1284 |
use_none=True)
|
| 1285 |
if not result[os_name]: |
| 1286 |
del result[os_name] # we removed all parameters |
| 1287 |
return result
|
| 1288 |
|
| 1289 |
self.new_osp = _GetNewParams(cluster.osparams,
|
| 1290 |
self.op.osparams)
|
| 1291 |
self.new_osp_private = _GetNewParams(cluster.osparams_private_cluster,
|
| 1292 |
self.op.osparams_private_cluster)
|
| 1293 |
|
| 1294 |
# Remove os validity check
|
| 1295 |
changed_oses = (set(self.new_osp.keys()) | set(self.new_osp_private.keys())) |
| 1296 |
for os_name in changed_oses: |
| 1297 |
os_params = cluster.SimpleFillOS( |
| 1298 |
os_name, |
| 1299 |
self.new_osp.get(os_name, {}),
|
| 1300 |
os_params_private=self.new_osp_private.get(os_name, {})
|
| 1301 |
) |
| 1302 |
# check the parameter validity (remote check)
|
| 1303 |
CheckOSParams(self, False, [self.cfg.GetMasterNode()], |
| 1304 |
os_name, os_params) |
| 1305 |
|
| 1306 |
def _CheckDiskTemplateConsistency(self): |
| 1307 |
"""Check whether the disk templates that are going to be disabled
|
| 1308 |
are still in use by some instances.
|
| 1309 |
|
| 1310 |
"""
|
| 1311 |
if self.op.enabled_disk_templates: |
| 1312 |
cluster = self.cfg.GetClusterInfo()
|
| 1313 |
instances = self.cfg.GetAllInstancesInfo()
|
| 1314 |
|
| 1315 |
disk_templates_to_remove = set(cluster.enabled_disk_templates) \
|
| 1316 |
- set(self.op.enabled_disk_templates) |
| 1317 |
for instance in instances.itervalues(): |
| 1318 |
if instance.disk_template in disk_templates_to_remove: |
| 1319 |
raise errors.OpPrereqError("Cannot disable disk template '%s'," |
| 1320 |
" because instance '%s' is using it." %
|
| 1321 |
(instance.disk_template, instance.name)) |
| 1322 |
|
| 1323 |
def _SetVgName(self, feedback_fn): |
| 1324 |
"""Determines and sets the new volume group name.
|
| 1325 |
|
| 1326 |
"""
|
| 1327 |
if self.op.vg_name is not None: |
| 1328 |
new_volume = self.op.vg_name
|
| 1329 |
if not new_volume: |
| 1330 |
new_volume = None
|
| 1331 |
if new_volume != self.cfg.GetVGName(): |
| 1332 |
self.cfg.SetVGName(new_volume)
|
| 1333 |
else:
|
| 1334 |
feedback_fn("Cluster LVM configuration already in desired"
|
| 1335 |
" state, not changing")
|
| 1336 |
|
| 1337 |
def _SetFileStorageDir(self, feedback_fn): |
| 1338 |
"""Set the file storage directory.
|
| 1339 |
|
| 1340 |
"""
|
| 1341 |
if self.op.file_storage_dir is not None: |
| 1342 |
if self.cluster.file_storage_dir == self.op.file_storage_dir: |
| 1343 |
feedback_fn("Global file storage dir already set to value '%s'"
|
| 1344 |
% self.cluster.file_storage_dir)
|
| 1345 |
else:
|
| 1346 |
self.cluster.file_storage_dir = self.op.file_storage_dir |
| 1347 |
|
| 1348 |
def _SetDrbdHelper(self, feedback_fn): |
| 1349 |
"""Set the DRBD usermode helper.
|
| 1350 |
|
| 1351 |
"""
|
| 1352 |
if self.op.drbd_helper is not None: |
| 1353 |
if not constants.DT_DRBD8 in self.cluster.enabled_disk_templates: |
| 1354 |
feedback_fn("Note that you specified a drbd user helper, but did not"
|
| 1355 |
" enable the drbd disk template.")
|
| 1356 |
new_helper = self.op.drbd_helper
|
| 1357 |
if not new_helper: |
| 1358 |
new_helper = None
|
| 1359 |
if new_helper != self.cfg.GetDRBDHelper(): |
| 1360 |
self.cfg.SetDRBDHelper(new_helper)
|
| 1361 |
else:
|
| 1362 |
feedback_fn("Cluster DRBD helper already in desired state,"
|
| 1363 |
" not changing")
|
| 1364 |
|
| 1365 |
@staticmethod
|
| 1366 |
def _EnsureInstanceCommunicationNetwork(cfg, network_name): |
| 1367 |
"""Ensure that the instance communication network exists and is
|
| 1368 |
connected to all groups.
|
| 1369 |
|
| 1370 |
The instance communication network given by L{network_name} it is
|
| 1371 |
created, if necessary, via the opcode 'OpNetworkAdd'. Also, the
|
| 1372 |
instance communication network is connected to all existing node
|
| 1373 |
groups, if necessary, via the opcode 'OpNetworkConnect'.
|
| 1374 |
|
| 1375 |
@type cfg: L{config.ConfigWriter}
|
| 1376 |
@param cfg: cluster configuration
|
| 1377 |
|
| 1378 |
@type network_name: string
|
| 1379 |
@param network_name: instance communication network name
|
| 1380 |
|
| 1381 |
@rtype: L{ganeti.cmdlib.ResultWithJobs} or L{None}
|
| 1382 |
@return: L{ganeti.cmdlib.ResultWithJobs} if the instance
|
| 1383 |
communication needs to be created or it needs to be
|
| 1384 |
connected to a group, otherwise L{None}
|
| 1385 |
|
| 1386 |
"""
|
| 1387 |
jobs = [] |
| 1388 |
|
| 1389 |
try:
|
| 1390 |
network_uuid = cfg.LookupNetwork(network_name) |
| 1391 |
network_exists = True
|
| 1392 |
except errors.OpPrereqError:
|
| 1393 |
network_exists = False
|
| 1394 |
|
| 1395 |
if not network_exists: |
| 1396 |
jobs.append(AddInstanceCommunicationNetworkOp(network_name)) |
| 1397 |
|
| 1398 |
for group_uuid in cfg.GetNodeGroupList(): |
| 1399 |
group = cfg.GetNodeGroup(group_uuid) |
| 1400 |
|
| 1401 |
if network_exists:
|
| 1402 |
network_connected = network_uuid in group.networks
|
| 1403 |
else:
|
| 1404 |
# The network was created asynchronously by the previous
|
| 1405 |
# opcode and, therefore, we don't have access to its
|
| 1406 |
# network_uuid. As a result, we assume that the network is
|
| 1407 |
# not connected to any group yet.
|
| 1408 |
network_connected = False
|
| 1409 |
|
| 1410 |
if not network_connected: |
| 1411 |
op = ConnectInstanceCommunicationNetworkOp(group_uuid, network_name) |
| 1412 |
jobs.append(op) |
| 1413 |
|
| 1414 |
if jobs:
|
| 1415 |
return ResultWithJobs([jobs])
|
| 1416 |
else:
|
| 1417 |
return None |
| 1418 |
|
| 1419 |
@staticmethod
|
| 1420 |
def _ModifyInstanceCommunicationNetwork(cfg, cluster, network_name, |
| 1421 |
feedback_fn): |
| 1422 |
"""Update the instance communication network stored in the cluster
|
| 1423 |
configuration.
|
| 1424 |
|
| 1425 |
Compares the user-supplied instance communication network against
|
| 1426 |
the one stored in the Ganeti cluster configuration. If there is a
|
| 1427 |
change, the instance communication network may be possibly created
|
| 1428 |
and connected to all groups (see
|
| 1429 |
L{LUClusterSetParams._EnsureInstanceCommunicationNetwork}).
|
| 1430 |
|
| 1431 |
@type cfg: L{config.ConfigWriter}
|
| 1432 |
@param cfg: cluster configuration
|
| 1433 |
|
| 1434 |
@type cluster: L{ganeti.objects.Cluster}
|
| 1435 |
@param cluster: Ganeti cluster
|
| 1436 |
|
| 1437 |
@type network_name: string
|
| 1438 |
@param network_name: instance communication network name
|
| 1439 |
|
| 1440 |
@type feedback_fn: function
|
| 1441 |
@param feedback_fn: see L{ganeti.cmdlist.base.LogicalUnit}
|
| 1442 |
|
| 1443 |
@rtype: L{LUClusterSetParams._EnsureInstanceCommunicationNetwork} or L{None}
|
| 1444 |
@return: see L{LUClusterSetParams._EnsureInstanceCommunicationNetwork}
|
| 1445 |
|
| 1446 |
"""
|
| 1447 |
config_network_name = cfg.GetInstanceCommunicationNetwork() |
| 1448 |
|
| 1449 |
if network_name == config_network_name:
|
| 1450 |
feedback_fn("Instance communication network already is '%s', nothing to"
|
| 1451 |
" do." % network_name)
|
| 1452 |
else:
|
| 1453 |
try:
|
| 1454 |
cfg.LookupNetwork(config_network_name) |
| 1455 |
feedback_fn("Previous instance communication network '%s'"
|
| 1456 |
" should be removed manually." % config_network_name)
|
| 1457 |
except errors.OpPrereqError:
|
| 1458 |
pass
|
| 1459 |
|
| 1460 |
if network_name:
|
| 1461 |
feedback_fn("Changing instance communication network to '%s', only new"
|
| 1462 |
" instances will be affected."
|
| 1463 |
% network_name) |
| 1464 |
else:
|
| 1465 |
feedback_fn("Disabling instance communication network, only new"
|
| 1466 |
" instances will be affected.")
|
| 1467 |
|
| 1468 |
cluster.instance_communication_network = network_name |
| 1469 |
|
| 1470 |
if network_name:
|
| 1471 |
return LUClusterSetParams._EnsureInstanceCommunicationNetwork(
|
| 1472 |
cfg, |
| 1473 |
network_name) |
| 1474 |
else:
|
| 1475 |
return None |
| 1476 |
|
| 1477 |
def Exec(self, feedback_fn): |
| 1478 |
"""Change the parameters of the cluster.
|
| 1479 |
|
| 1480 |
"""
|
| 1481 |
if self.op.enabled_disk_templates: |
| 1482 |
self.cluster.enabled_disk_templates = \
|
| 1483 |
list(self.op.enabled_disk_templates) |
| 1484 |
|
| 1485 |
self._SetVgName(feedback_fn)
|
| 1486 |
self._SetFileStorageDir(feedback_fn)
|
| 1487 |
self._SetDrbdHelper(feedback_fn)
|
| 1488 |
|
| 1489 |
if self.op.hvparams: |
| 1490 |
self.cluster.hvparams = self.new_hvparams |
| 1491 |
if self.op.os_hvp: |
| 1492 |
self.cluster.os_hvp = self.new_os_hvp |
| 1493 |
if self.op.enabled_hypervisors is not None: |
| 1494 |
self.cluster.hvparams = self.new_hvparams |
| 1495 |
self.cluster.enabled_hypervisors = self.op.enabled_hypervisors |
| 1496 |
if self.op.beparams: |
| 1497 |
self.cluster.beparams[constants.PP_DEFAULT] = self.new_beparams |
| 1498 |
if self.op.nicparams: |
| 1499 |
self.cluster.nicparams[constants.PP_DEFAULT] = self.new_nicparams |
| 1500 |
if self.op.ipolicy: |
| 1501 |
self.cluster.ipolicy = self.new_ipolicy |
| 1502 |
if self.op.osparams: |
| 1503 |
self.cluster.osparams = self.new_osp |
| 1504 |
if self.op.osparams_private_cluster: |
| 1505 |
self.cluster.osparams_private_cluster = self.new_osp_private |
| 1506 |
if self.op.ndparams: |
| 1507 |
self.cluster.ndparams = self.new_ndparams |
| 1508 |
if self.op.diskparams: |
| 1509 |
self.cluster.diskparams = self.new_diskparams |
| 1510 |
if self.op.hv_state: |
| 1511 |
self.cluster.hv_state_static = self.new_hv_state |
| 1512 |
if self.op.disk_state: |
| 1513 |
self.cluster.disk_state_static = self.new_disk_state |
| 1514 |
|
| 1515 |
if self.op.candidate_pool_size is not None: |
| 1516 |
self.cluster.candidate_pool_size = self.op.candidate_pool_size |
| 1517 |
# we need to update the pool size here, otherwise the save will fail
|
| 1518 |
AdjustCandidatePool(self, [], feedback_fn)
|
| 1519 |
|
| 1520 |
if self.op.max_running_jobs is not None: |
| 1521 |
self.cluster.max_running_jobs = self.op.max_running_jobs |
| 1522 |
|
| 1523 |
if self.op.maintain_node_health is not None: |
| 1524 |
if self.op.maintain_node_health and not constants.ENABLE_CONFD: |
| 1525 |
feedback_fn("Note: CONFD was disabled at build time, node health"
|
| 1526 |
" maintenance is not useful (still enabling it)")
|
| 1527 |
self.cluster.maintain_node_health = self.op.maintain_node_health |
| 1528 |
|
| 1529 |
if self.op.modify_etc_hosts is not None: |
| 1530 |
self.cluster.modify_etc_hosts = self.op.modify_etc_hosts |
| 1531 |
|
| 1532 |
if self.op.prealloc_wipe_disks is not None: |
| 1533 |
self.cluster.prealloc_wipe_disks = self.op.prealloc_wipe_disks |
| 1534 |
|
| 1535 |
if self.op.add_uids is not None: |
| 1536 |
uidpool.AddToUidPool(self.cluster.uid_pool, self.op.add_uids) |
| 1537 |
|
| 1538 |
if self.op.remove_uids is not None: |
| 1539 |
uidpool.RemoveFromUidPool(self.cluster.uid_pool, self.op.remove_uids) |
| 1540 |
|
| 1541 |
if self.op.uid_pool is not None: |
| 1542 |
self.cluster.uid_pool = self.op.uid_pool |
| 1543 |
|
| 1544 |
if self.op.default_iallocator is not None: |
| 1545 |
self.cluster.default_iallocator = self.op.default_iallocator |
| 1546 |
|
| 1547 |
if self.op.default_iallocator_params is not None: |
| 1548 |
self.cluster.default_iallocator_params = self.op.default_iallocator_params |
| 1549 |
|
| 1550 |
if self.op.reserved_lvs is not None: |
| 1551 |
self.cluster.reserved_lvs = self.op.reserved_lvs |
| 1552 |
|
| 1553 |
if self.op.use_external_mip_script is not None: |
| 1554 |
self.cluster.use_external_mip_script = self.op.use_external_mip_script |
| 1555 |
|
| 1556 |
def helper_os(aname, mods, desc): |
| 1557 |
desc += " OS list"
|
| 1558 |
lst = getattr(self.cluster, aname) |
| 1559 |
for key, val in mods: |
| 1560 |
if key == constants.DDM_ADD:
|
| 1561 |
if val in lst: |
| 1562 |
feedback_fn("OS %s already in %s, ignoring" % (val, desc))
|
| 1563 |
else:
|
| 1564 |
lst.append(val) |
| 1565 |
elif key == constants.DDM_REMOVE:
|
| 1566 |
if val in lst: |
| 1567 |
lst.remove(val) |
| 1568 |
else:
|
| 1569 |
feedback_fn("OS %s not found in %s, ignoring" % (val, desc))
|
| 1570 |
else:
|
| 1571 |
raise errors.ProgrammerError("Invalid modification '%s'" % key) |
| 1572 |
|
| 1573 |
if self.op.hidden_os: |
| 1574 |
helper_os("hidden_os", self.op.hidden_os, "hidden") |
| 1575 |
|
| 1576 |
if self.op.blacklisted_os: |
| 1577 |
helper_os("blacklisted_os", self.op.blacklisted_os, "blacklisted") |
| 1578 |
|
| 1579 |
if self.op.master_netdev: |
| 1580 |
master_params = self.cfg.GetMasterNetworkParameters()
|
| 1581 |
ems = self.cfg.GetUseExternalMipScript()
|
| 1582 |
feedback_fn("Shutting down master ip on the current netdev (%s)" %
|
| 1583 |
self.cluster.master_netdev)
|
| 1584 |
result = self.rpc.call_node_deactivate_master_ip(master_params.uuid,
|
| 1585 |
master_params, ems) |
| 1586 |
if not self.op.force: |
| 1587 |
result.Raise("Could not disable the master ip")
|
| 1588 |
else:
|
| 1589 |
if result.fail_msg:
|
| 1590 |
msg = ("Could not disable the master ip (continuing anyway): %s" %
|
| 1591 |
result.fail_msg) |
| 1592 |
feedback_fn(msg) |
| 1593 |
feedback_fn("Changing master_netdev from %s to %s" %
|
| 1594 |
(master_params.netdev, self.op.master_netdev))
|
| 1595 |
self.cluster.master_netdev = self.op.master_netdev |
| 1596 |
|
| 1597 |
if self.op.master_netmask: |
| 1598 |
master_params = self.cfg.GetMasterNetworkParameters()
|
| 1599 |
feedback_fn("Changing master IP netmask to %s" % self.op.master_netmask) |
| 1600 |
result = self.rpc.call_node_change_master_netmask(
|
| 1601 |
master_params.uuid, master_params.netmask, |
| 1602 |
self.op.master_netmask, master_params.ip,
|
| 1603 |
master_params.netdev) |
| 1604 |
result.Warn("Could not change the master IP netmask", feedback_fn)
|
| 1605 |
self.cluster.master_netmask = self.op.master_netmask |
| 1606 |
|
| 1607 |
self.cfg.Update(self.cluster, feedback_fn) |
| 1608 |
|
| 1609 |
if self.op.master_netdev: |
| 1610 |
master_params = self.cfg.GetMasterNetworkParameters()
|
| 1611 |
feedback_fn("Starting the master ip on the new master netdev (%s)" %
|
| 1612 |
self.op.master_netdev)
|
| 1613 |
ems = self.cfg.GetUseExternalMipScript()
|
| 1614 |
result = self.rpc.call_node_activate_master_ip(master_params.uuid,
|
| 1615 |
master_params, ems) |
| 1616 |
result.Warn("Could not re-enable the master ip on the master,"
|
| 1617 |
" please restart manually", self.LogWarning) |
| 1618 |
|
| 1619 |
network_name = self.op.instance_communication_network
|
| 1620 |
if network_name is not None: |
| 1621 |
return self._ModifyInstanceCommunicationNetwork(self.cfg, self.cluster, |
| 1622 |
network_name, feedback_fn) |
| 1623 |
else:
|
| 1624 |
return None |
| 1625 |
|
| 1626 |
|
| 1627 |
class LUClusterVerify(NoHooksLU): |
| 1628 |
"""Submits all jobs necessary to verify the cluster.
|
| 1629 |
|
| 1630 |
"""
|
| 1631 |
REQ_BGL = False
|
| 1632 |
|
| 1633 |
def ExpandNames(self): |
| 1634 |
self.needed_locks = {}
|
| 1635 |
|
| 1636 |
def Exec(self, feedback_fn): |
| 1637 |
jobs = [] |
| 1638 |
|
| 1639 |
if self.op.group_name: |
| 1640 |
groups = [self.op.group_name]
|
| 1641 |
depends_fn = lambda: None |
| 1642 |
else:
|
| 1643 |
groups = self.cfg.GetNodeGroupList()
|
| 1644 |
|
| 1645 |
# Verify global configuration
|
| 1646 |
jobs.append([ |
| 1647 |
opcodes.OpClusterVerifyConfig(ignore_errors=self.op.ignore_errors),
|
| 1648 |
]) |
| 1649 |
|
| 1650 |
# Always depend on global verification
|
| 1651 |
depends_fn = lambda: [(-len(jobs), [])] |
| 1652 |
|
| 1653 |
jobs.extend( |
| 1654 |
[opcodes.OpClusterVerifyGroup(group_name=group, |
| 1655 |
ignore_errors=self.op.ignore_errors,
|
| 1656 |
depends=depends_fn())] |
| 1657 |
for group in groups) |
| 1658 |
|
| 1659 |
# Fix up all parameters
|
| 1660 |
for op in itertools.chain(*jobs): # pylint: disable=W0142 |
| 1661 |
op.debug_simulate_errors = self.op.debug_simulate_errors
|
| 1662 |
op.verbose = self.op.verbose
|
| 1663 |
op.error_codes = self.op.error_codes
|
| 1664 |
try:
|
| 1665 |
op.skip_checks = self.op.skip_checks
|
| 1666 |
except AttributeError: |
| 1667 |
assert not isinstance(op, opcodes.OpClusterVerifyGroup) |
| 1668 |
|
| 1669 |
return ResultWithJobs(jobs)
|
| 1670 |
|
| 1671 |
|
| 1672 |
class _VerifyErrors(object): |
| 1673 |
"""Mix-in for cluster/group verify LUs.
|
| 1674 |
|
| 1675 |
It provides _Error and _ErrorIf, and updates the self.bad boolean. (Expects
|
| 1676 |
self.op and self._feedback_fn to be available.)
|
| 1677 |
|
| 1678 |
"""
|
| 1679 |
|
| 1680 |
ETYPE_FIELD = "code"
|
| 1681 |
ETYPE_ERROR = constants.CV_ERROR |
| 1682 |
ETYPE_WARNING = constants.CV_WARNING |
| 1683 |
|
| 1684 |
def _Error(self, ecode, item, msg, *args, **kwargs): |
| 1685 |
"""Format an error message.
|
| 1686 |
|
| 1687 |
Based on the opcode's error_codes parameter, either format a
|
| 1688 |
parseable error code, or a simpler error string.
|
| 1689 |
|
| 1690 |
This must be called only from Exec and functions called from Exec.
|
| 1691 |
|
| 1692 |
"""
|
| 1693 |
ltype = kwargs.get(self.ETYPE_FIELD, self.ETYPE_ERROR) |
| 1694 |
itype, etxt, _ = ecode |
| 1695 |
# If the error code is in the list of ignored errors, demote the error to a
|
| 1696 |
# warning
|
| 1697 |
if etxt in self.op.ignore_errors: # pylint: disable=E1101 |
| 1698 |
ltype = self.ETYPE_WARNING
|
| 1699 |
# first complete the msg
|
| 1700 |
if args:
|
| 1701 |
msg = msg % args |
| 1702 |
# then format the whole message
|
| 1703 |
if self.op.error_codes: # This is a mix-in. pylint: disable=E1101 |
| 1704 |
msg = "%s:%s:%s:%s:%s" % (ltype, etxt, itype, item, msg)
|
| 1705 |
else:
|
| 1706 |
if item:
|
| 1707 |
item = " " + item
|
| 1708 |
else:
|
| 1709 |
item = ""
|
| 1710 |
msg = "%s: %s%s: %s" % (ltype, itype, item, msg)
|
| 1711 |
# and finally report it via the feedback_fn
|
| 1712 |
self._feedback_fn(" - %s" % msg) # Mix-in. pylint: disable=E1101 |
| 1713 |
# do not mark the operation as failed for WARN cases only
|
| 1714 |
if ltype == self.ETYPE_ERROR: |
| 1715 |
self.bad = True |
| 1716 |
|
| 1717 |
def _ErrorIf(self, cond, *args, **kwargs): |
| 1718 |
"""Log an error message if the passed condition is True.
|
| 1719 |
|
| 1720 |
"""
|
| 1721 |
if (bool(cond) |
| 1722 |
or self.op.debug_simulate_errors): # pylint: disable=E1101 |
| 1723 |
self._Error(*args, **kwargs)
|
| 1724 |
|
| 1725 |
|
| 1726 |
def _GetAllHypervisorParameters(cluster, instances): |
| 1727 |
"""Compute the set of all hypervisor parameters.
|
| 1728 |
|
| 1729 |
@type cluster: L{objects.Cluster}
|
| 1730 |
@param cluster: the cluster object
|
| 1731 |
@param instances: list of L{objects.Instance}
|
| 1732 |
@param instances: additional instances from which to obtain parameters
|
| 1733 |
@rtype: list of (origin, hypervisor, parameters)
|
| 1734 |
@return: a list with all parameters found, indicating the hypervisor they
|
| 1735 |
apply to, and the origin (can be "cluster", "os X", or "instance Y")
|
| 1736 |
|
| 1737 |
"""
|
| 1738 |
hvp_data = [] |
| 1739 |
|
| 1740 |
for hv_name in cluster.enabled_hypervisors: |
| 1741 |
hvp_data.append(("cluster", hv_name, cluster.GetHVDefaults(hv_name)))
|
| 1742 |
|
| 1743 |
for os_name, os_hvp in cluster.os_hvp.items(): |
| 1744 |
for hv_name, hv_params in os_hvp.items(): |
| 1745 |
if hv_params:
|
| 1746 |
full_params = cluster.GetHVDefaults(hv_name, os_name=os_name) |
| 1747 |
hvp_data.append(("os %s" % os_name, hv_name, full_params))
|
| 1748 |
|
| 1749 |
# TODO: collapse identical parameter values in a single one
|
| 1750 |
for instance in instances: |
| 1751 |
if instance.hvparams:
|
| 1752 |
hvp_data.append(("instance %s" % instance.name, instance.hypervisor,
|
| 1753 |
cluster.FillHV(instance))) |
| 1754 |
|
| 1755 |
return hvp_data
|
| 1756 |
|
| 1757 |
|
| 1758 |
class LUClusterVerifyConfig(NoHooksLU, _VerifyErrors): |
| 1759 |
"""Verifies the cluster config.
|
| 1760 |
|
| 1761 |
"""
|
| 1762 |
REQ_BGL = False
|
| 1763 |
|
| 1764 |
def _VerifyHVP(self, hvp_data): |
| 1765 |
"""Verifies locally the syntax of the hypervisor parameters.
|
| 1766 |
|
| 1767 |
"""
|
| 1768 |
for item, hv_name, hv_params in hvp_data: |
| 1769 |
msg = ("hypervisor %s parameters syntax check (source %s): %%s" %
|
| 1770 |
(item, hv_name)) |
| 1771 |
try:
|
| 1772 |
hv_class = hypervisor.GetHypervisorClass(hv_name) |
| 1773 |
utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES) |
| 1774 |
hv_class.CheckParameterSyntax(hv_params) |
| 1775 |
except errors.GenericError, err:
|
| 1776 |
self._ErrorIf(True, constants.CV_ECLUSTERCFG, None, msg % str(err)) |
| 1777 |
|
| 1778 |
def ExpandNames(self): |
| 1779 |
self.needed_locks = dict.fromkeys(locking.LEVELS, locking.ALL_SET) |
| 1780 |
self.share_locks = ShareAll()
|
| 1781 |
|
| 1782 |
def CheckPrereq(self): |
| 1783 |
"""Check prerequisites.
|
| 1784 |
|
| 1785 |
"""
|
| 1786 |
# Retrieve all information
|
| 1787 |
self.all_group_info = self.cfg.GetAllNodeGroupsInfo() |
| 1788 |
self.all_node_info = self.cfg.GetAllNodesInfo() |
| 1789 |
self.all_inst_info = self.cfg.GetAllInstancesInfo() |
| 1790 |
|
| 1791 |
def Exec(self, feedback_fn): |
| 1792 |
"""Verify integrity of cluster, performing various test on nodes.
|
| 1793 |
|
| 1794 |
"""
|
| 1795 |
self.bad = False |
| 1796 |
self._feedback_fn = feedback_fn
|
| 1797 |
|
| 1798 |
feedback_fn("* Verifying cluster config")
|
| 1799 |
|
| 1800 |
for msg in self.cfg.VerifyConfig(): |
| 1801 |
self._ErrorIf(True, constants.CV_ECLUSTERCFG, None, msg) |
| 1802 |
|
| 1803 |
feedback_fn("* Verifying cluster certificate files")
|
| 1804 |
|
| 1805 |
for cert_filename in pathutils.ALL_CERT_FILES: |
| 1806 |
(errcode, msg) = utils.VerifyCertificate(cert_filename) |
| 1807 |
self._ErrorIf(errcode, constants.CV_ECLUSTERCERT, None, msg, code=errcode) |
| 1808 |
|
| 1809 |
self._ErrorIf(not utils.CanRead(constants.LUXID_USER, |
| 1810 |
pathutils.NODED_CERT_FILE), |
| 1811 |
constants.CV_ECLUSTERCERT, |
| 1812 |
None,
|
| 1813 |
pathutils.NODED_CERT_FILE + " must be accessible by the " +
|
| 1814 |
constants.LUXID_USER + " user")
|
| 1815 |
|
| 1816 |
feedback_fn("* Verifying hypervisor parameters")
|
| 1817 |
|
| 1818 |
self._VerifyHVP(_GetAllHypervisorParameters(self.cfg.GetClusterInfo(), |
| 1819 |
self.all_inst_info.values()))
|
| 1820 |
|
| 1821 |
feedback_fn("* Verifying all nodes belong to an existing group")
|
| 1822 |
|
| 1823 |
# We do this verification here because, should this bogus circumstance
|
| 1824 |
# occur, it would never be caught by VerifyGroup, which only acts on
|
| 1825 |
# nodes/instances reachable from existing node groups.
|
| 1826 |
|
| 1827 |
dangling_nodes = set(node for node in self.all_node_info.values() |
| 1828 |
if node.group not in self.all_group_info) |
| 1829 |
|
| 1830 |
dangling_instances = {}
|
| 1831 |
no_node_instances = [] |
| 1832 |
|
| 1833 |
for inst in self.all_inst_info.values(): |
| 1834 |
if inst.primary_node in [node.uuid for node in dangling_nodes]: |
| 1835 |
dangling_instances.setdefault(inst.primary_node, []).append(inst) |
| 1836 |
elif inst.primary_node not in self.all_node_info: |
| 1837 |
no_node_instances.append(inst) |
| 1838 |
|
| 1839 |
pretty_dangling = [ |
| 1840 |
"%s (%s)" %
|
| 1841 |
(node.name, |
| 1842 |
utils.CommaJoin(inst.name for
|
| 1843 |
inst in dangling_instances.get(node.uuid, [])))
|
| 1844 |
for node in dangling_nodes] |
| 1845 |
|
| 1846 |
self._ErrorIf(bool(dangling_nodes), constants.CV_ECLUSTERDANGLINGNODES, |
| 1847 |
None,
|
| 1848 |
"the following nodes (and their instances) belong to a non"
|
| 1849 |
" existing group: %s", utils.CommaJoin(pretty_dangling))
|
| 1850 |
|
| 1851 |
self._ErrorIf(bool(no_node_instances), constants.CV_ECLUSTERDANGLINGINST, |
| 1852 |
None,
|
| 1853 |
"the following instances have a non-existing primary-node:"
|
| 1854 |
" %s", utils.CommaJoin(inst.name for |
| 1855 |
inst in no_node_instances))
|
| 1856 |
|
| 1857 |
return not self.bad |
| 1858 |
|
| 1859 |
|
| 1860 |
class LUClusterVerifyGroup(LogicalUnit, _VerifyErrors): |
| 1861 |
"""Verifies the status of a node group.
|
| 1862 |
|
| 1863 |
"""
|
| 1864 |
HPATH = "cluster-verify"
|
| 1865 |
HTYPE = constants.HTYPE_CLUSTER |
| 1866 |
REQ_BGL = False
|
| 1867 |
|
| 1868 |
_HOOKS_INDENT_RE = re.compile("^", re.M)
|
| 1869 |
|
| 1870 |
class NodeImage(object): |
| 1871 |
"""A class representing the logical and physical status of a node.
|
| 1872 |
|
| 1873 |
@type uuid: string
|
| 1874 |
@ivar uuid: the node UUID to which this object refers
|
| 1875 |
@ivar volumes: a structure as returned from
|
| 1876 |
L{ganeti.backend.GetVolumeList} (runtime)
|
| 1877 |
@ivar instances: a list of running instances (runtime)
|
| 1878 |
@ivar pinst: list of configured primary instances (config)
|
| 1879 |
@ivar sinst: list of configured secondary instances (config)
|
| 1880 |
@ivar sbp: dictionary of {primary-node: list of instances} for all
|
| 1881 |
instances for which this node is secondary (config)
|
| 1882 |
@ivar mfree: free memory, as reported by hypervisor (runtime)
|
| 1883 |
@ivar dfree: free disk, as reported by the node (runtime)
|
| 1884 |
@ivar offline: the offline status (config)
|
| 1885 |
@type rpc_fail: boolean
|
| 1886 |
@ivar rpc_fail: whether the RPC verify call was successfull (overall,
|
| 1887 |
not whether the individual keys were correct) (runtime)
|
| 1888 |
@type lvm_fail: boolean
|
| 1889 |
@ivar lvm_fail: whether the RPC call didn't return valid LVM data
|
| 1890 |
@type hyp_fail: boolean
|
| 1891 |
@ivar hyp_fail: whether the RPC call didn't return the instance list
|
| 1892 |
@type ghost: boolean
|
| 1893 |
@ivar ghost: whether this is a known node or not (config)
|
| 1894 |
@type os_fail: boolean
|
| 1895 |
@ivar os_fail: whether the RPC call didn't return valid OS data
|
| 1896 |
@type oslist: list
|
| 1897 |
@ivar oslist: list of OSes as diagnosed by DiagnoseOS
|
| 1898 |
@type vm_capable: boolean
|
| 1899 |
@ivar vm_capable: whether the node can host instances
|
| 1900 |
@type pv_min: float
|
| 1901 |
@ivar pv_min: size in MiB of the smallest PVs
|
| 1902 |
@type pv_max: float
|
| 1903 |
@ivar pv_max: size in MiB of the biggest PVs
|
| 1904 |
|
| 1905 |
"""
|
| 1906 |
def __init__(self, offline=False, uuid=None, vm_capable=True): |
| 1907 |
self.uuid = uuid
|
| 1908 |
self.volumes = {}
|
| 1909 |
self.instances = []
|
| 1910 |
self.pinst = []
|
| 1911 |
self.sinst = []
|
| 1912 |
self.sbp = {}
|
| 1913 |
self.mfree = 0 |
| 1914 |
self.dfree = 0 |
| 1915 |
self.offline = offline
|
| 1916 |
self.vm_capable = vm_capable
|
| 1917 |
self.rpc_fail = False |
| 1918 |
self.lvm_fail = False |
| 1919 |
self.hyp_fail = False |
| 1920 |
self.ghost = False |
| 1921 |
self.os_fail = False |
| 1922 |
self.oslist = {}
|
| 1923 |
self.pv_min = None |
| 1924 |
self.pv_max = None |
| 1925 |
|
| 1926 |
def ExpandNames(self): |
| 1927 |
# This raises errors.OpPrereqError on its own:
|
| 1928 |
self.group_uuid = self.cfg.LookupNodeGroup(self.op.group_name) |
| 1929 |
|
| 1930 |
# Get instances in node group; this is unsafe and needs verification later
|
| 1931 |
inst_uuids = \ |
| 1932 |
self.cfg.GetNodeGroupInstances(self.group_uuid, primary_only=True) |
| 1933 |
|
| 1934 |
self.needed_locks = {
|
| 1935 |
locking.LEVEL_INSTANCE: self.cfg.GetInstanceNames(inst_uuids),
|
| 1936 |
locking.LEVEL_NODEGROUP: [self.group_uuid],
|
| 1937 |
locking.LEVEL_NODE: [], |
| 1938 |
|
| 1939 |
# This opcode is run by watcher every five minutes and acquires all nodes
|
| 1940 |
# for a group. It doesn't run for a long time, so it's better to acquire
|
| 1941 |
# the node allocation lock as well.
|
| 1942 |
locking.LEVEL_NODE_ALLOC: locking.ALL_SET, |
| 1943 |
} |
| 1944 |
|
| 1945 |
self.share_locks = ShareAll()
|
| 1946 |
|
| 1947 |
def DeclareLocks(self, level): |
| 1948 |
if level == locking.LEVEL_NODE:
|
| 1949 |
# Get members of node group; this is unsafe and needs verification later
|
| 1950 |
nodes = set(self.cfg.GetNodeGroup(self.group_uuid).members) |
| 1951 |
|
| 1952 |
# In Exec(), we warn about mirrored instances that have primary and
|
| 1953 |
# secondary living in separate node groups. To fully verify that
|
| 1954 |
# volumes for these instances are healthy, we will need to do an
|
| 1955 |
# extra call to their secondaries. We ensure here those nodes will
|
| 1956 |
# be locked.
|
| 1957 |
for inst_name in self.owned_locks(locking.LEVEL_INSTANCE): |
| 1958 |
# Important: access only the instances whose lock is owned
|
| 1959 |
instance = self.cfg.GetInstanceInfoByName(inst_name)
|
| 1960 |
if instance.disk_template in constants.DTS_INT_MIRROR: |
| 1961 |
nodes.update(self.cfg.GetInstanceSecondaryNodes(instance))
|
| 1962 |
|
| 1963 |
self.needed_locks[locking.LEVEL_NODE] = nodes
|
| 1964 |
|
| 1965 |
def CheckPrereq(self): |
| 1966 |
assert self.group_uuid in self.owned_locks(locking.LEVEL_NODEGROUP) |
| 1967 |
self.group_info = self.cfg.GetNodeGroup(self.group_uuid) |
| 1968 |
|
| 1969 |
group_node_uuids = set(self.group_info.members) |
| 1970 |
group_inst_uuids = \ |
| 1971 |
self.cfg.GetNodeGroupInstances(self.group_uuid, primary_only=True) |
| 1972 |
|
| 1973 |
unlocked_node_uuids = \ |
| 1974 |
group_node_uuids.difference(self.owned_locks(locking.LEVEL_NODE))
|
| 1975 |
|
| 1976 |
unlocked_inst_uuids = \ |
| 1977 |
group_inst_uuids.difference( |
| 1978 |
[self.cfg.GetInstanceInfoByName(name).uuid
|
| 1979 |
for name in self.owned_locks(locking.LEVEL_INSTANCE)]) |
| 1980 |
|
| 1981 |
if unlocked_node_uuids:
|
| 1982 |
raise errors.OpPrereqError(
|
| 1983 |
"Missing lock for nodes: %s" %
|
| 1984 |
utils.CommaJoin(self.cfg.GetNodeNames(unlocked_node_uuids)),
|
| 1985 |
errors.ECODE_STATE) |
| 1986 |
|
| 1987 |
if unlocked_inst_uuids:
|
| 1988 |
raise errors.OpPrereqError(
|
| 1989 |
"Missing lock for instances: %s" %
|
| 1990 |
utils.CommaJoin(self.cfg.GetInstanceNames(unlocked_inst_uuids)),
|
| 1991 |
errors.ECODE_STATE) |
| 1992 |
|
| 1993 |
self.all_node_info = self.cfg.GetAllNodesInfo() |
| 1994 |
self.all_inst_info = self.cfg.GetAllInstancesInfo() |
| 1995 |
|
| 1996 |
self.my_node_uuids = group_node_uuids
|
| 1997 |
self.my_node_info = dict((node_uuid, self.all_node_info[node_uuid]) |
| 1998 |
for node_uuid in group_node_uuids) |
| 1999 |
|
| 2000 |
self.my_inst_uuids = group_inst_uuids
|
| 2001 |
self.my_inst_info = dict((inst_uuid, self.all_inst_info[inst_uuid]) |
| 2002 |
for inst_uuid in group_inst_uuids) |
| 2003 |
|
| 2004 |
# We detect here the nodes that will need the extra RPC calls for verifying
|
| 2005 |
# split LV volumes; they should be locked.
|
| 2006 |
extra_lv_nodes = set()
|
| 2007 |
|
| 2008 |
for inst in self.my_inst_info.values(): |
| 2009 |
if inst.disk_template in constants.DTS_INT_MIRROR: |
| 2010 |
inst_nodes = self.cfg.GetInstanceNodes(inst)
|
| 2011 |
for nuuid in inst_nodes: |
| 2012 |
if self.all_node_info[nuuid].group != self.group_uuid: |
| 2013 |
extra_lv_nodes.add(nuuid) |
| 2014 |
|
| 2015 |
unlocked_lv_nodes = \ |
| 2016 |
extra_lv_nodes.difference(self.owned_locks(locking.LEVEL_NODE))
|
| 2017 |
|
| 2018 |
if unlocked_lv_nodes:
|
| 2019 |
raise errors.OpPrereqError("Missing node locks for LV check: %s" % |
| 2020 |
utils.CommaJoin(unlocked_lv_nodes), |
| 2021 |
errors.ECODE_STATE) |
| 2022 |
self.extra_lv_nodes = list(extra_lv_nodes) |
| 2023 |
|
| 2024 |
def _VerifyNode(self, ninfo, nresult): |
| 2025 |
"""Perform some basic validation on data returned from a node.
|
| 2026 |
|
| 2027 |
- check the result data structure is well formed and has all the
|
| 2028 |
mandatory fields
|
| 2029 |
- check ganeti version
|
| 2030 |
|
| 2031 |
@type ninfo: L{objects.Node}
|
| 2032 |
@param ninfo: the node to check
|
| 2033 |
@param nresult: the results from the node
|
| 2034 |
@rtype: boolean
|
| 2035 |
@return: whether overall this call was successful (and we can expect
|
| 2036 |
reasonable values in the respose)
|
| 2037 |
|
| 2038 |
"""
|
| 2039 |
# main result, nresult should be a non-empty dict
|
| 2040 |
test = not nresult or not isinstance(nresult, dict) |
| 2041 |
self._ErrorIf(test, constants.CV_ENODERPC, ninfo.name,
|
| 2042 |
"unable to verify node: no data returned")
|
| 2043 |
if test:
|
| 2044 |
return False |
| 2045 |
|
| 2046 |
# compares ganeti version
|
| 2047 |
local_version = constants.PROTOCOL_VERSION |
| 2048 |
remote_version = nresult.get("version", None) |
| 2049 |
test = not (remote_version and |
| 2050 |
isinstance(remote_version, (list, tuple)) and |
| 2051 |
len(remote_version) == 2) |
| 2052 |
self._ErrorIf(test, constants.CV_ENODERPC, ninfo.name,
|
| 2053 |
"connection to node returned invalid data")
|
| 2054 |
if test:
|
| 2055 |
return False |
| 2056 |
|
| 2057 |
test = local_version != remote_version[0]
|
| 2058 |
self._ErrorIf(test, constants.CV_ENODEVERSION, ninfo.name,
|
| 2059 |
"incompatible protocol versions: master %s,"
|
| 2060 |
" node %s", local_version, remote_version[0]) |
| 2061 |
if test:
|
| 2062 |
return False |
| 2063 |
|
| 2064 |
# node seems compatible, we can actually try to look into its results
|
| 2065 |
|
| 2066 |
# full package version
|
| 2067 |
self._ErrorIf(constants.RELEASE_VERSION != remote_version[1], |
| 2068 |
constants.CV_ENODEVERSION, ninfo.name, |
| 2069 |
"software version mismatch: master %s, node %s",
|
| 2070 |
constants.RELEASE_VERSION, remote_version[1],
|
| 2071 |
code=self.ETYPE_WARNING)
|
| 2072 |
|
| 2073 |
hyp_result = nresult.get(constants.NV_HYPERVISOR, None)
|
| 2074 |
if ninfo.vm_capable and isinstance(hyp_result, dict): |
| 2075 |
for hv_name, hv_result in hyp_result.iteritems(): |
| 2076 |
test = hv_result is not None |
| 2077 |
self._ErrorIf(test, constants.CV_ENODEHV, ninfo.name,
|
| 2078 |
"hypervisor %s verify failure: '%s'", hv_name, hv_result)
|
| 2079 |
|
| 2080 |
hvp_result = nresult.get(constants.NV_HVPARAMS, None)
|
| 2081 |
if ninfo.vm_capable and isinstance(hvp_result, list): |
| 2082 |
for item, hv_name, hv_result in hvp_result: |
| 2083 |
self._ErrorIf(True, constants.CV_ENODEHV, ninfo.name, |
| 2084 |
"hypervisor %s parameter verify failure (source %s): %s",
|
| 2085 |
hv_name, item, hv_result) |
| 2086 |
|
| 2087 |
test = nresult.get(constants.NV_NODESETUP, |
| 2088 |
["Missing NODESETUP results"])
|
| 2089 |
self._ErrorIf(test, constants.CV_ENODESETUP, ninfo.name,
|
| 2090 |
"node setup error: %s", "; ".join(test)) |
| 2091 |
|
| 2092 |
return True |
| 2093 |
|
| 2094 |
def _VerifyNodeTime(self, ninfo, nresult, |
| 2095 |
nvinfo_starttime, nvinfo_endtime): |
| 2096 |
"""Check the node time.
|
| 2097 |
|
| 2098 |
@type ninfo: L{objects.Node}
|
| 2099 |
@param ninfo: the node to check
|
| 2100 |
@param nresult: the remote results for the node
|
| 2101 |
@param nvinfo_starttime: the start time of the RPC call
|
| 2102 |
@param nvinfo_endtime: the end time of the RPC call
|
| 2103 |
|
| 2104 |
"""
|
| 2105 |
ntime = nresult.get(constants.NV_TIME, None)
|
| 2106 |
try:
|
| 2107 |
ntime_merged = utils.MergeTime(ntime) |
| 2108 |
except (ValueError, TypeError): |
| 2109 |
self._ErrorIf(True, constants.CV_ENODETIME, ninfo.name, |
| 2110 |
"Node returned invalid time")
|
| 2111 |
return
|
| 2112 |
|
| 2113 |
if ntime_merged < (nvinfo_starttime - constants.NODE_MAX_CLOCK_SKEW):
|
| 2114 |
ntime_diff = "%.01fs" % abs(nvinfo_starttime - ntime_merged) |
| 2115 |
elif ntime_merged > (nvinfo_endtime + constants.NODE_MAX_CLOCK_SKEW):
|
| 2116 |
ntime_diff = "%.01fs" % abs(ntime_merged - nvinfo_endtime) |
| 2117 |
else:
|
| 2118 |
ntime_diff = None
|
| 2119 |
|
| 2120 |
self._ErrorIf(ntime_diff is not None, constants.CV_ENODETIME, ninfo.name, |
| 2121 |
"Node time diverges by at least %s from master node time",
|
| 2122 |
ntime_diff) |
| 2123 |
|
| 2124 |
def _UpdateVerifyNodeLVM(self, ninfo, nresult, vg_name, nimg): |
| 2125 |
"""Check the node LVM results and update info for cross-node checks.
|
| 2126 |
|
| 2127 |
@type ninfo: L{objects.Node}
|
| 2128 |
@param ninfo: the node to check
|
| 2129 |
@param nresult: the remote results for the node
|
| 2130 |
@param vg_name: the configured VG name
|
| 2131 |
@type nimg: L{NodeImage}
|
| 2132 |
@param nimg: node image
|
| 2133 |
|
| 2134 |
"""
|
| 2135 |
if vg_name is None: |
| 2136 |
return
|
| 2137 |
|
| 2138 |
# checks vg existence and size > 20G
|
| 2139 |
vglist = nresult.get(constants.NV_VGLIST, None)
|
| 2140 |
test = not vglist
|
| 2141 |
self._ErrorIf(test, constants.CV_ENODELVM, ninfo.name,
|
| 2142 |
"unable to check volume groups")
|
| 2143 |
if not test: |
| 2144 |
vgstatus = utils.CheckVolumeGroupSize(vglist, vg_name, |
| 2145 |
constants.MIN_VG_SIZE) |
| 2146 |
self._ErrorIf(vgstatus, constants.CV_ENODELVM, ninfo.name, vgstatus)
|
| 2147 |
|
| 2148 |
# Check PVs
|
| 2149 |
(errmsgs, pvminmax) = CheckNodePVs(nresult, self._exclusive_storage)
|
| 2150 |
for em in errmsgs: |
| 2151 |
self._Error(constants.CV_ENODELVM, ninfo.name, em)
|
| 2152 |
if pvminmax is not None: |
| 2153 |
(nimg.pv_min, nimg.pv_max) = pvminmax |
| 2154 |
|
| 2155 |
def _VerifyGroupDRBDVersion(self, node_verify_infos): |
| 2156 |
"""Check cross-node DRBD version consistency.
|
| 2157 |
|
| 2158 |
@type node_verify_infos: dict
|
| 2159 |
@param node_verify_infos: infos about nodes as returned from the
|
| 2160 |
node_verify call.
|
| 2161 |
|
| 2162 |
"""
|
| 2163 |
node_versions = {}
|
| 2164 |
for node_uuid, ndata in node_verify_infos.items(): |
| 2165 |
nresult = ndata.payload |
| 2166 |
if nresult:
|
| 2167 |
version = nresult.get(constants.NV_DRBDVERSION, "Missing DRBD version")
|
| 2168 |
node_versions[node_uuid] = version |
| 2169 |
|
| 2170 |
if len(set(node_versions.values())) > 1: |
| 2171 |
for node_uuid, version in sorted(node_versions.items()): |
| 2172 |
msg = "DRBD version mismatch: %s" % version
|
| 2173 |
self._Error(constants.CV_ENODEDRBDHELPER, node_uuid, msg,
|
| 2174 |
code=self.ETYPE_WARNING)
|
| 2175 |
|
| 2176 |
def _VerifyGroupLVM(self, node_image, vg_name): |
| 2177 |
"""Check cross-node consistency in LVM.
|
| 2178 |
|
| 2179 |
@type node_image: dict
|
| 2180 |
@param node_image: info about nodes, mapping from node to names to
|
| 2181 |
L{NodeImage} objects
|
| 2182 |
@param vg_name: the configured VG name
|
| 2183 |
|
| 2184 |
"""
|
| 2185 |
if vg_name is None: |
| 2186 |
return
|
| 2187 |
|
| 2188 |
# Only exclusive storage needs this kind of checks
|
| 2189 |
if not self._exclusive_storage: |
| 2190 |
return
|
| 2191 |
|
| 2192 |
# exclusive_storage wants all PVs to have the same size (approximately),
|
| 2193 |
# if the smallest and the biggest ones are okay, everything is fine.
|
| 2194 |
# pv_min is None iff pv_max is None
|
| 2195 |
vals = filter((lambda ni: ni.pv_min is not None), node_image.values()) |
| 2196 |
if not vals: |
| 2197 |
return
|
| 2198 |
(pvmin, minnode_uuid) = min((ni.pv_min, ni.uuid) for ni in vals) |
| 2199 |
(pvmax, maxnode_uuid) = max((ni.pv_max, ni.uuid) for ni in vals) |
| 2200 |
bad = utils.LvmExclusiveTestBadPvSizes(pvmin, pvmax) |
| 2201 |
self._ErrorIf(bad, constants.CV_EGROUPDIFFERENTPVSIZE, self.group_info.name, |
| 2202 |
"PV sizes differ too much in the group; smallest (%s MB) is"
|
| 2203 |
" on %s, biggest (%s MB) is on %s",
|
| 2204 |
pvmin, self.cfg.GetNodeName(minnode_uuid),
|
| 2205 |
pvmax, self.cfg.GetNodeName(maxnode_uuid))
|
| 2206 |
|
| 2207 |
def _VerifyNodeBridges(self, ninfo, nresult, bridges): |
| 2208 |
"""Check the node bridges.
|
| 2209 |
|
| 2210 |
@type ninfo: L{objects.Node}
|
| 2211 |
@param ninfo: the node to check
|
| 2212 |
@param nresult: the remote results for the node
|
| 2213 |
@param bridges: the expected list of bridges
|
| 2214 |
|
| 2215 |
"""
|
| 2216 |
if not bridges: |
| 2217 |
return
|
| 2218 |
|
| 2219 |
missing = nresult.get(constants.NV_BRIDGES, None)
|
| 2220 |
test = not isinstance(missing, list) |
| 2221 |
self._ErrorIf(test, constants.CV_ENODENET, ninfo.name,
|
| 2222 |
"did not return valid bridge information")
|
| 2223 |
if not test: |
| 2224 |
self._ErrorIf(bool(missing), constants.CV_ENODENET, ninfo.name, |
| 2225 |
"missing bridges: %s" % utils.CommaJoin(sorted(missing))) |
| 2226 |
|
| 2227 |
def _VerifyNodeUserScripts(self, ninfo, nresult): |
| 2228 |
"""Check the results of user scripts presence and executability on the node
|
| 2229 |
|
| 2230 |
@type ninfo: L{objects.Node}
|
| 2231 |
@param ninfo: the node to check
|
| 2232 |
@param nresult: the remote results for the node
|
| 2233 |
|
| 2234 |
"""
|
| 2235 |
test = not constants.NV_USERSCRIPTS in nresult |
| 2236 |
self._ErrorIf(test, constants.CV_ENODEUSERSCRIPTS, ninfo.name,
|
| 2237 |
"did not return user scripts information")
|
| 2238 |
|
| 2239 |
broken_scripts = nresult.get(constants.NV_USERSCRIPTS, None)
|
| 2240 |
if not test: |
| 2241 |
self._ErrorIf(broken_scripts, constants.CV_ENODEUSERSCRIPTS, ninfo.name,
|
| 2242 |
"user scripts not present or not executable: %s" %
|
| 2243 |
utils.CommaJoin(sorted(broken_scripts)))
|
| 2244 |
|
| 2245 |
def _VerifyNodeNetwork(self, ninfo, nresult): |
| 2246 |
"""Check the node network connectivity results.
|
| 2247 |
|
| 2248 |
@type ninfo: L{objects.Node}
|
| 2249 |
@param ninfo: the node to check
|
| 2250 |
@param nresult: the remote results for the node
|
| 2251 |
|
| 2252 |
"""
|
| 2253 |
test = constants.NV_NODELIST not in nresult |
| 2254 |
self._ErrorIf(test, constants.CV_ENODESSH, ninfo.name,
|
| 2255 |
"node hasn't returned node ssh connectivity data")
|
| 2256 |
if not test: |
| 2257 |
if nresult[constants.NV_NODELIST]:
|
| 2258 |
for a_node, a_msg in nresult[constants.NV_NODELIST].items(): |
| 2259 |
self._ErrorIf(True, constants.CV_ENODESSH, ninfo.name, |
| 2260 |
"ssh communication with node '%s': %s", a_node, a_msg)
|
| 2261 |
|
| 2262 |
test = constants.NV_NODENETTEST not in nresult |
| 2263 |
self._ErrorIf(test, constants.CV_ENODENET, ninfo.name,
|
| 2264 |
"node hasn't returned node tcp connectivity data")
|
| 2265 |
if not test: |
| 2266 |
if nresult[constants.NV_NODENETTEST]:
|
| 2267 |
nlist = utils.NiceSort(nresult[constants.NV_NODENETTEST].keys()) |
| 2268 |
for anode in nlist: |
| 2269 |
self._ErrorIf(True, constants.CV_ENODENET, ninfo.name, |
| 2270 |
"tcp communication with node '%s': %s",
|
| 2271 |
anode, nresult[constants.NV_NODENETTEST][anode]) |
| 2272 |
|
| 2273 |
test = constants.NV_MASTERIP not in nresult |
| 2274 |
self._ErrorIf(test, constants.CV_ENODENET, ninfo.name,
|
| 2275 |
"node hasn't returned node master IP reachability data")
|
| 2276 |
if not test: |
| 2277 |
if not nresult[constants.NV_MASTERIP]: |
| 2278 |
if ninfo.uuid == self.master_node: |
| 2279 |
msg = "the master node cannot reach the master IP (not configured?)"
|
| 2280 |
else:
|
| 2281 |
msg = "cannot reach the master IP"
|
| 2282 |
self._ErrorIf(True, constants.CV_ENODENET, ninfo.name, msg) |
| 2283 |
|
| 2284 |
def _VerifyInstance(self, instance, node_image, diskstatus): |
| 2285 |
"""Verify an instance.
|
| 2286 |
|
| 2287 |
This function checks to see if the required block devices are
|
| 2288 |
available on the instance's node, and that the nodes are in the correct
|
| 2289 |
state.
|
| 2290 |
|
| 2291 |
"""
|
| 2292 |
pnode_uuid = instance.primary_node |
| 2293 |
pnode_img = node_image[pnode_uuid] |
| 2294 |
groupinfo = self.cfg.GetAllNodeGroupsInfo()
|
| 2295 |
|
| 2296 |
node_vol_should = {}
|
| 2297 |
self.cfg.GetInstanceLVsByNode(instance, lvmap=node_vol_should)
|
| 2298 |
|
| 2299 |
cluster = self.cfg.GetClusterInfo()
|
| 2300 |
ipolicy = ganeti.masterd.instance.CalculateGroupIPolicy(cluster, |
| 2301 |
self.group_info)
|
| 2302 |
err = ComputeIPolicyInstanceViolation(ipolicy, instance, self.cfg)
|
| 2303 |
self._ErrorIf(err, constants.CV_EINSTANCEPOLICY, instance.name,
|
| 2304 |
utils.CommaJoin(err), code=self.ETYPE_WARNING)
|
| 2305 |
|
| 2306 |
for node_uuid in node_vol_should: |
| 2307 |
n_img = node_image[node_uuid] |
| 2308 |
if n_img.offline or n_img.rpc_fail or n_img.lvm_fail: |
| 2309 |
# ignore missing volumes on offline or broken nodes
|
| 2310 |
continue
|
| 2311 |
for volume in node_vol_should[node_uuid]: |
| 2312 |
test = volume not in n_img.volumes |
| 2313 |
self._ErrorIf(test, constants.CV_EINSTANCEMISSINGDISK, instance.name,
|
| 2314 |
"volume %s missing on node %s", volume,
|
| 2315 |
self.cfg.GetNodeName(node_uuid))
|
| 2316 |
|
| 2317 |
if instance.admin_state == constants.ADMINST_UP:
|
| 2318 |
test = instance.uuid not in pnode_img.instances and not pnode_img.offline |
| 2319 |
self._ErrorIf(test, constants.CV_EINSTANCEDOWN, instance.name,
|
| 2320 |
"instance not running on its primary node %s",
|
| 2321 |
self.cfg.GetNodeName(pnode_uuid))
|
| 2322 |
self._ErrorIf(pnode_img.offline, constants.CV_EINSTANCEBADNODE,
|
| 2323 |
instance.name, "instance is marked as running and lives on"
|
| 2324 |
" offline node %s", self.cfg.GetNodeName(pnode_uuid)) |
| 2325 |
|
| 2326 |
diskdata = [(nname, success, status, idx) |
| 2327 |
for (nname, disks) in diskstatus.items() |
| 2328 |
for idx, (success, status) in enumerate(disks)] |
| 2329 |
|
| 2330 |
for nname, success, bdev_status, idx in diskdata: |
| 2331 |
# the 'ghost node' construction in Exec() ensures that we have a
|
| 2332 |
# node here
|
| 2333 |
snode = node_image[nname] |
| 2334 |
bad_snode = snode.ghost or snode.offline
|
| 2335 |
self._ErrorIf(instance.disks_active and |
| 2336 |
not success and not bad_snode, |
| 2337 |
constants.CV_EINSTANCEFAULTYDISK, instance.name, |
| 2338 |
"couldn't retrieve status for disk/%s on %s: %s",
|
| 2339 |
idx, self.cfg.GetNodeName(nname), bdev_status)
|
| 2340 |
|
| 2341 |
if instance.disks_active and success and \ |
| 2342 |
(bdev_status.is_degraded or
|
| 2343 |
bdev_status.ldisk_status != constants.LDS_OKAY): |
| 2344 |
msg = "disk/%s on %s" % (idx, self.cfg.GetNodeName(nname)) |
| 2345 |
if bdev_status.is_degraded:
|
| 2346 |
msg += " is degraded"
|
| 2347 |
if bdev_status.ldisk_status != constants.LDS_OKAY:
|
| 2348 |
msg += "; state is '%s'" % \
|
| 2349 |
constants.LDS_NAMES[bdev_status.ldisk_status] |
| 2350 |
|
| 2351 |
self._Error(constants.CV_EINSTANCEFAULTYDISK, instance.name, msg)
|
| 2352 |
|
| 2353 |
self._ErrorIf(pnode_img.rpc_fail and not pnode_img.offline, |
| 2354 |
constants.CV_ENODERPC, self.cfg.GetNodeName(pnode_uuid),
|
| 2355 |
"instance %s, connection to primary node failed",
|
| 2356 |
instance.name) |
| 2357 |
|
| 2358 |
secondary_nodes = self.cfg.GetInstanceSecondaryNodes(instance)
|
| 2359 |
self._ErrorIf(len(secondary_nodes) > 1, |
| 2360 |
constants.CV_EINSTANCELAYOUT, instance.name, |
| 2361 |
"instance has multiple secondary nodes: %s",
|
| 2362 |
utils.CommaJoin(secondary_nodes), |
| 2363 |
code=self.ETYPE_WARNING)
|
| 2364 |
|
| 2365 |
inst_nodes = self.cfg.GetInstanceNodes(instance)
|
| 2366 |
es_flags = rpc.GetExclusiveStorageForNodes(self.cfg, inst_nodes)
|
| 2367 |
if any(es_flags.values()): |
| 2368 |
if instance.disk_template not in constants.DTS_EXCL_STORAGE: |
| 2369 |
# Disk template not compatible with exclusive_storage: no instance
|
| 2370 |
# node should have the flag set
|
| 2371 |
es_nodes = [n |
| 2372 |
for (n, es) in es_flags.items() |
| 2373 |
if es]
|
| 2374 |
self._Error(constants.CV_EINSTANCEUNSUITABLENODE, instance.name,
|
| 2375 |
"instance has template %s, which is not supported on nodes"
|
| 2376 |
" that have exclusive storage set: %s",
|
| 2377 |
instance.disk_template, |
| 2378 |
utils.CommaJoin(self.cfg.GetNodeNames(es_nodes)))
|
| 2379 |
for (idx, disk) in enumerate(instance.disks): |
| 2380 |
self._ErrorIf(disk.spindles is None, |
| 2381 |
constants.CV_EINSTANCEMISSINGCFGPARAMETER, instance.name, |
| 2382 |
"number of spindles not configured for disk %s while"
|
| 2383 |
" exclusive storage is enabled, try running"
|
| 2384 |
" gnt-cluster repair-disk-sizes", idx)
|
| 2385 |
|
| 2386 |
if instance.disk_template in constants.DTS_INT_MIRROR: |
| 2387 |
instance_nodes = utils.NiceSort(inst_nodes) |
| 2388 |
instance_groups = {}
|
| 2389 |
|
| 2390 |
for node_uuid in instance_nodes: |
| 2391 |
instance_groups.setdefault(self.all_node_info[node_uuid].group,
|
| 2392 |
[]).append(node_uuid) |
| 2393 |
|
| 2394 |
pretty_list = [ |
| 2395 |
"%s (group %s)" % (utils.CommaJoin(self.cfg.GetNodeNames(nodes)), |
| 2396 |
groupinfo[group].name) |
| 2397 |
# Sort so that we always list the primary node first.
|
| 2398 |
for group, nodes in sorted(instance_groups.items(), |
| 2399 |
key=lambda (_, nodes): pnode_uuid in nodes, |
| 2400 |
reverse=True)]
|
| 2401 |
|
| 2402 |
self._ErrorIf(len(instance_groups) > 1, |
| 2403 |
constants.CV_EINSTANCESPLITGROUPS, |
| 2404 |
instance.name, "instance has primary and secondary nodes in"
|
| 2405 |
" different groups: %s", utils.CommaJoin(pretty_list),
|
| 2406 |
code=self.ETYPE_WARNING)
|
| 2407 |
|
| 2408 |
inst_nodes_offline = [] |
| 2409 |
for snode in secondary_nodes: |
| 2410 |
s_img = node_image[snode] |
| 2411 |
self._ErrorIf(s_img.rpc_fail and not s_img.offline, constants.CV_ENODERPC, |
| 2412 |
self.cfg.GetNodeName(snode),
|
| 2413 |
"instance %s, connection to secondary node failed",
|
| 2414 |
instance.name) |
| 2415 |
|
| 2416 |
if s_img.offline:
|
| 2417 |
inst_nodes_offline.append(snode) |
| 2418 |
|
| 2419 |
# warn that the instance lives on offline nodes
|
| 2420 |
self._ErrorIf(inst_nodes_offline, constants.CV_EINSTANCEBADNODE,
|
| 2421 |
instance.name, "instance has offline secondary node(s) %s",
|
| 2422 |
utils.CommaJoin(self.cfg.GetNodeNames(inst_nodes_offline)))
|
| 2423 |
# ... or ghost/non-vm_capable nodes
|
| 2424 |
for node_uuid in inst_nodes: |
| 2425 |
self._ErrorIf(node_image[node_uuid].ghost, constants.CV_EINSTANCEBADNODE,
|
| 2426 |
instance.name, "instance lives on ghost node %s",
|
| 2427 |
self.cfg.GetNodeName(node_uuid))
|
| 2428 |
self._ErrorIf(not node_image[node_uuid].vm_capable, |
| 2429 |
constants.CV_EINSTANCEBADNODE, instance.name, |
| 2430 |
"instance lives on non-vm_capable node %s",
|
| 2431 |
self.cfg.GetNodeName(node_uuid))
|
| 2432 |
|
| 2433 |
def _VerifyOrphanVolumes(self, node_vol_should, node_image, reserved): |
| 2434 |
"""Verify if there are any unknown volumes in the cluster.
|
| 2435 |
|
| 2436 |
The .os, .swap and backup volumes are ignored. All other volumes are
|
| 2437 |
reported as unknown.
|
| 2438 |
|
| 2439 |
@type reserved: L{ganeti.utils.FieldSet}
|
| 2440 |
@param reserved: a FieldSet of reserved volume names
|
| 2441 |
|
| 2442 |
"""
|
| 2443 |
for node_uuid, n_img in node_image.items(): |
| 2444 |
if (n_img.offline or n_img.rpc_fail or n_img.lvm_fail or |
| 2445 |
self.all_node_info[node_uuid].group != self.group_uuid): |
| 2446 |
# skip non-healthy nodes
|
| 2447 |
continue
|
| 2448 |
for volume in n_img.volumes: |
| 2449 |
test = ((node_uuid not in node_vol_should or |
| 2450 |
volume not in node_vol_should[node_uuid]) and |
| 2451 |
not reserved.Matches(volume))
|
| 2452 |
self._ErrorIf(test, constants.CV_ENODEORPHANLV,
|
| 2453 |
self.cfg.GetNodeName(node_uuid),
|
| 2454 |
"volume %s is unknown", volume,
|
| 2455 |
code=_VerifyErrors.ETYPE_WARNING) |
| 2456 |
|
| 2457 |
def _VerifyNPlusOneMemory(self, node_image, all_insts): |
| 2458 |
"""Verify N+1 Memory Resilience.
|
| 2459 |
|
| 2460 |
Check that if one single node dies we can still start all the
|
| 2461 |
instances it was primary for.
|
| 2462 |
|
| 2463 |
"""
|
| 2464 |
cluster_info = self.cfg.GetClusterInfo()
|
| 2465 |
for node_uuid, n_img in node_image.items(): |
| 2466 |
# This code checks that every node which is now listed as
|
| 2467 |
# secondary has enough memory to host all instances it is
|
| 2468 |
# supposed to should a single other node in the cluster fail.
|
| 2469 |
# FIXME: not ready for failover to an arbitrary node
|
| 2470 |
# FIXME: does not support file-backed instances
|
| 2471 |
# WARNING: we currently take into account down instances as well
|
| 2472 |
# as up ones, considering that even if they're down someone
|
| 2473 |
# might want to start them even in the event of a node failure.
|
| 2474 |
if n_img.offline or \ |
| 2475 |
self.all_node_info[node_uuid].group != self.group_uuid: |
| 2476 |
# we're skipping nodes marked offline and nodes in other groups from
|
| 2477 |
# the N+1 warning, since most likely we don't have good memory
|
| 2478 |
# information from them; we already list instances living on such
|
| 2479 |
# nodes, and that's enough warning
|
| 2480 |
continue
|
| 2481 |
#TODO(dynmem): also consider ballooning out other instances
|
| 2482 |
for prinode, inst_uuids in n_img.sbp.items(): |
| 2483 |
needed_mem = 0
|
| 2484 |
for inst_uuid in inst_uuids: |
| 2485 |
bep = cluster_info.FillBE(all_insts[inst_uuid]) |
| 2486 |
if bep[constants.BE_AUTO_BALANCE]:
|
| 2487 |
needed_mem += bep[constants.BE_MINMEM] |
| 2488 |
test = n_img.mfree < needed_mem |
| 2489 |
self._ErrorIf(test, constants.CV_ENODEN1,
|
| 2490 |
self.cfg.GetNodeName(node_uuid),
|
| 2491 |
"not enough memory to accomodate instance failovers"
|
| 2492 |
" should node %s fail (%dMiB needed, %dMiB available)",
|
| 2493 |
self.cfg.GetNodeName(prinode), needed_mem, n_img.mfree)
|
| 2494 |
|
| 2495 |
def _VerifyClientCertificates(self, nodes, all_nvinfo): |
| 2496 |
"""Verifies the consistency of the client certificates.
|
| 2497 |
|
| 2498 |
This includes several aspects:
|
| 2499 |
- the individual validation of all nodes' certificates
|
| 2500 |
- the consistency of the master candidate certificate map
|
| 2501 |
- the consistency of the master candidate certificate map with the
|
| 2502 |
certificates that the master candidates are actually using.
|
| 2503 |
|
| 2504 |
@param nodes: the list of nodes to consider in this verification
|
| 2505 |
@param all_nvinfo: the map of results of the verify_node call to
|
| 2506 |
all nodes
|
| 2507 |
|
| 2508 |
"""
|
| 2509 |
candidate_certs = self.cfg.GetClusterInfo().candidate_certs
|
| 2510 |
if candidate_certs is None or len(candidate_certs) == 0: |
| 2511 |
self._ErrorIf(
|
| 2512 |
True, constants.CV_ECLUSTERCLIENTCERT, None, |
| 2513 |
"The cluster's list of master candidate certificates is empty."
|
| 2514 |
"If you just updated the cluster, please run"
|
| 2515 |
" 'gnt-cluster renew-crypto --new-node-certificates'.")
|
| 2516 |
return
|
| 2517 |
|
| 2518 |
self._ErrorIf(
|
| 2519 |
len(candidate_certs) != len(set(candidate_certs.values())), |
| 2520 |
constants.CV_ECLUSTERCLIENTCERT, None,
|
| 2521 |
"There are at least two master candidates configured to use the same"
|
| 2522 |
" certificate.")
|
| 2523 |
|
| 2524 |
# collect the client certificate
|
| 2525 |
for node in nodes: |
| 2526 |
if node.offline:
|
| 2527 |
continue
|
| 2528 |
|
| 2529 |
nresult = all_nvinfo[node.uuid] |
| 2530 |
if nresult.fail_msg or not nresult.payload: |
| 2531 |
continue
|
| 2532 |
|
| 2533 |
(errcode, msg) = nresult.payload.get(constants.NV_CLIENT_CERT, None)
|
| 2534 |
|
| 2535 |
self._ErrorIf(
|
| 2536 |
errcode is not None, constants.CV_ECLUSTERCLIENTCERT, None, |
| 2537 |
"Client certificate of node '%s' failed validation: %s (code '%s')",
|
| 2538 |
node.uuid, msg, errcode) |
| 2539 |
|
| 2540 |
if not errcode: |
| 2541 |
digest = msg |
| 2542 |
if node.master_candidate:
|
| 2543 |
if node.uuid in candidate_certs: |
| 2544 |
self._ErrorIf(
|
| 2545 |
digest != candidate_certs[node.uuid], |
| 2546 |
constants.CV_ECLUSTERCLIENTCERT, None,
|
| 2547 |
"Client certificate digest of master candidate '%s' does not"
|
| 2548 |
" match its entry in the cluster's map of master candidate"
|
| 2549 |
" certificates. Expected: %s Got: %s", node.uuid,
|
| 2550 |
digest, candidate_certs[node.uuid]) |
| 2551 |
else:
|
| 2552 |
self._ErrorIf(
|
| 2553 |
True, constants.CV_ECLUSTERCLIENTCERT, None, |
| 2554 |
"The master candidate '%s' does not have an entry in the"
|
| 2555 |
" map of candidate certificates.", node.uuid)
|
| 2556 |
self._ErrorIf(
|
| 2557 |
digest in candidate_certs.values(),
|
| 2558 |
constants.CV_ECLUSTERCLIENTCERT, None,
|
| 2559 |
"Master candidate '%s' is using a certificate of another node.",
|
| 2560 |
node.uuid) |
| 2561 |
else:
|
| 2562 |
self._ErrorIf(
|
| 2563 |
node.uuid in candidate_certs,
|
| 2564 |
constants.CV_ECLUSTERCLIENTCERT, None,
|
| 2565 |
"Node '%s' is not a master candidate, but still listed in the"
|
| 2566 |
" map of master candidate certificates.", node.uuid)
|
| 2567 |
self._ErrorIf(
|
| 2568 |
(node.uuid not in candidate_certs) and |
| 2569 |
(digest in candidate_certs.values()),
|
| 2570 |
constants.CV_ECLUSTERCLIENTCERT, None,
|
| 2571 |
"Node '%s' is not a master candidate and is incorrectly using a"
|
| 2572 |
" certificate of another node which is master candidate.",
|
| 2573 |
node.uuid) |
| 2574 |
|
| 2575 |
def _VerifyFiles(self, nodes, master_node_uuid, all_nvinfo, |
| 2576 |
(files_all, files_opt, files_mc, files_vm)): |
| 2577 |
"""Verifies file checksums collected from all nodes.
|
| 2578 |
|
| 2579 |
@param nodes: List of L{objects.Node} objects
|
| 2580 |
@param master_node_uuid: UUID of master node
|
| 2581 |
@param all_nvinfo: RPC results
|
| 2582 |
|
| 2583 |
"""
|
| 2584 |
# Define functions determining which nodes to consider for a file
|
| 2585 |
files2nodefn = [ |
| 2586 |
(files_all, None),
|
| 2587 |
(files_mc, lambda node: (node.master_candidate or |
| 2588 |
node.uuid == master_node_uuid)), |
| 2589 |
(files_vm, lambda node: node.vm_capable),
|
| 2590 |
] |
| 2591 |
|
| 2592 |
# Build mapping from filename to list of nodes which should have the file
|
| 2593 |
nodefiles = {}
|
| 2594 |
for (files, fn) in files2nodefn: |
| 2595 |
if fn is None: |
| 2596 |
filenodes = nodes |
| 2597 |
else:
|
| 2598 |
filenodes = filter(fn, nodes)
|
| 2599 |
nodefiles.update((filename, |
| 2600 |
frozenset(map(operator.attrgetter("uuid"), filenodes))) |
| 2601 |
for filename in files) |
| 2602 |
|
| 2603 |
assert set(nodefiles) == (files_all | files_mc | files_vm) |
| 2604 |
|
| 2605 |
fileinfo = dict((filename, {}) for filename in nodefiles) |
| 2606 |
ignore_nodes = set()
|
| 2607 |
|
| 2608 |
for node in nodes: |
| 2609 |
if node.offline:
|
| 2610 |
ignore_nodes.add(node.uuid) |
| 2611 |
continue
|
| 2612 |
|
| 2613 |
nresult = all_nvinfo[node.uuid] |
| 2614 |
|
| 2615 |
if nresult.fail_msg or not nresult.payload: |
| 2616 |
node_files = None
|
| 2617 |
else:
|
| 2618 |
fingerprints = nresult.payload.get(constants.NV_FILELIST, {})
|
| 2619 |
node_files = dict((vcluster.LocalizeVirtualPath(key), value)
|
| 2620 |
for (key, value) in fingerprints.items()) |
| 2621 |
del fingerprints
|
| 2622 |
|
| 2623 |
test = not (node_files and isinstance(node_files, dict)) |
| 2624 |
self._ErrorIf(test, constants.CV_ENODEFILECHECK, node.name,
|
| 2625 |
"Node did not return file checksum data")
|
| 2626 |
if test:
|
| 2627 |
ignore_nodes.add(node.uuid) |
| 2628 |
continue
|
| 2629 |
|
| 2630 |
# Build per-checksum mapping from filename to nodes having it
|
| 2631 |
for (filename, checksum) in node_files.items(): |
| 2632 |
assert filename in nodefiles |
| 2633 |
fileinfo[filename].setdefault(checksum, set()).add(node.uuid)
|
| 2634 |
|
| 2635 |
for (filename, checksums) in fileinfo.items(): |
| 2636 |
assert compat.all(len(i) > 10 for i in checksums), "Invalid checksum" |
| 2637 |
|
| 2638 |
# Nodes having the file
|
| 2639 |
with_file = frozenset(node_uuid
|
| 2640 |
for node_uuids in fileinfo[filename].values() |
| 2641 |
for node_uuid in node_uuids) - ignore_nodes |
| 2642 |
|
| 2643 |
expected_nodes = nodefiles[filename] - ignore_nodes |
| 2644 |
|
| 2645 |
# Nodes missing file
|
| 2646 |
missing_file = expected_nodes - with_file |
| 2647 |
|
| 2648 |
if filename in files_opt: |
| 2649 |
# All or no nodes
|
| 2650 |
self._ErrorIf(missing_file and missing_file != expected_nodes, |
| 2651 |
constants.CV_ECLUSTERFILECHECK, None,
|
| 2652 |
"File %s is optional, but it must exist on all or no"
|
| 2653 |
" nodes (not found on %s)",
|
| 2654 |
filename, |
| 2655 |
utils.CommaJoin( |
| 2656 |
utils.NiceSort( |
| 2657 |
map(self.cfg.GetNodeName, missing_file)))) |
| 2658 |
else:
|
| 2659 |
self._ErrorIf(missing_file, constants.CV_ECLUSTERFILECHECK, None, |
| 2660 |
"File %s is missing from node(s) %s", filename,
|
| 2661 |
utils.CommaJoin( |
| 2662 |
utils.NiceSort( |
| 2663 |
map(self.cfg.GetNodeName, missing_file)))) |
| 2664 |
|
| 2665 |
# Warn if a node has a file it shouldn't
|
| 2666 |
unexpected = with_file - expected_nodes |
| 2667 |
self._ErrorIf(unexpected,
|
| 2668 |
constants.CV_ECLUSTERFILECHECK, None,
|
| 2669 |
"File %s should not exist on node(s) %s",
|
| 2670 |
filename, utils.CommaJoin( |
| 2671 |
utils.NiceSort(map(self.cfg.GetNodeName, unexpected)))) |
| 2672 |
|
| 2673 |
# See if there are multiple versions of the file
|
| 2674 |
test = len(checksums) > 1 |
| 2675 |
if test:
|
| 2676 |
variants = ["variant %s on %s" %
|
| 2677 |
(idx + 1,
|
| 2678 |
utils.CommaJoin(utils.NiceSort( |
| 2679 |
map(self.cfg.GetNodeName, node_uuids)))) |
| 2680 |
for (idx, (checksum, node_uuids)) in |
| 2681 |
enumerate(sorted(checksums.items()))] |
| 2682 |
else:
|
| 2683 |
variants = [] |
| 2684 |
|
| 2685 |
self._ErrorIf(test, constants.CV_ECLUSTERFILECHECK, None, |
| 2686 |
"File %s found with %s different checksums (%s)",
|
| 2687 |
filename, len(checksums), "; ".join(variants)) |
| 2688 |
|
| 2689 |
def _VerifyNodeDrbdHelper(self, ninfo, nresult, drbd_helper): |
| 2690 |
"""Verify the drbd helper.
|
| 2691 |
|
| 2692 |
"""
|
| 2693 |
if drbd_helper:
|
| 2694 |
helper_result = nresult.get(constants.NV_DRBDHELPER, None)
|
| 2695 |
test = (helper_result is None) |
| 2696 |
self._ErrorIf(test, constants.CV_ENODEDRBDHELPER, ninfo.name,
|
| 2697 |
"no drbd usermode helper returned")
|
| 2698 |
if helper_result:
|
| 2699 |
status, payload = helper_result |
| 2700 |
test = not status
|
| 2701 |
self._ErrorIf(test, constants.CV_ENODEDRBDHELPER, ninfo.name,
|
| 2702 |
"drbd usermode helper check unsuccessful: %s", payload)
|
| 2703 |
test = status and (payload != drbd_helper)
|
| 2704 |
self._ErrorIf(test, constants.CV_ENODEDRBDHELPER, ninfo.name,
|
| 2705 |
"wrong drbd usermode helper: %s", payload)
|
| 2706 |
|
| 2707 |
def _VerifyNodeDrbd(self, ninfo, nresult, instanceinfo, drbd_helper, |
| 2708 |
drbd_map): |
| 2709 |
"""Verifies and the node DRBD status.
|
| 2710 |
|
| 2711 |
@type ninfo: L{objects.Node}
|
| 2712 |
@param ninfo: the node to check
|
| 2713 |
@param nresult: the remote results for the node
|
| 2714 |
@param instanceinfo: the dict of instances
|
| 2715 |
@param drbd_helper: the configured DRBD usermode helper
|
| 2716 |
@param drbd_map: the DRBD map as returned by
|
| 2717 |
L{ganeti.config.ConfigWriter.ComputeDRBDMap}
|
| 2718 |
|
| 2719 |
"""
|
| 2720 |
self._VerifyNodeDrbdHelper(ninfo, nresult, drbd_helper)
|
| 2721 |
|
| 2722 |
# compute the DRBD minors
|
| 2723 |
node_drbd = {}
|
| 2724 |
for minor, inst_uuid in drbd_map[ninfo.uuid].items(): |
| 2725 |
test = inst_uuid not in instanceinfo |
| 2726 |
self._ErrorIf(test, constants.CV_ECLUSTERCFG, None, |
| 2727 |
"ghost instance '%s' in temporary DRBD map", inst_uuid)
|
| 2728 |
# ghost instance should not be running, but otherwise we
|
| 2729 |
# don't give double warnings (both ghost instance and
|
| 2730 |
# unallocated minor in use)
|
| 2731 |
if test:
|
| 2732 |
node_drbd[minor] = (inst_uuid, False)
|
| 2733 |
else:
|
| 2734 |
instance = instanceinfo[inst_uuid] |
| 2735 |
node_drbd[minor] = (inst_uuid, instance.disks_active) |
| 2736 |
|
| 2737 |
# and now check them
|
| 2738 |
used_minors = nresult.get(constants.NV_DRBDLIST, []) |
| 2739 |
test = not isinstance(used_minors, (tuple, list)) |
| 2740 |
self._ErrorIf(test, constants.CV_ENODEDRBD, ninfo.name,
|
| 2741 |
"cannot parse drbd status file: %s", str(used_minors)) |
| 2742 |
if test:
|
| 2743 |
# we cannot check drbd status
|
| 2744 |
return
|
| 2745 |
|
| 2746 |
for minor, (inst_uuid, must_exist) in node_drbd.items(): |
| 2747 |
test = minor not in used_minors and must_exist |
| 2748 |
self._ErrorIf(test, constants.CV_ENODEDRBD, ninfo.name,
|
| 2749 |
"drbd minor %d of instance %s is not active", minor,
|
| 2750 |
self.cfg.GetInstanceName(inst_uuid))
|
| 2751 |
for minor in used_minors: |
| 2752 |
test = minor not in node_drbd |
| 2753 |
self._ErrorIf(test, constants.CV_ENODEDRBD, ninfo.name,
|
| 2754 |
"unallocated drbd minor %d is in use", minor)
|
| 2755 |
|
| 2756 |
def _UpdateNodeOS(self, ninfo, nresult, nimg): |
| 2757 |
"""Builds the node OS structures.
|
| 2758 |
|
| 2759 |
@type ninfo: L{objects.Node}
|
| 2760 |
@param ninfo: the node to check
|
| 2761 |
@param nresult: the remote results for the node
|
| 2762 |
@param nimg: the node image object
|
| 2763 |
|
| 2764 |
"""
|
| 2765 |
remote_os = nresult.get(constants.NV_OSLIST, None)
|
| 2766 |
test = (not isinstance(remote_os, list) or |
| 2767 |
not compat.all(isinstance(v, list) and len(v) == 7 |
| 2768 |
for v in remote_os)) |
| 2769 |
|
| 2770 |
self._ErrorIf(test, constants.CV_ENODEOS, ninfo.name,
|
| 2771 |
"node hasn't returned valid OS data")
|
| 2772 |
|
| 2773 |
nimg.os_fail = test |
| 2774 |
|
| 2775 |
if test:
|
| 2776 |
return
|
| 2777 |
|
| 2778 |
os_dict = {}
|
| 2779 |
|
| 2780 |
for (name, os_path, status, diagnose,
|
| 2781 |
variants, parameters, api_ver) in nresult[constants.NV_OSLIST]:
|
| 2782 |
|
| 2783 |
if name not in os_dict: |
| 2784 |
os_dict[name] = [] |
| 2785 |
|
| 2786 |
# parameters is a list of lists instead of list of tuples due to
|
| 2787 |
# JSON lacking a real tuple type, fix it:
|
| 2788 |
parameters = [tuple(v) for v in parameters] |
| 2789 |
os_dict[name].append((os_path, status, diagnose, |
| 2790 |
set(variants), set(parameters), set(api_ver))) |
| 2791 |
|
| 2792 |
nimg.oslist = os_dict |
| 2793 |
|
| 2794 |
def _VerifyNodeOS(self, ninfo, nimg, base): |
| 2795 |
"""Verifies the node OS list.
|
| 2796 |
|
| 2797 |
@type ninfo: L{objects.Node}
|
| 2798 |
@param ninfo: the node to check
|
| 2799 |
@param nimg: the node image object
|
| 2800 |
@param base: the 'template' node we match against (e.g. from the master)
|
| 2801 |
|
| 2802 |
"""
|
| 2803 |
assert not nimg.os_fail, "Entered _VerifyNodeOS with failed OS rpc?" |
| 2804 |
|
| 2805 |
beautify_params = lambda l: ["%s: %s" % (k, v) for (k, v) in l] |
| 2806 |
for os_name, os_data in nimg.oslist.items(): |
| 2807 |
assert os_data, "Empty OS status for OS %s?!" % os_name |
| 2808 |
f_path, f_status, f_diag, f_var, f_param, f_api = os_data[0]
|
| 2809 |
self._ErrorIf(not f_status, constants.CV_ENODEOS, ninfo.name, |
| 2810 |
"Invalid OS %s (located at %s): %s",
|
| 2811 |
os_name, f_path, f_diag) |
| 2812 |
self._ErrorIf(len(os_data) > 1, constants.CV_ENODEOS, ninfo.name, |
| 2813 |
"OS '%s' has multiple entries"
|
| 2814 |
" (first one shadows the rest): %s",
|
| 2815 |
os_name, utils.CommaJoin([v[0] for v in os_data])) |
| 2816 |
# comparisons with the 'base' image
|
| 2817 |
test = os_name not in base.oslist |
| 2818 |
self._ErrorIf(test, constants.CV_ENODEOS, ninfo.name,
|
| 2819 |
"Extra OS %s not present on reference node (%s)",
|
| 2820 |
os_name, self.cfg.GetNodeName(base.uuid))
|
| 2821 |
if test:
|
| 2822 |
continue
|
| 2823 |
assert base.oslist[os_name], "Base node has empty OS status?" |
| 2824 |
_, b_status, _, b_var, b_param, b_api = base.oslist[os_name][0]
|
| 2825 |
if not b_status: |
| 2826 |
# base OS is invalid, skipping
|
| 2827 |
continue
|
| 2828 |
for kind, a, b in [("API version", f_api, b_api), |
| 2829 |
("variants list", f_var, b_var),
|
| 2830 |
("parameters", beautify_params(f_param),
|
| 2831 |
beautify_params(b_param))]: |
| 2832 |
self._ErrorIf(a != b, constants.CV_ENODEOS, ninfo.name,
|
| 2833 |
"OS %s for %s differs from reference node %s:"
|
| 2834 |
" [%s] vs. [%s]", kind, os_name,
|
| 2835 |
self.cfg.GetNodeName(base.uuid),
|
| 2836 |
utils.CommaJoin(sorted(a)), utils.CommaJoin(sorted(b))) |
| 2837 |
|
| 2838 |
# check any missing OSes
|
| 2839 |
missing = set(base.oslist.keys()).difference(nimg.oslist.keys())
|
| 2840 |
self._ErrorIf(missing, constants.CV_ENODEOS, ninfo.name,
|
| 2841 |
"OSes present on reference node %s"
|
| 2842 |
" but missing on this node: %s",
|
| 2843 |
self.cfg.GetNodeName(base.uuid), utils.CommaJoin(missing))
|
| 2844 |
|
| 2845 |
def _VerifyAcceptedFileStoragePaths(self, ninfo, nresult, is_master): |
| 2846 |
"""Verifies paths in L{pathutils.FILE_STORAGE_PATHS_FILE}.
|
| 2847 |
|
| 2848 |
@type ninfo: L{objects.Node}
|
| 2849 |
@param ninfo: the node to check
|
| 2850 |
@param nresult: the remote results for the node
|
| 2851 |
@type is_master: bool
|
| 2852 |
@param is_master: Whether node is the master node
|
| 2853 |
|
| 2854 |
"""
|
| 2855 |
cluster = self.cfg.GetClusterInfo()
|
| 2856 |
if (is_master and |
| 2857 |
(cluster.IsFileStorageEnabled() or
|
| 2858 |
cluster.IsSharedFileStorageEnabled())): |
| 2859 |
try:
|
| 2860 |
fspaths = nresult[constants.NV_ACCEPTED_STORAGE_PATHS] |
| 2861 |
except KeyError: |
| 2862 |
# This should never happen
|
| 2863 |
self._ErrorIf(True, constants.CV_ENODEFILESTORAGEPATHS, ninfo.name, |
| 2864 |
"Node did not return forbidden file storage paths")
|
| 2865 |
else:
|
| 2866 |
self._ErrorIf(fspaths, constants.CV_ENODEFILESTORAGEPATHS, ninfo.name,
|
| 2867 |
"Found forbidden file storage paths: %s",
|
| 2868 |
utils.CommaJoin(fspaths)) |
| 2869 |
else:
|
| 2870 |
self._ErrorIf(constants.NV_ACCEPTED_STORAGE_PATHS in nresult, |
| 2871 |
constants.CV_ENODEFILESTORAGEPATHS, ninfo.name, |
| 2872 |
"Node should not have returned forbidden file storage"
|
| 2873 |
" paths")
|
| 2874 |
|
| 2875 |
def _VerifyStoragePaths(self, ninfo, nresult, file_disk_template, |
| 2876 |
verify_key, error_key): |
| 2877 |
"""Verifies (file) storage paths.
|
| 2878 |
|
| 2879 |
@type ninfo: L{objects.Node}
|
| 2880 |
@param ninfo: the node to check
|
| 2881 |
@param nresult: the remote results for the node
|
| 2882 |
@type file_disk_template: string
|
| 2883 |
@param file_disk_template: file-based disk template, whose directory
|
| 2884 |
is supposed to be verified
|
| 2885 |
@type verify_key: string
|
| 2886 |
@param verify_key: key for the verification map of this file
|
| 2887 |
verification step
|
| 2888 |
@param error_key: error key to be added to the verification results
|
| 2889 |
in case something goes wrong in this verification step
|
| 2890 |
|
| 2891 |
"""
|
| 2892 |
assert (file_disk_template in utils.storage.GetDiskTemplatesOfStorageTypes( |
| 2893 |
constants.ST_FILE, constants.ST_SHARED_FILE |
| 2894 |
)) |
| 2895 |
|
| 2896 |
cluster = self.cfg.GetClusterInfo()
|
| 2897 |
if cluster.IsDiskTemplateEnabled(file_disk_template):
|
| 2898 |
self._ErrorIf(
|
| 2899 |
verify_key in nresult,
|
| 2900 |
error_key, ninfo.name, |
| 2901 |
"The configured %s storage path is unusable: %s" %
|
| 2902 |
(file_disk_template, nresult.get(verify_key))) |
| 2903 |
|
| 2904 |
def _VerifyFileStoragePaths(self, ninfo, nresult): |
| 2905 |
"""Verifies (file) storage paths.
|
| 2906 |
|
| 2907 |
@see: C{_VerifyStoragePaths}
|
| 2908 |
|
| 2909 |
"""
|
| 2910 |
self._VerifyStoragePaths(
|
| 2911 |
ninfo, nresult, constants.DT_FILE, |
| 2912 |
constants.NV_FILE_STORAGE_PATH, |
| 2913 |
constants.CV_ENODEFILESTORAGEPATHUNUSABLE) |
| 2914 |
|
| 2915 |
def _VerifySharedFileStoragePaths(self, ninfo, nresult): |
| 2916 |
"""Verifies (file) storage paths.
|
| 2917 |
|
| 2918 |
@see: C{_VerifyStoragePaths}
|
| 2919 |
|
| 2920 |
"""
|
| 2921 |
self._VerifyStoragePaths(
|
| 2922 |
ninfo, nresult, constants.DT_SHARED_FILE, |
| 2923 |
constants.NV_SHARED_FILE_STORAGE_PATH, |
| 2924 |
constants.CV_ENODESHAREDFILESTORAGEPATHUNUSABLE) |
| 2925 |
|
| 2926 |
def _VerifyOob(self, ninfo, nresult): |
| 2927 |
"""Verifies out of band functionality of a node.
|
| 2928 |
|
| 2929 |
@type ninfo: L{objects.Node}
|
| 2930 |
@param ninfo: the node to check
|
| 2931 |
@param nresult: the remote results for the node
|
| 2932 |
|
| 2933 |
"""
|
| 2934 |
# We just have to verify the paths on master and/or master candidates
|
| 2935 |
# as the oob helper is invoked on the master
|
| 2936 |
if ((ninfo.master_candidate or ninfo.master_capable) and |
| 2937 |
constants.NV_OOB_PATHS in nresult):
|
| 2938 |
for path_result in nresult[constants.NV_OOB_PATHS]: |
| 2939 |
self._ErrorIf(path_result, constants.CV_ENODEOOBPATH,
|
| 2940 |
ninfo.name, path_result) |
| 2941 |
|
| 2942 |
def _UpdateNodeVolumes(self, ninfo, nresult, nimg, vg_name): |
| 2943 |
"""Verifies and updates the node volume data.
|
| 2944 |
|
| 2945 |
This function will update a L{NodeImage}'s internal structures
|
| 2946 |
with data from the remote call.
|
| 2947 |
|
| 2948 |
@type ninfo: L{objects.Node}
|
| 2949 |
@param ninfo: the node to check
|
| 2950 |
@param nresult: the remote results for the node
|
| 2951 |
@param nimg: the node image object
|
| 2952 |
@param vg_name: the configured VG name
|
| 2953 |
|
| 2954 |
"""
|
| 2955 |
nimg.lvm_fail = True
|
| 2956 |
lvdata = nresult.get(constants.NV_LVLIST, "Missing LV data")
|
| 2957 |
if vg_name is None: |
| 2958 |
pass
|
| 2959 |
elif isinstance(lvdata, basestring): |
| 2960 |
self._ErrorIf(True, constants.CV_ENODELVM, ninfo.name, |
| 2961 |
"LVM problem on node: %s", utils.SafeEncode(lvdata))
|
| 2962 |
elif not isinstance(lvdata, dict): |
| 2963 |
self._ErrorIf(True, constants.CV_ENODELVM, ninfo.name, |
| 2964 |
"rpc call to node failed (lvlist)")
|
| 2965 |
else:
|
| 2966 |
nimg.volumes = lvdata |
| 2967 |
nimg.lvm_fail = False
|
| 2968 |
|
| 2969 |
def _UpdateNodeInstances(self, ninfo, nresult, nimg): |
| 2970 |
"""Verifies and updates the node instance list.
|
| 2971 |
|
| 2972 |
If the listing was successful, then updates this node's instance
|
| 2973 |
list. Otherwise, it marks the RPC call as failed for the instance
|
| 2974 |
list key.
|
| 2975 |
|
| 2976 |
@type ninfo: L{objects.Node}
|
| 2977 |
@param ninfo: the node to check
|
| 2978 |
@param nresult: the remote results for the node
|
| 2979 |
@param nimg: the node image object
|
| 2980 |
|
| 2981 |
"""
|
| 2982 |
idata = nresult.get(constants.NV_INSTANCELIST, None)
|
| 2983 |
test = not isinstance(idata, list) |
| 2984 |
self._ErrorIf(test, constants.CV_ENODEHV, ninfo.name,
|
| 2985 |
"rpc call to node failed (instancelist): %s",
|
| 2986 |
utils.SafeEncode(str(idata)))
|
| 2987 |
if test:
|
| 2988 |
nimg.hyp_fail = True
|
| 2989 |
else:
|
| 2990 |
nimg.instances = [inst.uuid for (_, inst) in |
| 2991 |
self.cfg.GetMultiInstanceInfoByName(idata)]
|
| 2992 |
|
| 2993 |
def _UpdateNodeInfo(self, ninfo, nresult, nimg, vg_name): |
| 2994 |
"""Verifies and computes a node information map
|
| 2995 |
|
| 2996 |
@type ninfo: L{objects.Node}
|
| 2997 |
@param ninfo: the node to check
|
| 2998 |
@param nresult: the remote results for the node
|
| 2999 |
@param nimg: the node image object
|
| 3000 |
@param vg_name: the configured VG name
|
| 3001 |
|
| 3002 |
"""
|
| 3003 |
# try to read free memory (from the hypervisor)
|
| 3004 |
hv_info = nresult.get(constants.NV_HVINFO, None)
|
| 3005 |
test = not isinstance(hv_info, dict) or "memory_free" not in hv_info |
| 3006 |
self._ErrorIf(test, constants.CV_ENODEHV, ninfo.name,
|
| 3007 |
"rpc call to node failed (hvinfo)")
|
| 3008 |
if not test: |
| 3009 |
try:
|
| 3010 |
nimg.mfree = int(hv_info["memory_free"]) |
| 3011 |
except (ValueError, TypeError): |
| 3012 |
self._ErrorIf(True, constants.CV_ENODERPC, ninfo.name, |
| 3013 |
"node returned invalid nodeinfo, check hypervisor")
|
| 3014 |
|
| 3015 |
# FIXME: devise a free space model for file based instances as well
|
| 3016 |
if vg_name is not None: |
| 3017 |
test = (constants.NV_VGLIST not in nresult or |
| 3018 |
vg_name not in nresult[constants.NV_VGLIST]) |
| 3019 |
self._ErrorIf(test, constants.CV_ENODELVM, ninfo.name,
|
| 3020 |
"node didn't return data for the volume group '%s'"
|
| 3021 |
" - it is either missing or broken", vg_name)
|
| 3022 |
if not test: |
| 3023 |
try:
|
| 3024 |
nimg.dfree = int(nresult[constants.NV_VGLIST][vg_name])
|
| 3025 |
except (ValueError, TypeError): |
| 3026 |
self._ErrorIf(True, constants.CV_ENODERPC, ninfo.name, |
| 3027 |
"node returned invalid LVM info, check LVM status")
|
| 3028 |
|
| 3029 |
def _CollectDiskInfo(self, node_uuids, node_image, instanceinfo): |
| 3030 |
"""Gets per-disk status information for all instances.
|
| 3031 |
|
| 3032 |
@type node_uuids: list of strings
|
| 3033 |
@param node_uuids: Node UUIDs
|
| 3034 |
@type node_image: dict of (UUID, L{objects.Node})
|
| 3035 |
@param node_image: Node objects
|
| 3036 |
@type instanceinfo: dict of (UUID, L{objects.Instance})
|
| 3037 |
@param instanceinfo: Instance objects
|
| 3038 |
@rtype: {instance: {node: [(succes, payload)]}}
|
| 3039 |
@return: a dictionary of per-instance dictionaries with nodes as
|
| 3040 |
keys and disk information as values; the disk information is a
|
| 3041 |
list of tuples (success, payload)
|
| 3042 |
|
| 3043 |
"""
|
| 3044 |
node_disks = {}
|
| 3045 |
node_disks_dev_inst_only = {}
|
| 3046 |
diskless_instances = set()
|
| 3047 |
nodisk_instances = set()
|
| 3048 |
diskless = constants.DT_DISKLESS |
| 3049 |
|
| 3050 |
for nuuid in node_uuids: |
| 3051 |
node_inst_uuids = list(itertools.chain(node_image[nuuid].pinst,
|
| 3052 |
node_image[nuuid].sinst)) |
| 3053 |
diskless_instances.update(uuid for uuid in node_inst_uuids |
| 3054 |
if instanceinfo[uuid].disk_template == diskless)
|
| 3055 |
disks = [(inst_uuid, disk) |
| 3056 |
for inst_uuid in node_inst_uuids |
| 3057 |
for disk in instanceinfo[inst_uuid].disks] |
| 3058 |
|
| 3059 |
if not disks: |
| 3060 |
nodisk_instances.update(uuid for uuid in node_inst_uuids |
| 3061 |
if instanceinfo[uuid].disk_template != diskless)
|
| 3062 |
# No need to collect data
|
| 3063 |
continue
|
| 3064 |
|
| 3065 |
node_disks[nuuid] = disks |
| 3066 |
|
| 3067 |
# _AnnotateDiskParams makes already copies of the disks
|
| 3068 |
dev_inst_only = [] |
| 3069 |
for (inst_uuid, dev) in disks: |
| 3070 |
(anno_disk,) = AnnotateDiskParams(instanceinfo[inst_uuid], [dev], |
| 3071 |
self.cfg)
|
| 3072 |
dev_inst_only.append((anno_disk, instanceinfo[inst_uuid])) |
| 3073 |
|
| 3074 |
node_disks_dev_inst_only[nuuid] = dev_inst_only |
| 3075 |
|
| 3076 |
assert len(node_disks) == len(node_disks_dev_inst_only) |
| 3077 |
|
| 3078 |
# Collect data from all nodes with disks
|
| 3079 |
result = self.rpc.call_blockdev_getmirrorstatus_multi(
|
| 3080 |
node_disks.keys(), node_disks_dev_inst_only) |
| 3081 |
|
| 3082 |
assert len(result) == len(node_disks) |
| 3083 |
|
| 3084 |
instdisk = {}
|
| 3085 |
|
| 3086 |
for (nuuid, nres) in result.items(): |
| 3087 |
node = self.cfg.GetNodeInfo(nuuid)
|
| 3088 |
disks = node_disks[node.uuid] |
| 3089 |
|
| 3090 |
if nres.offline:
|
| 3091 |
# No data from this node
|
| 3092 |
data = len(disks) * [(False, "node offline")] |
| 3093 |
else:
|
| 3094 |
msg = nres.fail_msg |
| 3095 |
self._ErrorIf(msg, constants.CV_ENODERPC, node.name,
|
| 3096 |
"while getting disk information: %s", msg)
|
| 3097 |
if msg:
|
| 3098 |
# No data from this node
|
| 3099 |
data = len(disks) * [(False, msg)] |
| 3100 |
else:
|
| 3101 |
data = [] |
| 3102 |
for idx, i in enumerate(nres.payload): |
| 3103 |
if isinstance(i, (tuple, list)) and len(i) == 2: |
| 3104 |
data.append(i) |
| 3105 |
else:
|
| 3106 |
logging.warning("Invalid result from node %s, entry %d: %s",
|
| 3107 |
node.name, idx, i) |
| 3108 |
data.append((False, "Invalid result from the remote node")) |
| 3109 |
|
| 3110 |
for ((inst_uuid, _), status) in zip(disks, data): |
| 3111 |
instdisk.setdefault(inst_uuid, {}).setdefault(node.uuid, []) \
|
| 3112 |
.append(status) |
| 3113 |
|
| 3114 |
# Add empty entries for diskless instances.
|
| 3115 |
for inst_uuid in diskless_instances: |
| 3116 |
assert inst_uuid not in instdisk |
| 3117 |
instdisk[inst_uuid] = {}
|
| 3118 |
# ...and disk-full instances that happen to have no disks
|
| 3119 |
for inst_uuid in nodisk_instances: |
| 3120 |
assert inst_uuid not in instdisk |
| 3121 |
instdisk[inst_uuid] = {}
|
| 3122 |
|
| 3123 |
assert compat.all(len(statuses) == len(instanceinfo[inst].disks) and |
| 3124 |
len(nuuids) <= len( |
| 3125 |
self.cfg.GetInstanceNodes(instanceinfo[inst])) and |
| 3126 |
compat.all(isinstance(s, (tuple, list)) and |
| 3127 |
len(s) == 2 for s in statuses) |
| 3128 |
for inst, nuuids in instdisk.items() |
| 3129 |
for nuuid, statuses in nuuids.items()) |
| 3130 |
if __debug__:
|
| 3131 |
instdisk_keys = set(instdisk)
|
| 3132 |
instanceinfo_keys = set(instanceinfo)
|
| 3133 |
assert instdisk_keys == instanceinfo_keys, \
|
| 3134 |
("instdisk keys (%s) do not match instanceinfo keys (%s)" %
|
| 3135 |
(instdisk_keys, instanceinfo_keys)) |
| 3136 |
|
| 3137 |
return instdisk
|
| 3138 |
|
| 3139 |
@staticmethod
|
| 3140 |
def _SshNodeSelector(group_uuid, all_nodes): |
| 3141 |
"""Create endless iterators for all potential SSH check hosts.
|
| 3142 |
|
| 3143 |
"""
|
| 3144 |
nodes = [node for node in all_nodes |
| 3145 |
if (node.group != group_uuid and |
| 3146 |
not node.offline)]
|
| 3147 |
keyfunc = operator.attrgetter("group")
|
| 3148 |
|
| 3149 |
return map(itertools.cycle, |
| 3150 |
[sorted(map(operator.attrgetter("name"), names)) |
| 3151 |
for _, names in itertools.groupby(sorted(nodes, key=keyfunc), |
| 3152 |
keyfunc)]) |
| 3153 |
|
| 3154 |
@classmethod
|
| 3155 |
def _SelectSshCheckNodes(cls, group_nodes, group_uuid, all_nodes): |
| 3156 |
"""Choose which nodes should talk to which other nodes.
|
| 3157 |
|
| 3158 |
We will make nodes contact all nodes in their group, and one node from
|
| 3159 |
every other group.
|
| 3160 |
|
| 3161 |
@warning: This algorithm has a known issue if one node group is much
|
| 3162 |
smaller than others (e.g. just one node). In such a case all other
|
| 3163 |
nodes will talk to the single node.
|
| 3164 |
|
| 3165 |
"""
|
| 3166 |
online_nodes = sorted(node.name for node in group_nodes if not node.offline) |
| 3167 |
sel = cls._SshNodeSelector(group_uuid, all_nodes) |
| 3168 |
|
| 3169 |
return (online_nodes,
|
| 3170 |
dict((name, sorted([i.next() for i in sel])) |
| 3171 |
for name in online_nodes)) |
| 3172 |
|
| 3173 |
def BuildHooksEnv(self): |
| 3174 |
"""Build hooks env.
|
| 3175 |
|
| 3176 |
Cluster-Verify hooks just ran in the post phase and their failure makes
|
| 3177 |
the output be logged in the verify output and the verification to fail.
|
| 3178 |
|
| 3179 |
"""
|
| 3180 |
env = {
|
| 3181 |
"CLUSTER_TAGS": " ".join(self.cfg.GetClusterInfo().GetTags()), |
| 3182 |
} |
| 3183 |
|
| 3184 |
env.update(("NODE_TAGS_%s" % node.name, " ".join(node.GetTags())) |
| 3185 |
for node in self.my_node_info.values()) |
| 3186 |
|
| 3187 |
return env
|
| 3188 |
|
| 3189 |
def BuildHooksNodes(self): |
| 3190 |
"""Build hooks nodes.
|
| 3191 |
|
| 3192 |
"""
|
| 3193 |
return ([], list(self.my_node_info.keys())) |
| 3194 |
|
| 3195 |
def Exec(self, feedback_fn): |
| 3196 |
"""Verify integrity of the node group, performing various test on nodes.
|
| 3197 |
|
| 3198 |
"""
|
| 3199 |
# This method has too many local variables. pylint: disable=R0914
|
| 3200 |
feedback_fn("* Verifying group '%s'" % self.group_info.name) |
| 3201 |
|
| 3202 |
if not self.my_node_uuids: |
| 3203 |
# empty node group
|
| 3204 |
feedback_fn("* Empty node group, skipping verification")
|
| 3205 |
return True |
| 3206 |
|
| 3207 |
self.bad = False |
| 3208 |
verbose = self.op.verbose
|
| 3209 |
self._feedback_fn = feedback_fn
|
| 3210 |
|
| 3211 |
vg_name = self.cfg.GetVGName()
|
| 3212 |
drbd_helper = self.cfg.GetDRBDHelper()
|
| 3213 |
cluster = self.cfg.GetClusterInfo()
|
| 3214 |
hypervisors = cluster.enabled_hypervisors |
| 3215 |
node_data_list = self.my_node_info.values()
|
| 3216 |
|
| 3217 |
i_non_redundant = [] # Non redundant instances
|
| 3218 |
i_non_a_balanced = [] # Non auto-balanced instances
|
| 3219 |
i_offline = 0 # Count of offline instances |
| 3220 |
n_offline = 0 # Count of offline nodes |
| 3221 |
n_drained = 0 # Count of nodes being drained |
| 3222 |
node_vol_should = {}
|
| 3223 |
|
| 3224 |
# FIXME: verify OS list
|
| 3225 |
|
| 3226 |
# File verification
|
| 3227 |
filemap = ComputeAncillaryFiles(cluster, False)
|
| 3228 |
|
| 3229 |
# do local checksums
|
| 3230 |
master_node_uuid = self.master_node = self.cfg.GetMasterNode() |
| 3231 |
master_ip = self.cfg.GetMasterIP()
|
| 3232 |
|
| 3233 |
feedback_fn("* Gathering data (%d nodes)" % len(self.my_node_uuids)) |
| 3234 |
|
| 3235 |
user_scripts = [] |
| 3236 |
if self.cfg.GetUseExternalMipScript(): |
| 3237 |
user_scripts.append(pathutils.EXTERNAL_MASTER_SETUP_SCRIPT) |
| 3238 |
|
| 3239 |
node_verify_param = {
|
| 3240 |
constants.NV_FILELIST: |
| 3241 |
map(vcluster.MakeVirtualPath,
|
| 3242 |
utils.UniqueSequence(filename |
| 3243 |
for files in filemap |
| 3244 |
for filename in files)), |
| 3245 |
constants.NV_NODELIST: |
| 3246 |
self._SelectSshCheckNodes(node_data_list, self.group_uuid, |
| 3247 |
self.all_node_info.values()),
|
| 3248 |
constants.NV_HYPERVISOR: hypervisors, |
| 3249 |
constants.NV_HVPARAMS: |
| 3250 |
_GetAllHypervisorParameters(cluster, self.all_inst_info.values()),
|
| 3251 |
constants.NV_NODENETTEST: [(node.name, node.primary_ip, node.secondary_ip) |
| 3252 |
for node in node_data_list |
| 3253 |
if not node.offline], |
| 3254 |
constants.NV_INSTANCELIST: hypervisors, |
| 3255 |
constants.NV_VERSION: None,
|
| 3256 |
constants.NV_HVINFO: self.cfg.GetHypervisorType(),
|
| 3257 |
constants.NV_NODESETUP: None,
|
| 3258 |
constants.NV_TIME: None,
|
| 3259 |
constants.NV_MASTERIP: (self.cfg.GetMasterNodeName(), master_ip),
|
| 3260 |
constants.NV_OSLIST: None,
|
| 3261 |
constants.NV_VMNODES: self.cfg.GetNonVmCapableNodeList(),
|
| 3262 |
constants.NV_USERSCRIPTS: user_scripts, |
| 3263 |
constants.NV_CLIENT_CERT: None,
|
| 3264 |
} |
| 3265 |
|
| 3266 |
if vg_name is not None: |
| 3267 |
node_verify_param[constants.NV_VGLIST] = None
|
| 3268 |
node_verify_param[constants.NV_LVLIST] = vg_name |
| 3269 |
node_verify_param[constants.NV_PVLIST] = [vg_name] |
| 3270 |
|
| 3271 |
if cluster.IsDiskTemplateEnabled(constants.DT_DRBD8):
|
| 3272 |
if drbd_helper:
|
| 3273 |
node_verify_param[constants.NV_DRBDVERSION] = None
|
| 3274 |
node_verify_param[constants.NV_DRBDLIST] = None
|
| 3275 |
node_verify_param[constants.NV_DRBDHELPER] = drbd_helper |
| 3276 |
|
| 3277 |
if cluster.IsFileStorageEnabled() or \ |
| 3278 |
cluster.IsSharedFileStorageEnabled(): |
| 3279 |
# Load file storage paths only from master node
|
| 3280 |
node_verify_param[constants.NV_ACCEPTED_STORAGE_PATHS] = \ |
| 3281 |
self.cfg.GetMasterNodeName()
|
| 3282 |
if cluster.IsFileStorageEnabled():
|
| 3283 |
node_verify_param[constants.NV_FILE_STORAGE_PATH] = \ |
| 3284 |
cluster.file_storage_dir |
| 3285 |
|
| 3286 |
# bridge checks
|
| 3287 |
# FIXME: this needs to be changed per node-group, not cluster-wide
|
| 3288 |
bridges = set()
|
| 3289 |
default_nicpp = cluster.nicparams[constants.PP_DEFAULT] |
| 3290 |
if default_nicpp[constants.NIC_MODE] == constants.NIC_MODE_BRIDGED:
|
| 3291 |
bridges.add(default_nicpp[constants.NIC_LINK]) |
| 3292 |
for inst_uuid in self.my_inst_info.values(): |
| 3293 |
for nic in inst_uuid.nics: |
| 3294 |
full_nic = cluster.SimpleFillNIC(nic.nicparams) |
| 3295 |
if full_nic[constants.NIC_MODE] == constants.NIC_MODE_BRIDGED:
|
| 3296 |
bridges.add(full_nic[constants.NIC_LINK]) |
| 3297 |
|
| 3298 |
if bridges:
|
| 3299 |
node_verify_param[constants.NV_BRIDGES] = list(bridges)
|
| 3300 |
|
| 3301 |
# Build our expected cluster state
|
| 3302 |
node_image = dict((node.uuid, self.NodeImage(offline=node.offline, |
| 3303 |
uuid=node.uuid, |
| 3304 |
vm_capable=node.vm_capable)) |
| 3305 |
for node in node_data_list) |
| 3306 |
|
| 3307 |
# Gather OOB paths
|
| 3308 |
oob_paths = [] |
| 3309 |
for node in self.all_node_info.values(): |
| 3310 |
path = SupportsOob(self.cfg, node)
|
| 3311 |
if path and path not in oob_paths: |
| 3312 |
oob_paths.append(path) |
| 3313 |
|
| 3314 |
if oob_paths:
|
| 3315 |
node_verify_param[constants.NV_OOB_PATHS] = oob_paths |
| 3316 |
|
| 3317 |
for inst_uuid in self.my_inst_uuids: |
| 3318 |
instance = self.my_inst_info[inst_uuid]
|
| 3319 |
if instance.admin_state == constants.ADMINST_OFFLINE:
|
| 3320 |
i_offline += 1
|
| 3321 |
|
| 3322 |
inst_nodes = self.cfg.GetInstanceNodes(instance)
|
| 3323 |
for nuuid in inst_nodes: |
| 3324 |
if nuuid not in node_image: |
| 3325 |
gnode = self.NodeImage(uuid=nuuid)
|
| 3326 |
gnode.ghost = (nuuid not in self.all_node_info) |
| 3327 |
node_image[nuuid] = gnode |
| 3328 |
|
| 3329 |
self.cfg.GetInstanceLVsByNode(instance, lvmap=node_vol_should)
|
| 3330 |
|
| 3331 |
pnode = instance.primary_node |
| 3332 |
node_image[pnode].pinst.append(instance.uuid) |
| 3333 |
|
| 3334 |
for snode in self.cfg.GetInstanceSecondaryNodes(instance): |
| 3335 |
nimg = node_image[snode] |
| 3336 |
nimg.sinst.append(instance.uuid) |
| 3337 |
if pnode not in nimg.sbp: |
| 3338 |
nimg.sbp[pnode] = [] |
| 3339 |
nimg.sbp[pnode].append(instance.uuid) |
| 3340 |
|
| 3341 |
es_flags = rpc.GetExclusiveStorageForNodes(self.cfg,
|
| 3342 |
self.my_node_info.keys())
|
| 3343 |
# The value of exclusive_storage should be the same across the group, so if
|
| 3344 |
# it's True for at least a node, we act as if it were set for all the nodes
|
| 3345 |
self._exclusive_storage = compat.any(es_flags.values())
|
| 3346 |
if self._exclusive_storage: |
| 3347 |
node_verify_param[constants.NV_EXCLUSIVEPVS] = True
|
| 3348 |
|
| 3349 |
node_group_uuids = dict(map(lambda n: (n.name, n.group), |
| 3350 |
self.cfg.GetAllNodesInfo().values()))
|
| 3351 |
groups_config = self.cfg.GetAllNodeGroupsInfoDict()
|
| 3352 |
|
| 3353 |
# At this point, we have the in-memory data structures complete,
|
| 3354 |
# except for the runtime information, which we'll gather next
|
| 3355 |
|
| 3356 |
# Due to the way our RPC system works, exact response times cannot be
|
| 3357 |
# guaranteed (e.g. a broken node could run into a timeout). By keeping the
|
| 3358 |
# time before and after executing the request, we can at least have a time
|
| 3359 |
# window.
|
| 3360 |
nvinfo_starttime = time.time() |
| 3361 |
all_nvinfo = self.rpc.call_node_verify(self.my_node_uuids, |
| 3362 |
node_verify_param, |
| 3363 |
self.cfg.GetClusterName(),
|
| 3364 |
self.cfg.GetClusterInfo().hvparams,
|
| 3365 |
node_group_uuids, |
| 3366 |
groups_config) |
| 3367 |
nvinfo_endtime = time.time() |
| 3368 |
|
| 3369 |
if self.extra_lv_nodes and vg_name is not None: |
| 3370 |
extra_lv_nvinfo = \ |
| 3371 |
self.rpc.call_node_verify(self.extra_lv_nodes, |
| 3372 |
{constants.NV_LVLIST: vg_name},
|
| 3373 |
self.cfg.GetClusterName(),
|
| 3374 |
self.cfg.GetClusterInfo().hvparams,
|
| 3375 |
node_group_uuids, |
| 3376 |
groups_config) |
| 3377 |
else:
|
| 3378 |
extra_lv_nvinfo = {}
|
| 3379 |
|
| 3380 |
all_drbd_map = self.cfg.ComputeDRBDMap()
|
| 3381 |
|
| 3382 |
feedback_fn("* Gathering disk information (%s nodes)" %
|
| 3383 |
len(self.my_node_uuids)) |
| 3384 |
instdisk = self._CollectDiskInfo(self.my_node_info.keys(), node_image, |
| 3385 |
self.my_inst_info)
|
| 3386 |
|
| 3387 |
feedback_fn("* Verifying configuration file consistency")
|
| 3388 |
|
| 3389 |
self._VerifyClientCertificates(self.my_node_info.values(), all_nvinfo) |
| 3390 |
# If not all nodes are being checked, we need to make sure the master node
|
| 3391 |
# and a non-checked vm_capable node are in the list.
|
| 3392 |
absent_node_uuids = set(self.all_node_info).difference(self.my_node_info) |
| 3393 |
if absent_node_uuids:
|
| 3394 |
vf_nvinfo = all_nvinfo.copy() |
| 3395 |
vf_node_info = list(self.my_node_info.values()) |
| 3396 |
additional_node_uuids = [] |
| 3397 |
if master_node_uuid not in self.my_node_info: |
| 3398 |
additional_node_uuids.append(master_node_uuid) |
| 3399 |
vf_node_info.append(self.all_node_info[master_node_uuid])
|
| 3400 |
# Add the first vm_capable node we find which is not included,
|
| 3401 |
# excluding the master node (which we already have)
|
| 3402 |
for node_uuid in absent_node_uuids: |
| 3403 |
nodeinfo = self.all_node_info[node_uuid]
|
| 3404 |
if (nodeinfo.vm_capable and not nodeinfo.offline and |
| 3405 |
node_uuid != master_node_uuid): |
| 3406 |
additional_node_uuids.append(node_uuid) |
| 3407 |
vf_node_info.append(self.all_node_info[node_uuid])
|
| 3408 |
break
|
| 3409 |
key = constants.NV_FILELIST |
| 3410 |
vf_nvinfo.update(self.rpc.call_node_verify(
|
| 3411 |
additional_node_uuids, {key: node_verify_param[key]},
|
| 3412 |
self.cfg.GetClusterName(), self.cfg.GetClusterInfo().hvparams, |
| 3413 |
node_group_uuids, |
| 3414 |
groups_config)) |
| 3415 |
else:
|
| 3416 |
vf_nvinfo = all_nvinfo |
| 3417 |
vf_node_info = self.my_node_info.values()
|
| 3418 |
|
| 3419 |
self._VerifyFiles(vf_node_info, master_node_uuid, vf_nvinfo, filemap)
|
| 3420 |
|
| 3421 |
feedback_fn("* Verifying node status")
|
| 3422 |
|
| 3423 |
refos_img = None
|
| 3424 |
|
| 3425 |
for node_i in node_data_list: |
| 3426 |
nimg = node_image[node_i.uuid] |
| 3427 |
|
| 3428 |
if node_i.offline:
|
| 3429 |
if verbose:
|
| 3430 |
feedback_fn("* Skipping offline node %s" % (node_i.name,))
|
| 3431 |
n_offline += 1
|
| 3432 |
continue
|
| 3433 |
|
| 3434 |
if node_i.uuid == master_node_uuid:
|
| 3435 |
ntype = "master"
|
| 3436 |
elif node_i.master_candidate:
|
| 3437 |
ntype = "master candidate"
|
| 3438 |
elif node_i.drained:
|
| 3439 |
ntype = "drained"
|
| 3440 |
n_drained += 1
|
| 3441 |
else:
|
| 3442 |
ntype = "regular"
|
| 3443 |
if verbose:
|
| 3444 |
feedback_fn("* Verifying node %s (%s)" % (node_i.name, ntype))
|
| 3445 |
|
| 3446 |
msg = all_nvinfo[node_i.uuid].fail_msg |
| 3447 |
self._ErrorIf(msg, constants.CV_ENODERPC, node_i.name,
|
| 3448 |
"while contacting node: %s", msg)
|
| 3449 |
if msg:
|
| 3450 |
nimg.rpc_fail = True
|
| 3451 |
continue
|
| 3452 |
|
| 3453 |
nresult = all_nvinfo[node_i.uuid].payload |
| 3454 |
|
| 3455 |
nimg.call_ok = self._VerifyNode(node_i, nresult)
|
| 3456 |
self._VerifyNodeTime(node_i, nresult, nvinfo_starttime, nvinfo_endtime)
|
| 3457 |
self._VerifyNodeNetwork(node_i, nresult)
|
| 3458 |
self._VerifyNodeUserScripts(node_i, nresult)
|
| 3459 |
self._VerifyOob(node_i, nresult)
|
| 3460 |
self._VerifyAcceptedFileStoragePaths(node_i, nresult,
|
| 3461 |
node_i.uuid == master_node_uuid) |
| 3462 |
self._VerifyFileStoragePaths(node_i, nresult)
|
| 3463 |
self._VerifySharedFileStoragePaths(node_i, nresult)
|
| 3464 |
|
| 3465 |
if nimg.vm_capable:
|
| 3466 |
self._UpdateVerifyNodeLVM(node_i, nresult, vg_name, nimg)
|
| 3467 |
self._VerifyNodeDrbd(node_i, nresult, self.all_inst_info, drbd_helper, |
| 3468 |
all_drbd_map) |
| 3469 |
|
| 3470 |
self._UpdateNodeVolumes(node_i, nresult, nimg, vg_name)
|
| 3471 |
self._UpdateNodeInstances(node_i, nresult, nimg)
|
| 3472 |
self._UpdateNodeInfo(node_i, nresult, nimg, vg_name)
|
| 3473 |
self._UpdateNodeOS(node_i, nresult, nimg)
|
| 3474 |
|
| 3475 |
if not nimg.os_fail: |
| 3476 |
if refos_img is None: |
| 3477 |
refos_img = nimg |
| 3478 |
self._VerifyNodeOS(node_i, nimg, refos_img)
|
| 3479 |
self._VerifyNodeBridges(node_i, nresult, bridges)
|
| 3480 |
|
| 3481 |
# Check whether all running instances are primary for the node. (This
|
| 3482 |
# can no longer be done from _VerifyInstance below, since some of the
|
| 3483 |
# wrong instances could be from other node groups.)
|
| 3484 |
non_primary_inst_uuids = set(nimg.instances).difference(nimg.pinst)
|
| 3485 |
|
| 3486 |
for inst_uuid in non_primary_inst_uuids: |
| 3487 |
test = inst_uuid in self.all_inst_info |
| 3488 |
self._ErrorIf(test, constants.CV_EINSTANCEWRONGNODE,
|
| 3489 |
self.cfg.GetInstanceName(inst_uuid),
|
| 3490 |
"instance should not run on node %s", node_i.name)
|
| 3491 |
self._ErrorIf(not test, constants.CV_ENODEORPHANINSTANCE, node_i.name, |
| 3492 |
"node is running unknown instance %s", inst_uuid)
|
| 3493 |
|
| 3494 |
self._VerifyGroupDRBDVersion(all_nvinfo)
|
| 3495 |
self._VerifyGroupLVM(node_image, vg_name)
|
| 3496 |
|
| 3497 |
for node_uuid, result in extra_lv_nvinfo.items(): |
| 3498 |
self._UpdateNodeVolumes(self.all_node_info[node_uuid], result.payload, |
| 3499 |
node_image[node_uuid], vg_name) |
| 3500 |
|
| 3501 |
feedback_fn("* Verifying instance status")
|
| 3502 |
for inst_uuid in self.my_inst_uuids: |
| 3503 |
instance = self.my_inst_info[inst_uuid]
|
| 3504 |
if verbose:
|
| 3505 |
feedback_fn("* Verifying instance %s" % instance.name)
|
| 3506 |
self._VerifyInstance(instance, node_image, instdisk[inst_uuid])
|
| 3507 |
|
| 3508 |
# If the instance is non-redundant we cannot survive losing its primary
|
| 3509 |
# node, so we are not N+1 compliant.
|
| 3510 |
if instance.disk_template not in constants.DTS_MIRRORED: |
| 3511 |
i_non_redundant.append(instance) |
| 3512 |
|
| 3513 |
if not cluster.FillBE(instance)[constants.BE_AUTO_BALANCE]: |
| 3514 |
i_non_a_balanced.append(instance) |
| 3515 |
|
| 3516 |
feedback_fn("* Verifying orphan volumes")
|
| 3517 |
reserved = utils.FieldSet(*cluster.reserved_lvs) |
| 3518 |
|
| 3519 |
# We will get spurious "unknown volume" warnings if any node of this group
|
| 3520 |
# is secondary for an instance whose primary is in another group. To avoid
|
| 3521 |
# them, we find these instances and add their volumes to node_vol_should.
|
| 3522 |
for instance in self.all_inst_info.values(): |
| 3523 |
for secondary in self.cfg.GetInstanceSecondaryNodes(instance): |
| 3524 |
if (secondary in self.my_node_info |
| 3525 |
and instance.name not in self.my_inst_info): |
| 3526 |
self.cfg.GetInstanceLVsByNode(instance, lvmap=node_vol_should)
|
| 3527 |
break
|
| 3528 |
|
| 3529 |
self._VerifyOrphanVolumes(node_vol_should, node_image, reserved)
|
| 3530 |
|
| 3531 |
if constants.VERIFY_NPLUSONE_MEM not in self.op.skip_checks: |
| 3532 |
feedback_fn("* Verifying N+1 Memory redundancy")
|
| 3533 |
self._VerifyNPlusOneMemory(node_image, self.my_inst_info) |
| 3534 |
|
| 3535 |
feedback_fn("* Other Notes")
|
| 3536 |
if i_non_redundant:
|
| 3537 |
feedback_fn(" - NOTICE: %d non-redundant instance(s) found."
|
| 3538 |
% len(i_non_redundant))
|
| 3539 |
|
| 3540 |
if i_non_a_balanced:
|
| 3541 |
feedback_fn(" - NOTICE: %d non-auto-balanced instance(s) found."
|
| 3542 |
% len(i_non_a_balanced))
|
| 3543 |
|
| 3544 |
if i_offline:
|
| 3545 |
feedback_fn(" - NOTICE: %d offline instance(s) found." % i_offline)
|
| 3546 |
|
| 3547 |
if n_offline:
|
| 3548 |
feedback_fn(" - NOTICE: %d offline node(s) found." % n_offline)
|
| 3549 |
|
| 3550 |
if n_drained:
|
| 3551 |
feedback_fn(" - NOTICE: %d drained node(s) found." % n_drained)
|
| 3552 |
|
| 3553 |
return not self.bad |
| 3554 |
|
| 3555 |
def HooksCallBack(self, phase, hooks_results, feedback_fn, lu_result): |
| 3556 |
"""Analyze the post-hooks' result
|
| 3557 |
|
| 3558 |
This method analyses the hook result, handles it, and sends some
|
| 3559 |
nicely-formatted feedback back to the user.
|
| 3560 |
|
| 3561 |
@param phase: one of L{constants.HOOKS_PHASE_POST} or
|
| 3562 |
L{constants.HOOKS_PHASE_PRE}; it denotes the hooks phase
|
| 3563 |
@param hooks_results: the results of the multi-node hooks rpc call
|
| 3564 |
@param feedback_fn: function used send feedback back to the caller
|
| 3565 |
@param lu_result: previous Exec result
|
| 3566 |
@return: the new Exec result, based on the previous result
|
| 3567 |
and hook results
|
| 3568 |
|
| 3569 |
"""
|
| 3570 |
# We only really run POST phase hooks, only for non-empty groups,
|
| 3571 |
# and are only interested in their results
|
| 3572 |
if not self.my_node_uuids: |
| 3573 |
# empty node group
|
| 3574 |
pass
|
| 3575 |
elif phase == constants.HOOKS_PHASE_POST:
|
| 3576 |
# Used to change hooks' output to proper indentation
|
| 3577 |
feedback_fn("* Hooks Results")
|
| 3578 |
assert hooks_results, "invalid result from hooks" |
| 3579 |
|
| 3580 |
for node_name in hooks_results: |
| 3581 |
res = hooks_results[node_name] |
| 3582 |
msg = res.fail_msg |
| 3583 |
test = msg and not res.offline |
| 3584 |
self._ErrorIf(test, constants.CV_ENODEHOOKS, node_name,
|
| 3585 |
"Communication failure in hooks execution: %s", msg)
|
| 3586 |
if res.offline or msg: |
| 3587 |
# No need to investigate payload if node is offline or gave
|
| 3588 |
# an error.
|
| 3589 |
continue
|
| 3590 |
for script, hkr, output in res.payload: |
| 3591 |
test = hkr == constants.HKR_FAIL |
| 3592 |
self._ErrorIf(test, constants.CV_ENODEHOOKS, node_name,
|
| 3593 |
"Script %s failed, output:", script)
|
| 3594 |
if test:
|
| 3595 |
output = self._HOOKS_INDENT_RE.sub(" ", output) |
| 3596 |
feedback_fn("%s" % output)
|
| 3597 |
lu_result = False
|
| 3598 |
|
| 3599 |
return lu_result
|
| 3600 |
|
| 3601 |
|
| 3602 |
class LUClusterVerifyDisks(NoHooksLU): |
| 3603 |
"""Verifies the cluster disks status.
|
| 3604 |
|
| 3605 |
"""
|
| 3606 |
REQ_BGL = False
|
| 3607 |
|
| 3608 |
def ExpandNames(self): |
| 3609 |
self.share_locks = ShareAll()
|
| 3610 |
self.needed_locks = {
|
| 3611 |
locking.LEVEL_NODEGROUP: locking.ALL_SET, |
| 3612 |
} |
| 3613 |
|
| 3614 |
def Exec(self, feedback_fn): |
| 3615 |
group_names = self.owned_locks(locking.LEVEL_NODEGROUP)
|
| 3616 |
|
| 3617 |
# Submit one instance of L{opcodes.OpGroupVerifyDisks} per node group
|
| 3618 |
return ResultWithJobs([[opcodes.OpGroupVerifyDisks(group_name=group)]
|
| 3619 |
for group in group_names]) |