cluster.py 111 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
#
#

# Copyright (C) 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013 Google Inc.
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful, but
# WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
# General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
# 02110-1301, USA.


"""Logical units dealing with the cluster."""

import OpenSSL

import copy
import itertools
import logging
import operator
import os
import re
import time

from ganeti import compat
from ganeti import constants
from ganeti import errors
from ganeti import hypervisor
from ganeti import locking
from ganeti import masterd
from ganeti import netutils
from ganeti import objects
from ganeti import opcodes
from ganeti import pathutils
from ganeti import query
from ganeti import rpc
from ganeti import runtime
from ganeti import ssh
from ganeti import uidpool
from ganeti import utils
from ganeti import vcluster

52
from ganeti.cmdlib.base import NoHooksLU, QueryBase, LogicalUnit, \
53
  ResultWithJobs
54
55
56
57
58
59
from ganeti.cmdlib.common import ShareAll, RunPostHook, \
  ComputeAncillaryFiles, RedistributeAncillaryFiles, UploadHelper, \
  GetWantedInstances, MergeAndVerifyHvState, MergeAndVerifyDiskState, \
  GetUpdatedIPolicy, ComputeNewInstanceViolations, GetUpdatedParams, \
  CheckOSParams, CheckHVParams, AdjustCandidatePool, CheckNodePVs, \
  ComputeIPolicyInstanceViolation, AnnotateDiskParams, SupportsOob
60
61
62
63
64
65
66
67
68
69
70
71
72
73

import ganeti.masterd.instance


class LUClusterActivateMasterIp(NoHooksLU):
  """Activate the master IP on the master node.

  """
  def Exec(self, feedback_fn):
    """Activate the master IP.

    """
    master_params = self.cfg.GetMasterNetworkParameters()
    ems = self.cfg.GetUseExternalMipScript()
Thomas Thrainer's avatar
Thomas Thrainer committed
74
    result = self.rpc.call_node_activate_master_ip(master_params.uuid,
75
76
77
78
79
80
81
82
83
84
85
86
87
88
                                                   master_params, ems)
    result.Raise("Could not activate the master IP")


class LUClusterDeactivateMasterIp(NoHooksLU):
  """Deactivate the master IP on the master node.

  """
  def Exec(self, feedback_fn):
    """Deactivate the master IP.

    """
    master_params = self.cfg.GetMasterNetworkParameters()
    ems = self.cfg.GetUseExternalMipScript()
Thomas Thrainer's avatar
Thomas Thrainer committed
89
    result = self.rpc.call_node_deactivate_master_ip(master_params.uuid,
90
91
92
93
94
95
96
97
98
99
100
                                                     master_params, ems)
    result.Raise("Could not deactivate the master IP")


class LUClusterConfigQuery(NoHooksLU):
  """Return configuration values.

  """
  REQ_BGL = False

  def CheckArguments(self):
101
    self.cq = ClusterQuery(None, self.op.output_fields, False)
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165

  def ExpandNames(self):
    self.cq.ExpandNames(self)

  def DeclareLocks(self, level):
    self.cq.DeclareLocks(self, level)

  def Exec(self, feedback_fn):
    result = self.cq.OldStyleQuery(self)

    assert len(result) == 1

    return result[0]


class LUClusterDestroy(LogicalUnit):
  """Logical unit for destroying the cluster.

  """
  HPATH = "cluster-destroy"
  HTYPE = constants.HTYPE_CLUSTER

  def BuildHooksEnv(self):
    """Build hooks env.

    """
    return {
      "OP_TARGET": self.cfg.GetClusterName(),
      }

  def BuildHooksNodes(self):
    """Build hooks nodes.

    """
    return ([], [])

  def CheckPrereq(self):
    """Check prerequisites.

    This checks whether the cluster is empty.

    Any errors are signaled by raising errors.OpPrereqError.

    """
    master = self.cfg.GetMasterNode()

    nodelist = self.cfg.GetNodeList()
    if len(nodelist) != 1 or nodelist[0] != master:
      raise errors.OpPrereqError("There are still %d node(s) in"
                                 " this cluster." % (len(nodelist) - 1),
                                 errors.ECODE_INVAL)
    instancelist = self.cfg.GetInstanceList()
    if instancelist:
      raise errors.OpPrereqError("There are still %d instance(s) in"
                                 " this cluster." % len(instancelist),
                                 errors.ECODE_INVAL)

  def Exec(self, feedback_fn):
    """Destroys the cluster.

    """
    master_params = self.cfg.GetMasterNetworkParameters()

    # Run post hooks on master node before it's removed
Thomas Thrainer's avatar
Thomas Thrainer committed
166
    RunPostHook(self, self.cfg.GetNodeName(master_params.uuid))
167
168

    ems = self.cfg.GetUseExternalMipScript()
Thomas Thrainer's avatar
Thomas Thrainer committed
169
    result = self.rpc.call_node_deactivate_master_ip(master_params.uuid,
170
                                                     master_params, ems)
171
    result.Warn("Error disabling the master IP address", self.LogWarning)
Thomas Thrainer's avatar
Thomas Thrainer committed
172
    return master_params.uuid
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202


class LUClusterPostInit(LogicalUnit):
  """Logical unit for running hooks after cluster initialization.

  """
  HPATH = "cluster-init"
  HTYPE = constants.HTYPE_CLUSTER

  def BuildHooksEnv(self):
    """Build hooks env.

    """
    return {
      "OP_TARGET": self.cfg.GetClusterName(),
      }

  def BuildHooksNodes(self):
    """Build hooks nodes.

    """
    return ([], [self.cfg.GetMasterNode()])

  def Exec(self, feedback_fn):
    """Nothing to do.

    """
    return True


203
class ClusterQuery(QueryBase):
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
  FIELDS = query.CLUSTER_FIELDS

  #: Do not sort (there is only one item)
  SORT_FIELD = None

  def ExpandNames(self, lu):
    lu.needed_locks = {}

    # The following variables interact with _QueryBase._GetNames
    self.wanted = locking.ALL_SET
    self.do_locking = self.use_locking

    if self.do_locking:
      raise errors.OpPrereqError("Can not use locking for cluster queries",
                                 errors.ECODE_INVAL)

  def DeclareLocks(self, lu, level):
    pass

  def _GetQueryData(self, lu):
    """Computes the list of nodes and their attributes.

    """
    # Locking is not used
    assert not (compat.any(lu.glm.is_owned(level)
                           for level in locking.LEVELS
                           if level != locking.LEVEL_CLUSTER) or
                self.do_locking or self.use_locking)

    if query.CQ_CONFIG in self.requested_data:
      cluster = lu.cfg.GetClusterInfo()
Thomas Thrainer's avatar
Thomas Thrainer committed
235
      nodes = lu.cfg.GetAllNodesInfo()
236
237
    else:
      cluster = NotImplemented
Thomas Thrainer's avatar
Thomas Thrainer committed
238
      nodes = NotImplemented
239
240
241
242
243
244
245

    if query.CQ_QUEUE_DRAINED in self.requested_data:
      drain_flag = os.path.exists(pathutils.JOB_QUEUE_DRAIN_FILE)
    else:
      drain_flag = NotImplemented

    if query.CQ_WATCHER_PAUSE in self.requested_data:
Thomas Thrainer's avatar
Thomas Thrainer committed
246
      master_node_uuid = lu.cfg.GetMasterNode()
247

Thomas Thrainer's avatar
Thomas Thrainer committed
248
      result = lu.rpc.call_get_watcher_pause(master_node_uuid)
249
      result.Raise("Can't retrieve watcher pause from master node '%s'" %
Thomas Thrainer's avatar
Thomas Thrainer committed
250
                   lu.cfg.GetMasterNodeName())
251
252
253
254
255

      watcher_pause = result.payload
    else:
      watcher_pause = NotImplemented

Thomas Thrainer's avatar
Thomas Thrainer committed
256
    return query.ClusterQueryData(cluster, nodes, drain_flag, watcher_pause)
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294


class LUClusterQuery(NoHooksLU):
  """Query cluster configuration.

  """
  REQ_BGL = False

  def ExpandNames(self):
    self.needed_locks = {}

  def Exec(self, feedback_fn):
    """Return cluster config.

    """
    cluster = self.cfg.GetClusterInfo()
    os_hvp = {}

    # Filter just for enabled hypervisors
    for os_name, hv_dict in cluster.os_hvp.items():
      os_hvp[os_name] = {}
      for hv_name, hv_params in hv_dict.items():
        if hv_name in cluster.enabled_hypervisors:
          os_hvp[os_name][hv_name] = hv_params

    # Convert ip_family to ip_version
    primary_ip_version = constants.IP4_VERSION
    if cluster.primary_ip_family == netutils.IP6Address.family:
      primary_ip_version = constants.IP6_VERSION

    result = {
      "software_version": constants.RELEASE_VERSION,
      "protocol_version": constants.PROTOCOL_VERSION,
      "config_version": constants.CONFIG_VERSION,
      "os_api_version": max(constants.OS_API_VERSIONS),
      "export_version": constants.EXPORT_VERSION,
      "architecture": runtime.GetArchInfo(),
      "name": cluster.cluster_name,
Thomas Thrainer's avatar
Thomas Thrainer committed
295
      "master": self.cfg.GetMasterNodeName(),
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
      "default_hypervisor": cluster.primary_hypervisor,
      "enabled_hypervisors": cluster.enabled_hypervisors,
      "hvparams": dict([(hypervisor_name, cluster.hvparams[hypervisor_name])
                        for hypervisor_name in cluster.enabled_hypervisors]),
      "os_hvp": os_hvp,
      "beparams": cluster.beparams,
      "osparams": cluster.osparams,
      "ipolicy": cluster.ipolicy,
      "nicparams": cluster.nicparams,
      "ndparams": cluster.ndparams,
      "diskparams": cluster.diskparams,
      "candidate_pool_size": cluster.candidate_pool_size,
      "master_netdev": cluster.master_netdev,
      "master_netmask": cluster.master_netmask,
      "use_external_mip_script": cluster.use_external_mip_script,
      "volume_group_name": cluster.volume_group_name,
      "drbd_usermode_helper": cluster.drbd_usermode_helper,
      "file_storage_dir": cluster.file_storage_dir,
      "shared_file_storage_dir": cluster.shared_file_storage_dir,
      "maintain_node_health": cluster.maintain_node_health,
      "ctime": cluster.ctime,
      "mtime": cluster.mtime,
      "uuid": cluster.uuid,
      "tags": list(cluster.GetTags()),
      "uid_pool": cluster.uid_pool,
      "default_iallocator": cluster.default_iallocator,
      "reserved_lvs": cluster.reserved_lvs,
      "primary_ip_version": primary_ip_version,
      "prealloc_wipe_disks": cluster.prealloc_wipe_disks,
      "hidden_os": cluster.hidden_os,
      "blacklisted_os": cluster.blacklisted_os,
      }

    return result


class LUClusterRedistConf(NoHooksLU):
  """Force the redistribution of cluster configuration.

  This is a very simple LU.

  """
  REQ_BGL = False

  def ExpandNames(self):
    self.needed_locks = {
      locking.LEVEL_NODE: locking.ALL_SET,
      locking.LEVEL_NODE_ALLOC: locking.ALL_SET,
    }
345
    self.share_locks = ShareAll()
346
347
348
349
350
351

  def Exec(self, feedback_fn):
    """Redistribute the configuration.

    """
    self.cfg.Update(self.cfg.GetClusterInfo(), feedback_fn)
352
    RedistributeAncillaryFiles(self)
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409


class LUClusterRename(LogicalUnit):
  """Rename the cluster.

  """
  HPATH = "cluster-rename"
  HTYPE = constants.HTYPE_CLUSTER

  def BuildHooksEnv(self):
    """Build hooks env.

    """
    return {
      "OP_TARGET": self.cfg.GetClusterName(),
      "NEW_NAME": self.op.name,
      }

  def BuildHooksNodes(self):
    """Build hooks nodes.

    """
    return ([self.cfg.GetMasterNode()], self.cfg.GetNodeList())

  def CheckPrereq(self):
    """Verify that the passed name is a valid one.

    """
    hostname = netutils.GetHostname(name=self.op.name,
                                    family=self.cfg.GetPrimaryIPFamily())

    new_name = hostname.name
    self.ip = new_ip = hostname.ip
    old_name = self.cfg.GetClusterName()
    old_ip = self.cfg.GetMasterIP()
    if new_name == old_name and new_ip == old_ip:
      raise errors.OpPrereqError("Neither the name nor the IP address of the"
                                 " cluster has changed",
                                 errors.ECODE_INVAL)
    if new_ip != old_ip:
      if netutils.TcpPing(new_ip, constants.DEFAULT_NODED_PORT):
        raise errors.OpPrereqError("The given cluster IP address (%s) is"
                                   " reachable on the network" %
                                   new_ip, errors.ECODE_NOTUNIQUE)

    self.op.name = new_name

  def Exec(self, feedback_fn):
    """Rename the cluster.

    """
    clustername = self.op.name
    new_ip = self.ip

    # shutdown the master IP
    master_params = self.cfg.GetMasterNetworkParameters()
    ems = self.cfg.GetUseExternalMipScript()
Thomas Thrainer's avatar
Thomas Thrainer committed
410
    result = self.rpc.call_node_deactivate_master_ip(master_params.uuid,
411
412
413
414
415
416
417
418
419
420
421
422
423
                                                     master_params, ems)
    result.Raise("Could not disable the master role")

    try:
      cluster = self.cfg.GetClusterInfo()
      cluster.cluster_name = clustername
      cluster.master_ip = new_ip
      self.cfg.Update(cluster, feedback_fn)

      # update the known hosts file
      ssh.WriteKnownHostsFile(self.cfg, pathutils.SSH_KNOWN_HOSTS_FILE)
      node_list = self.cfg.GetOnlineNodeList()
      try:
Thomas Thrainer's avatar
Thomas Thrainer committed
424
        node_list.remove(master_params.uuid)
425
426
      except ValueError:
        pass
427
      UploadHelper(self, node_list, pathutils.SSH_KNOWN_HOSTS_FILE)
428
429
    finally:
      master_params.ip = new_ip
Thomas Thrainer's avatar
Thomas Thrainer committed
430
      result = self.rpc.call_node_activate_master_ip(master_params.uuid,
431
                                                     master_params, ems)
432
433
      result.Warn("Could not re-enable the master role on the master,"
                  " please restart manually", self.LogWarning)
434
435
436
437
438
439
440
441
442
443
444
445

    return clustername


class LUClusterRepairDiskSizes(NoHooksLU):
  """Verifies the cluster disks sizes.

  """
  REQ_BGL = False

  def ExpandNames(self):
    if self.op.instances:
446
      self.wanted_names = GetWantedInstances(self, self.op.instances)
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
      # Not getting the node allocation lock as only a specific set of
      # instances (and their nodes) is going to be acquired
      self.needed_locks = {
        locking.LEVEL_NODE_RES: [],
        locking.LEVEL_INSTANCE: self.wanted_names,
        }
      self.recalculate_locks[locking.LEVEL_NODE_RES] = constants.LOCKS_REPLACE
    else:
      self.wanted_names = None
      self.needed_locks = {
        locking.LEVEL_NODE_RES: locking.ALL_SET,
        locking.LEVEL_INSTANCE: locking.ALL_SET,

        # This opcode is acquires the node locks for all instances
        locking.LEVEL_NODE_ALLOC: locking.ALL_SET,
        }

    self.share_locks = {
      locking.LEVEL_NODE_RES: 1,
      locking.LEVEL_INSTANCE: 0,
      locking.LEVEL_NODE_ALLOC: 1,
      }

  def DeclareLocks(self, level):
    if level == locking.LEVEL_NODE_RES and self.wanted_names is not None:
      self._LockInstancesNodes(primary_only=True, level=level)

  def CheckPrereq(self):
    """Check prerequisites.

    This only checks the optional instance list against the existing names.

    """
    if self.wanted_names is None:
      self.wanted_names = self.owned_locks(locking.LEVEL_INSTANCE)

    self.wanted_instances = \
        map(compat.snd, self.cfg.GetMultiInstanceInfo(self.wanted_names))

  def _EnsureChildSizes(self, disk):
    """Ensure children of the disk have the needed disk size.

    This is valid mainly for DRBD8 and fixes an issue where the
    children have smaller disk size.

    @param disk: an L{ganeti.objects.Disk} object

    """
    if disk.dev_type == constants.LD_DRBD8:
      assert disk.children, "Empty children for DRBD8?"
      fchild = disk.children[0]
      mismatch = fchild.size < disk.size
      if mismatch:
        self.LogInfo("Child disk has size %d, parent %d, fixing",
                     fchild.size, disk.size)
        fchild.size = disk.size

      # and we recurse on this child only, not on the metadev
      return self._EnsureChildSizes(fchild) or mismatch
    else:
      return False

  def Exec(self, feedback_fn):
    """Verify the size of cluster disks.

    """
    # TODO: check child disks too
    # TODO: check differences in size between primary/secondary nodes
    per_node_disks = {}
    for instance in self.wanted_instances:
      pnode = instance.primary_node
      if pnode not in per_node_disks:
        per_node_disks[pnode] = []
      for idx, disk in enumerate(instance.disks):
        per_node_disks[pnode].append((instance, idx, disk))

    assert not (frozenset(per_node_disks.keys()) -
                self.owned_locks(locking.LEVEL_NODE_RES)), \
      "Not owning correct locks"
    assert not self.owned_locks(locking.LEVEL_NODE)

Thomas Thrainer's avatar
Thomas Thrainer committed
528
529
    es_flags = rpc.GetExclusiveStorageForNodes(self.cfg,
                                               per_node_disks.keys())
530

531
    changed = []
Thomas Thrainer's avatar
Thomas Thrainer committed
532
    for node_uuid, dskl in per_node_disks.items():
533
534
      newl = [v[2].Copy() for v in dskl]
      for dsk in newl:
Thomas Thrainer's avatar
Thomas Thrainer committed
535
536
537
        self.cfg.SetDiskID(dsk, node_uuid)
      node_name = self.cfg.GetNodeName(node_uuid)
      result = self.rpc.call_blockdev_getdimensions(node_uuid, newl)
538
      if result.fail_msg:
539
        self.LogWarning("Failure in blockdev_getdimensions call to node"
Thomas Thrainer's avatar
Thomas Thrainer committed
540
                        " %s, ignoring", node_name)
541
542
543
        continue
      if len(result.payload) != len(dskl):
        logging.warning("Invalid result from node %s: len(dksl)=%d,"
Thomas Thrainer's avatar
Thomas Thrainer committed
544
545
                        " result.payload=%s", node_name, len(dskl),
                        result.payload)
546
        self.LogWarning("Invalid result from node %s, ignoring node results",
Thomas Thrainer's avatar
Thomas Thrainer committed
547
                        node_name)
548
        continue
549
550
      for ((instance, idx, disk), dimensions) in zip(dskl, result.payload):
        if dimensions is None:
551
552
553
          self.LogWarning("Disk %d of instance %s did not return size"
                          " information, ignoring", idx, instance.name)
          continue
554
555
556
557
558
        if not isinstance(dimensions, (tuple, list)):
          self.LogWarning("Disk %d of instance %s did not return valid"
                          " dimension information, ignoring", idx,
                          instance.name)
          continue
559
        (size, spindles) = dimensions
560
561
562
563
564
565
566
567
568
569
570
        if not isinstance(size, (int, long)):
          self.LogWarning("Disk %d of instance %s did not return valid"
                          " size information, ignoring", idx, instance.name)
          continue
        size = size >> 20
        if size != disk.size:
          self.LogInfo("Disk %d of instance %s has mismatched size,"
                       " correcting: recorded %d, actual %d", idx,
                       instance.name, disk.size, size)
          disk.size = size
          self.cfg.Update(instance, feedback_fn)
571
          changed.append((instance.name, idx, "size", size))
Thomas Thrainer's avatar
Thomas Thrainer committed
572
        if es_flags[node_uuid]:
573
574
575
576
577
578
579
580
581
582
583
          if spindles is None:
            self.LogWarning("Disk %d of instance %s did not return valid"
                            " spindles information, ignoring", idx,
                            instance.name)
          elif disk.spindles is None or disk.spindles != spindles:
            self.LogInfo("Disk %d of instance %s has mismatched spindles,"
                         " correcting: recorded %s, actual %s",
                         idx, instance.name, disk.spindles, spindles)
            disk.spindles = spindles
            self.cfg.Update(instance, feedback_fn)
            changed.append((instance.name, idx, "spindles", disk.spindles))
584
585
        if self._EnsureChildSizes(disk):
          self.cfg.Update(instance, feedback_fn)
586
          changed.append((instance.name, idx, "size", disk.size))
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
    return changed


def _ValidateNetmask(cfg, netmask):
  """Checks if a netmask is valid.

  @type cfg: L{config.ConfigWriter}
  @param cfg: The cluster configuration
  @type netmask: int
  @param netmask: the netmask to be verified
  @raise errors.OpPrereqError: if the validation fails

  """
  ip_family = cfg.GetPrimaryIPFamily()
  try:
    ipcls = netutils.IPAddress.GetClassFromIpFamily(ip_family)
  except errors.ProgrammerError:
    raise errors.OpPrereqError("Invalid primary ip family: %s." %
                               ip_family, errors.ECODE_INVAL)
  if not ipcls.ValidateNetmask(netmask):
    raise errors.OpPrereqError("CIDR netmask (%s) not valid" %
                               (netmask), errors.ECODE_INVAL)


class LUClusterSetParams(LogicalUnit):
  """Change the parameters of the cluster.

  """
  HPATH = "cluster-modify"
  HTYPE = constants.HTYPE_CLUSTER
  REQ_BGL = False

  def CheckArguments(self):
    """Check parameters

    """
    if self.op.uid_pool:
      uidpool.CheckUidPool(self.op.uid_pool)

    if self.op.add_uids:
      uidpool.CheckUidPool(self.op.add_uids)

    if self.op.remove_uids:
      uidpool.CheckUidPool(self.op.remove_uids)

    if self.op.master_netmask is not None:
      _ValidateNetmask(self.cfg, self.op.master_netmask)

    if self.op.diskparams:
      for dt_params in self.op.diskparams.values():
        utils.ForceDictType(dt_params, constants.DISK_DT_TYPES)
      try:
        utils.VerifyDictOptions(self.op.diskparams, constants.DISK_DT_DEFAULTS)
      except errors.OpPrereqError, err:
        raise errors.OpPrereqError("While verify diskparams options: %s" % err,
                                   errors.ECODE_INVAL)

  def ExpandNames(self):
    # FIXME: in the future maybe other cluster params won't require checking on
    # all nodes to be modified.
    # FIXME: This opcode changes cluster-wide settings. Is acquiring all
    # resource locks the right thing, shouldn't it be the BGL instead?
    self.needed_locks = {
      locking.LEVEL_NODE: locking.ALL_SET,
      locking.LEVEL_INSTANCE: locking.ALL_SET,
      locking.LEVEL_NODEGROUP: locking.ALL_SET,
      locking.LEVEL_NODE_ALLOC: locking.ALL_SET,
    }
655
    self.share_locks = ShareAll()
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672

  def BuildHooksEnv(self):
    """Build hooks env.

    """
    return {
      "OP_TARGET": self.cfg.GetClusterName(),
      "NEW_VG_NAME": self.op.vg_name,
      }

  def BuildHooksNodes(self):
    """Build hooks nodes.

    """
    mn = self.cfg.GetMasterNode()
    return ([mn], [mn])

Thomas Thrainer's avatar
Thomas Thrainer committed
673
  def _CheckVgName(self, node_uuids, enabled_disk_templates,
674
675
676
                   new_enabled_disk_templates):
    """Check the consistency of the vg name on all nodes and in case it gets
       unset whether there are instances still using it.
677
678
679
680
681
682
683

    """
    if self.op.vg_name is not None and not self.op.vg_name:
      if self.cfg.HasAnyDiskOfType(constants.LD_LV):
        raise errors.OpPrereqError("Cannot disable lvm storage while lvm-based"
                                   " instances exist", errors.ECODE_INVAL)

684
685
686
687
688
    if (self.op.vg_name is not None and
        utils.IsLvmEnabled(enabled_disk_templates)) or \
           (self.cfg.GetVGName() is not None and
            utils.LvmGetsEnabled(enabled_disk_templates,
                                 new_enabled_disk_templates)):
Thomas Thrainer's avatar
Thomas Thrainer committed
689
      self._CheckVgNameOnNodes(node_uuids)
690

Thomas Thrainer's avatar
Thomas Thrainer committed
691
  def _CheckVgNameOnNodes(self, node_uuids):
692
693
694
    """Check the status of the volume group on each node.

    """
Thomas Thrainer's avatar
Thomas Thrainer committed
695
696
697
    vglist = self.rpc.call_vg_list(node_uuids)
    for node_uuid in node_uuids:
      msg = vglist[node_uuid].fail_msg
698
699
700
      if msg:
        # ignoring down node
        self.LogWarning("Error while gathering data on node %s"
Thomas Thrainer's avatar
Thomas Thrainer committed
701
702
                        " (ignoring node): %s",
                        self.cfg.GetNodeName(node_uuid), msg)
703
        continue
Thomas Thrainer's avatar
Thomas Thrainer committed
704
      vgstatus = utils.CheckVolumeGroupSize(vglist[node_uuid].payload,
705
706
707
708
                                            self.op.vg_name,
                                            constants.MIN_VG_SIZE)
      if vgstatus:
        raise errors.OpPrereqError("Error on node '%s': %s" %
Thomas Thrainer's avatar
Thomas Thrainer committed
709
710
                                   (self.cfg.GetNodeName(node_uuid), vgstatus),
                                   errors.ECODE_ENVIRON)
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734

  def _GetEnabledDiskTemplates(self, cluster):
    """Determines the enabled disk templates and the subset of disk templates
       that are newly enabled by this operation.

    """
    enabled_disk_templates = None
    new_enabled_disk_templates = []
    if self.op.enabled_disk_templates:
      enabled_disk_templates = self.op.enabled_disk_templates
      new_enabled_disk_templates = \
        list(set(enabled_disk_templates)
             - set(cluster.enabled_disk_templates))
    else:
      enabled_disk_templates = cluster.enabled_disk_templates
    return (enabled_disk_templates, new_enabled_disk_templates)

  def CheckPrereq(self):
    """Check prerequisites.

    This checks whether the given params don't conflict and
    if the given volume group is valid.

    """
735
736
737
738
739
740
    if self.op.drbd_helper is not None and not self.op.drbd_helper:
      if self.cfg.HasAnyDiskOfType(constants.LD_DRBD8):
        raise errors.OpPrereqError("Cannot disable drbd helper while"
                                   " drbd-based instances exist",
                                   errors.ECODE_INVAL)

Thomas Thrainer's avatar
Thomas Thrainer committed
741
    node_uuids = self.owned_locks(locking.LEVEL_NODE)
742
    self.cluster = cluster = self.cfg.GetClusterInfo()
743

Thomas Thrainer's avatar
Thomas Thrainer committed
744
745
746
    vm_capable_node_uuids = [node.uuid
                             for node in self.cfg.GetAllNodesInfo().values()
                             if node.uuid in node_uuids and node.vm_capable]
747

748
749
750
    (enabled_disk_templates, new_enabled_disk_templates) = \
      self._GetEnabledDiskTemplates(cluster)

Thomas Thrainer's avatar
Thomas Thrainer committed
751
    self._CheckVgName(vm_capable_node_uuids, enabled_disk_templates,
752
                      new_enabled_disk_templates)
753
754
755

    if self.op.drbd_helper:
      # checks given drbd helper on all nodes
Thomas Thrainer's avatar
Thomas Thrainer committed
756
757
      helpers = self.rpc.call_drbd_helper(node_uuids)
      for (_, ninfo) in self.cfg.GetMultiNodeInfo(node_uuids):
758
        if ninfo.offline:
Thomas Thrainer's avatar
Thomas Thrainer committed
759
760
          self.LogInfo("Not checking drbd helper on offline node %s",
                       ninfo.name)
761
          continue
Thomas Thrainer's avatar
Thomas Thrainer committed
762
        msg = helpers[ninfo.uuid].fail_msg
763
764
        if msg:
          raise errors.OpPrereqError("Error checking drbd helper on node"
Thomas Thrainer's avatar
Thomas Thrainer committed
765
                                     " '%s': %s" % (ninfo.name, msg),
766
                                     errors.ECODE_ENVIRON)
Thomas Thrainer's avatar
Thomas Thrainer committed
767
        node_helper = helpers[ninfo.uuid].payload
768
769
        if node_helper != self.op.drbd_helper:
          raise errors.OpPrereqError("Error on node '%s': drbd helper is %s" %
Thomas Thrainer's avatar
Thomas Thrainer committed
770
771
                                     (ninfo.name, node_helper),
                                     errors.ECODE_ENVIRON)
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789

    # validate params changes
    if self.op.beparams:
      objects.UpgradeBeParams(self.op.beparams)
      utils.ForceDictType(self.op.beparams, constants.BES_PARAMETER_TYPES)
      self.new_beparams = cluster.SimpleFillBE(self.op.beparams)

    if self.op.ndparams:
      utils.ForceDictType(self.op.ndparams, constants.NDS_PARAMETER_TYPES)
      self.new_ndparams = cluster.SimpleFillND(self.op.ndparams)

      # TODO: we need a more general way to handle resetting
      # cluster-level parameters to default values
      if self.new_ndparams["oob_program"] == "":
        self.new_ndparams["oob_program"] = \
            constants.NDC_DEFAULTS[constants.ND_OOB_PROGRAM]

    if self.op.hv_state:
790
791
      new_hv_state = MergeAndVerifyHvState(self.op.hv_state,
                                           self.cluster.hv_state_static)
792
793
794
795
      self.new_hv_state = dict((hv, cluster.SimpleFillHvState(values))
                               for hv, values in new_hv_state.items())

    if self.op.disk_state:
796
797
      new_disk_state = MergeAndVerifyDiskState(self.op.disk_state,
                                               self.cluster.disk_state_static)
798
799
800
801
802
803
      self.new_disk_state = \
        dict((storage, dict((name, cluster.SimpleFillDiskState(values))
                            for name, values in svalues.items()))
             for storage, svalues in new_disk_state.items())

    if self.op.ipolicy:
804
805
      self.new_ipolicy = GetUpdatedIPolicy(cluster.ipolicy, self.op.ipolicy,
                                           group_policy=False)
806
807
808
809
810

      all_instances = self.cfg.GetAllInstancesInfo().values()
      violations = set()
      for group in self.cfg.GetAllNodeGroupsInfo().values():
        instances = frozenset([inst for inst in all_instances
Thomas Thrainer's avatar
Thomas Thrainer committed
811
812
                               if compat.any(nuuid in group.members
                                             for nuuid in inst.all_nodes)])
813
814
        new_ipolicy = objects.FillIPolicy(self.new_ipolicy, group.ipolicy)
        ipol = masterd.instance.CalculateGroupIPolicy(cluster, group)
815
816
        new = ComputeNewInstanceViolations(ipol,
                                           new_ipolicy, instances, self.cfg)
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
        if new:
          violations.update(new)

      if violations:
        self.LogWarning("After the ipolicy change the following instances"
                        " violate them: %s",
                        utils.CommaJoin(utils.NiceSort(violations)))

    if self.op.nicparams:
      utils.ForceDictType(self.op.nicparams, constants.NICS_PARAMETER_TYPES)
      self.new_nicparams = cluster.SimpleFillNIC(self.op.nicparams)
      objects.NIC.CheckParameterSyntax(self.new_nicparams)
      nic_errors = []

      # check all instances for consistency
      for instance in self.cfg.GetAllInstancesInfo().values():
        for nic_idx, nic in enumerate(instance.nics):
          params_copy = copy.deepcopy(nic.nicparams)
          params_filled = objects.FillDict(self.new_nicparams, params_copy)

          # check parameter syntax
          try:
            objects.NIC.CheckParameterSyntax(params_filled)
          except errors.ConfigurationError, err:
            nic_errors.append("Instance %s, nic/%d: %s" %
                              (instance.name, nic_idx, err))

          # if we're moving instances to routed, check that they have an ip
          target_mode = params_filled[constants.NIC_MODE]
          if target_mode == constants.NIC_MODE_ROUTED and not nic.ip:
            nic_errors.append("Instance %s, nic/%d: routed NIC with no ip"
                              " address" % (instance.name, nic_idx))
      if nic_errors:
        raise errors.OpPrereqError("Cannot apply the change, errors:\n%s" %
                                   "\n".join(nic_errors), errors.ECODE_INVAL)

    # hypervisor list/parameters
    self.new_hvparams = new_hvp = objects.FillDict(cluster.hvparams, {})
    if self.op.hvparams:
      for hv_name, hv_dict in self.op.hvparams.items():
        if hv_name not in self.new_hvparams:
          self.new_hvparams[hv_name] = hv_dict
        else:
          self.new_hvparams[hv_name].update(hv_dict)

    # disk template parameters
    self.new_diskparams = objects.FillDict(cluster.diskparams, {})
    if self.op.diskparams:
      for dt_name, dt_params in self.op.diskparams.items():
        if dt_name not in self.op.diskparams:
          self.new_diskparams[dt_name] = dt_params
        else:
          self.new_diskparams[dt_name].update(dt_params)

    # os hypervisor parameters
    self.new_os_hvp = objects.FillDict(cluster.os_hvp, {})
    if self.op.os_hvp:
      for os_name, hvs in self.op.os_hvp.items():
        if os_name not in self.new_os_hvp:
          self.new_os_hvp[os_name] = hvs
        else:
          for hv_name, hv_dict in hvs.items():
            if hv_dict is None:
              # Delete if it exists
              self.new_os_hvp[os_name].pop(hv_name, None)
            elif hv_name not in self.new_os_hvp[os_name]:
              self.new_os_hvp[os_name][hv_name] = hv_dict
            else:
              self.new_os_hvp[os_name][hv_name].update(hv_dict)

    # os parameters
    self.new_osp = objects.FillDict(cluster.osparams, {})
    if self.op.osparams:
      for os_name, osp in self.op.osparams.items():
        if os_name not in self.new_osp:
          self.new_osp[os_name] = {}

894
895
        self.new_osp[os_name] = GetUpdatedParams(self.new_osp[os_name], osp,
                                                 use_none=True)
896
897
898
899
900
901

        if not self.new_osp[os_name]:
          # we removed all parameters
          del self.new_osp[os_name]
        else:
          # check the parameter validity (remote check)
902
903
          CheckOSParams(self, False, [self.cfg.GetMasterNode()],
                        os_name, self.new_osp[os_name])
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930

    # changes to the hypervisor list
    if self.op.enabled_hypervisors is not None:
      self.hv_list = self.op.enabled_hypervisors
      for hv in self.hv_list:
        # if the hypervisor doesn't already exist in the cluster
        # hvparams, we initialize it to empty, and then (in both
        # cases) we make sure to fill the defaults, as we might not
        # have a complete defaults list if the hypervisor wasn't
        # enabled before
        if hv not in new_hvp:
          new_hvp[hv] = {}
        new_hvp[hv] = objects.FillDict(constants.HVC_DEFAULTS[hv], new_hvp[hv])
        utils.ForceDictType(new_hvp[hv], constants.HVS_PARAMETER_TYPES)
    else:
      self.hv_list = cluster.enabled_hypervisors

    if self.op.hvparams or self.op.enabled_hypervisors is not None:
      # either the enabled list has changed, or the parameters have, validate
      for hv_name, hv_params in self.new_hvparams.items():
        if ((self.op.hvparams and hv_name in self.op.hvparams) or
            (self.op.enabled_hypervisors and
             hv_name in self.op.enabled_hypervisors)):
          # either this is a new hypervisor, or its parameters have changed
          hv_class = hypervisor.GetHypervisorClass(hv_name)
          utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES)
          hv_class.CheckParameterSyntax(hv_params)
Thomas Thrainer's avatar
Thomas Thrainer committed
931
          CheckHVParams(self, node_uuids, hv_name, hv_params)
932
933
934
935
936
937
938
939
940
941
942
943
944
945

    self._CheckDiskTemplateConsistency()

    if self.op.os_hvp:
      # no need to check any newly-enabled hypervisors, since the
      # defaults have already been checked in the above code-block
      for os_name, os_hvp in self.new_os_hvp.items():
        for hv_name, hv_params in os_hvp.items():
          utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES)
          # we need to fill in the new os_hvp on top of the actual hv_p
          cluster_defaults = self.new_hvparams.get(hv_name, {})
          new_osp = objects.FillDict(cluster_defaults, hv_params)
          hv_class = hypervisor.GetHypervisorClass(hv_name)
          hv_class.CheckParameterSyntax(new_osp)
Thomas Thrainer's avatar
Thomas Thrainer committed
946
          CheckHVParams(self, node_uuids, hv_name, new_osp)
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973

    if self.op.default_iallocator:
      alloc_script = utils.FindFile(self.op.default_iallocator,
                                    constants.IALLOCATOR_SEARCH_PATH,
                                    os.path.isfile)
      if alloc_script is None:
        raise errors.OpPrereqError("Invalid default iallocator script '%s'"
                                   " specified" % self.op.default_iallocator,
                                   errors.ECODE_INVAL)

  def _CheckDiskTemplateConsistency(self):
    """Check whether the disk templates that are going to be disabled
       are still in use by some instances.

    """
    if self.op.enabled_disk_templates:
      cluster = self.cfg.GetClusterInfo()
      instances = self.cfg.GetAllInstancesInfo()

      disk_templates_to_remove = set(cluster.enabled_disk_templates) \
        - set(self.op.enabled_disk_templates)
      for instance in instances.itervalues():
        if instance.disk_template in disk_templates_to_remove:
          raise errors.OpPrereqError("Cannot disable disk template '%s',"
                                     " because instance '%s' is using it." %
                                     (instance.disk_template, instance.name))

974
975
  def _SetVgName(self, feedback_fn):
    """Determines and sets the new volume group name.
976
977
978

    """
    if self.op.vg_name is not None:
979
980
981
982
      if self.op.vg_name and not \
           utils.IsLvmEnabled(self.cluster.enabled_disk_templates):
        feedback_fn("Note that you specified a volume group, but did not"
                    " enable any lvm disk template.")
983
984
      new_volume = self.op.vg_name
      if not new_volume:
985
986
987
        if utils.IsLvmEnabled(self.cluster.enabled_disk_templates):
          raise errors.OpPrereqError("Cannot unset volume group if lvm-based"
                                     " disk templates are enabled.")
988
989
990
991
992
993
        new_volume = None
      if new_volume != self.cfg.GetVGName():
        self.cfg.SetVGName(new_volume)
      else:
        feedback_fn("Cluster LVM configuration already in desired"
                    " state, not changing")
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
    else:
      if utils.IsLvmEnabled(self.cluster.enabled_disk_templates) and \
          not self.cfg.GetVGName():
        raise errors.OpPrereqError("Please specify a volume group when"
                                   " enabling lvm-based disk-templates.")

  def Exec(self, feedback_fn):
    """Change the parameters of the cluster.

    """
    if self.op.enabled_disk_templates:
      self.cluster.enabled_disk_templates = \
        list(set(self.op.enabled_disk_templates))

    self._SetVgName(feedback_fn)

1010
    if self.op.drbd_helper is not None:
1011
1012
1013
      if not constants.DT_DRBD8 in self.cluster.enabled_disk_templates:
        feedback_fn("Note that you specified a drbd user helper, but did"
                    " enabled the drbd disk template.")
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
      new_helper = self.op.drbd_helper
      if not new_helper:
        new_helper = None
      if new_helper != self.cfg.GetDRBDHelper():
        self.cfg.SetDRBDHelper(new_helper)
      else:
        feedback_fn("Cluster DRBD helper already in desired state,"
                    " not changing")
    if self.op.hvparams:
      self.cluster.hvparams = self.new_hvparams
    if self.op.os_hvp:
      self.cluster.os_hvp = self.new_os_hvp
    if self.op.enabled_hypervisors is not None:
      self.cluster.hvparams = self.new_hvparams
      self.cluster.enabled_hypervisors = self.op.enabled_hypervisors
    if self.op.beparams:
      self.cluster.beparams[constants.PP_DEFAULT] = self.new_beparams
    if self.op.nicparams:
      self.cluster.nicparams[constants.PP_DEFAULT] = self.new_nicparams
    if self.op.ipolicy:
      self.cluster.ipolicy = self.new_ipolicy
    if self.op.osparams:
      self.cluster.osparams = self.new_osp
    if self.op.ndparams:
      self.cluster.ndparams = self.new_ndparams
    if self.op.diskparams:
      self.cluster.diskparams = self.new_diskparams
    if self.op.hv_state:
      self.cluster.hv_state_static = self.new_hv_state
    if self.op.disk_state:
      self.cluster.disk_state_static = self.new_disk_state

    if self.op.candidate_pool_size is not None:
      self.cluster.candidate_pool_size = self.op.candidate_pool_size
      # we need to update the pool size here, otherwise the save will fail
1049
      AdjustCandidatePool(self, [])
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105

    if self.op.maintain_node_health is not None:
      if self.op.maintain_node_health and not constants.ENABLE_CONFD:
        feedback_fn("Note: CONFD was disabled at build time, node health"
                    " maintenance is not useful (still enabling it)")
      self.cluster.maintain_node_health = self.op.maintain_node_health

    if self.op.prealloc_wipe_disks is not None:
      self.cluster.prealloc_wipe_disks = self.op.prealloc_wipe_disks

    if self.op.add_uids is not None:
      uidpool.AddToUidPool(self.cluster.uid_pool, self.op.add_uids)

    if self.op.remove_uids is not None:
      uidpool.RemoveFromUidPool(self.cluster.uid_pool, self.op.remove_uids)

    if self.op.uid_pool is not None:
      self.cluster.uid_pool = self.op.uid_pool

    if self.op.default_iallocator is not None:
      self.cluster.default_iallocator = self.op.default_iallocator

    if self.op.reserved_lvs is not None:
      self.cluster.reserved_lvs = self.op.reserved_lvs

    if self.op.use_external_mip_script is not None:
      self.cluster.use_external_mip_script = self.op.use_external_mip_script

    def helper_os(aname, mods, desc):
      desc += " OS list"
      lst = getattr(self.cluster, aname)
      for key, val in mods:
        if key == constants.DDM_ADD:
          if val in lst:
            feedback_fn("OS %s already in %s, ignoring" % (val, desc))
          else:
            lst.append(val)
        elif key == constants.DDM_REMOVE:
          if val in lst:
            lst.remove(val)
          else:
            feedback_fn("OS %s not found in %s, ignoring" % (val, desc))
        else:
          raise errors.ProgrammerError("Invalid modification '%s'" % key)

    if self.op.hidden_os:
      helper_os("hidden_os", self.op.hidden_os, "hidden")

    if self.op.blacklisted_os:
      helper_os("blacklisted_os", self.op.blacklisted_os, "blacklisted")

    if self.op.master_netdev:
      master_params = self.cfg.GetMasterNetworkParameters()
      ems = self.cfg.GetUseExternalMipScript()
      feedback_fn("Shutting down master ip on the current netdev (%s)" %
                  self.cluster.master_netdev)
Thomas Thrainer's avatar
Thomas Thrainer committed
1106
      result = self.rpc.call_node_deactivate_master_ip(master_params.uuid,
1107
                                                       master_params, ems)
1108
1109
1110
1111
1112
1113
1114
      if not self.op.force:
        result.Raise("Could not disable the master ip")
      else:
        if result.fail_msg:
          msg = ("Could not disable the master ip (continuing anyway): %s" %
                 result.fail_msg)
          feedback_fn(msg)
1115
1116
1117
1118
1119
1120
1121
      feedback_fn("Changing master_netdev from %s to %s" %
                  (master_params.netdev, self.op.master_netdev))
      self.cluster.master_netdev = self.op.master_netdev

    if self.op.master_netmask:
      master_params = self.cfg.GetMasterNetworkParameters()
      feedback_fn("Changing master IP netmask to %s" % self.op.master_netmask)
Thomas Thrainer's avatar
Thomas Thrainer committed
1122
1123
1124
1125
      result = self.rpc.call_node_change_master_netmask(
                 master_params.uuid, master_params.netmask,
                 self.op.master_netmask, master_params.ip,
                 master_params.netdev)
1126
      result.Warn("Could not change the master IP netmask", feedback_fn)
1127
1128
1129
1130
1131
1132
1133
1134
1135
      self.cluster.master_netmask = self.op.master_netmask

    self.cfg.Update(self.cluster, feedback_fn)

    if self.op.master_netdev:
      master_params = self.cfg.GetMasterNetworkParameters()
      feedback_fn("Starting the master ip on the new master netdev (%s)" %
                  self.op.master_netdev)
      ems = self.cfg.GetUseExternalMipScript()
Thomas Thrainer's avatar
Thomas Thrainer committed
1136
      result = self.rpc.call_node_activate_master_ip(master_params.uuid,
1137
                                                     master_params, ems)
1138
1139
      result.Warn("Could not re-enable the master ip on the master,"
                  " please restart manually", self.LogWarning)
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327


class LUClusterVerify(NoHooksLU):
  """Submits all jobs necessary to verify the cluster.

  """
  REQ_BGL = False

  def ExpandNames(self):
    self.needed_locks = {}

  def Exec(self, feedback_fn):
    jobs = []

    if self.op.group_name:
      groups = [self.op.group_name]
      depends_fn = lambda: None
    else:
      groups = self.cfg.GetNodeGroupList()

      # Verify global configuration
      jobs.append([
        opcodes.OpClusterVerifyConfig(ignore_errors=self.op.ignore_errors),
        ])

      # Always depend on global verification
      depends_fn = lambda: [(-len(jobs), [])]

    jobs.extend(
      [opcodes.OpClusterVerifyGroup(group_name=group,
                                    ignore_errors=self.op.ignore_errors,
                                    depends=depends_fn())]
      for group in groups)

    # Fix up all parameters
    for op in itertools.chain(*jobs): # pylint: disable=W0142
      op.debug_simulate_errors = self.op.debug_simulate_errors
      op.verbose = self.op.verbose
      op.error_codes = self.op.error_codes
      try:
        op.skip_checks = self.op.skip_checks
      except AttributeError:
        assert not isinstance(op, opcodes.OpClusterVerifyGroup)

    return ResultWithJobs(jobs)


class _VerifyErrors(object):
  """Mix-in for cluster/group verify LUs.

  It provides _Error and _ErrorIf, and updates the self.bad boolean. (Expects
  self.op and self._feedback_fn to be available.)

  """

  ETYPE_FIELD = "code"
  ETYPE_ERROR = "ERROR"
  ETYPE_WARNING = "WARNING"

  def _Error(self, ecode, item, msg, *args, **kwargs):
    """Format an error message.

    Based on the opcode's error_codes parameter, either format a
    parseable error code, or a simpler error string.

    This must be called only from Exec and functions called from Exec.

    """
    ltype = kwargs.get(self.ETYPE_FIELD, self.ETYPE_ERROR)
    itype, etxt, _ = ecode
    # If the error code is in the list of ignored errors, demote the error to a
    # warning
    if etxt in self.op.ignore_errors:     # pylint: disable=E1101
      ltype = self.ETYPE_WARNING
    # first complete the msg
    if args:
      msg = msg % args
    # then format the whole message
    if self.op.error_codes: # This is a mix-in. pylint: disable=E1101
      msg = "%s:%s:%s:%s:%s" % (ltype, etxt, itype, item, msg)
    else:
      if item:
        item = " " + item
      else:
        item = ""
      msg = "%s: %s%s: %s" % (ltype, itype, item, msg)
    # and finally report it via the feedback_fn
    self._feedback_fn("  - %s" % msg) # Mix-in. pylint: disable=E1101
    # do not mark the operation as failed for WARN cases only
    if ltype == self.ETYPE_ERROR:
      self.bad = True

  def _ErrorIf(self, cond, *args, **kwargs):
    """Log an error message if the passed condition is True.

    """
    if (bool(cond)
        or self.op.debug_simulate_errors): # pylint: disable=E1101
      self._Error(*args, **kwargs)


def _VerifyCertificate(filename):
  """Verifies a certificate for L{LUClusterVerifyConfig}.

  @type filename: string
  @param filename: Path to PEM file

  """
  try:
    cert = OpenSSL.crypto.load_certificate(OpenSSL.crypto.FILETYPE_PEM,
                                           utils.ReadFile(filename))
  except Exception, err: # pylint: disable=W0703
    return (LUClusterVerifyConfig.ETYPE_ERROR,
            "Failed to load X509 certificate %s: %s" % (filename, err))

  (errcode, msg) = \
    utils.VerifyX509Certificate(cert, constants.SSL_CERT_EXPIRATION_WARN,
                                constants.SSL_CERT_EXPIRATION_ERROR)

  if msg:
    fnamemsg = "While verifying %s: %s" % (filename, msg)
  else:
    fnamemsg = None

  if errcode is None:
    return (None, fnamemsg)
  elif errcode == utils.CERT_WARNING:
    return (LUClusterVerifyConfig.ETYPE_WARNING, fnamemsg)
  elif errcode == utils.CERT_ERROR:
    return (LUClusterVerifyConfig.ETYPE_ERROR, fnamemsg)

  raise errors.ProgrammerError("Unhandled certificate error code %r" % errcode)


def _GetAllHypervisorParameters(cluster, instances):
  """Compute the set of all hypervisor parameters.

  @type cluster: L{objects.Cluster}
  @param cluster: the cluster object
  @param instances: list of L{objects.Instance}
  @param instances: additional instances from which to obtain parameters
  @rtype: list of (origin, hypervisor, parameters)
  @return: a list with all parameters found, indicating the hypervisor they
       apply to, and the origin (can be "cluster", "os X", or "instance Y")

  """
  hvp_data = []

  for hv_name in cluster.enabled_hypervisors:
    hvp_data.append(("cluster", hv_name, cluster.GetHVDefaults(hv_name)))

  for os_name, os_hvp in cluster.os_hvp.items():
    for hv_name, hv_params in os_hvp.items():
      if hv_params:
        full_params = cluster.GetHVDefaults(hv_name, os_name=os_name)
        hvp_data.append(("os %s" % os_name, hv_name, full_params))

  # TODO: collapse identical parameter values in a single one
  for instance in instances:
    if instance.hvparams:
      hvp_data.append(("instance %s" % instance.name, instance.hypervisor,
                       cluster.FillHV(instance)))

  return hvp_data


class LUClusterVerifyConfig(NoHooksLU, _VerifyErrors):
  """Verifies the cluster config.

  """
  REQ_BGL = False

  def _VerifyHVP(self, hvp_data):
    """Verifies locally the syntax of the hypervisor parameters.

    """
    for item, hv_name, hv_params in hvp_data:
      msg = ("hypervisor %s parameters syntax check (source %s): %%s" %
             (item, hv_name))
      try:
        hv_class = hypervisor.GetHypervisorClass(hv_name)
        utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES)
        hv_class.CheckParameterSyntax(hv_params)
      except errors.GenericError, err:
        self._ErrorIf(True, constants.CV_ECLUSTERCFG, None, msg % str(err))

  def ExpandNames(self):
    self.needed_locks = dict.fromkeys(locking.LEVELS, locking.ALL_SET)
1328
    self.share_locks = ShareAll()
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367

  def CheckPrereq(self):
    """Check prerequisites.

    """
    # Retrieve all information
    self.all_group_info = self.cfg.GetAllNodeGroupsInfo()
    self.all_node_info = self.cfg.GetAllNodesInfo()
    self.all_inst_info = self.cfg.GetAllInstancesInfo()

  def Exec(self, feedback_fn):
    """Verify integrity of cluster, performing various test on nodes.

    """
    self.bad = False
    self._feedback_fn = feedback_fn

    feedback_fn("* Verifying cluster config")

    for msg in self.cfg.VerifyConfig():
      self._ErrorIf(True, constants.CV_ECLUSTERCFG, None, msg)

    feedback_fn("* Verifying cluster certificate files")

    for cert_filename in pathutils.ALL_CERT_FILES:
      (errcode, msg) = _VerifyCertificate(cert_filename)
      self._ErrorIf(errcode, constants.CV_ECLUSTERCERT, None, msg, code=errcode)

    feedback_fn("* Verifying hypervisor parameters")

    self._VerifyHVP(_GetAllHypervisorParameters(self.cfg.GetClusterInfo(),
                                                self.all_inst_info.values()))

    feedback_fn("* Verifying all nodes belong to an existing group")

    # We do this verification here because, should this bogus circumstance
    # occur, it would never be caught by VerifyGroup, which only acts on
    # nodes/instances reachable from existing node groups.

Thomas Thrainer's avatar
Thomas Thrainer committed
1368
    dangling_nodes = set(node for node in self.all_node_info.values()
1369
1370
1371
1372
1373
1374
                         if node.group not in self.all_group_info)

    dangling_instances = {}
    no_node_instances = []

    for inst in self.all_inst_info.values():
Thomas Thrainer's avatar
Thomas Thrainer committed
1375
      if inst.primary_node in [node.uuid for node in dangling_nodes]:
1376
1377
1378
1379
1380
1381
1382
        dangling_instances.setdefault(inst.primary_node, []).append(inst.name)
      elif inst.primary_node not in self.all_node_info:
        no_node_instances.append(inst.name)

    pretty_dangling = [
        "%s (%s)" %
        (node.name,
Thomas Thrainer's avatar
Thomas Thrainer committed
1383
         utils.CommaJoin(dangling_instances.get(node.uuid,
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
                                                ["no instances"])))
        for node in dangling_nodes]

    self._ErrorIf(bool(dangling_nodes), constants.CV_ECLUSTERDANGLINGNODES,
                  None,
                  "the following nodes (and their instances) belong to a non"
                  " existing group: %s", utils.CommaJoin(pretty_dangling))

    self._ErrorIf(bool(no_node_instances), constants.CV_ECLUSTERDANGLINGINST,
                  None,
                  "the following instances have a non-existing primary-node:"
                  " %s", utils.CommaJoin(no_node_instances))

    return not self.bad


class LUClusterVerifyGroup(LogicalUnit, _VerifyErrors):
  """Verifies the status of a node group.

  """
  HPATH = "cluster-verify"
  HTYPE = constants.HTYPE_CLUSTER
  REQ_BGL = False

  _HOOKS_INDENT_RE = re.compile("^", re.M)

  class NodeImage(object):
    """A class representing the logical and physical status of a node.

Thomas Thrainer's avatar
Thomas Thrainer committed
1413
1414
    @type uuid: string
    @ivar uuid: the node UUID to which this object refers
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
    @ivar volumes: a structure as returned from
        L{ganeti.backend.GetVolumeList} (runtime)
    @ivar instances: a list of running instances (runtime)
    @ivar pinst: list of configured primary instances (config)
    @ivar sinst: list of configured secondary instances (config)
    @ivar sbp: dictionary of {primary-node: list of instances} for all
        instances for which this node is secondary (config)
    @ivar mfree: free memory, as reported by hypervisor (runtime)
    @ivar dfree: free disk, as reported by the node (runtime)
    @ivar offline: the offline status (config)
    @type rpc_fail: boolean
    @ivar rpc_fail: whether the RPC verify call was successfull (overall,
        not whether the individual keys were correct) (runtime)
    @type lvm_fail: boolean
    @ivar lvm_fail: whether the RPC call didn't return valid LVM data
    @type hyp_fail: boolean
    @ivar hyp_fail: whether the RPC call didn't return the instance list
    @type ghost: boolean
    @ivar ghost: whether this is a known node or not (config)
    @type os_fail: boolean
    @ivar os_fail: whether the RPC call didn't return valid OS data
    @type oslist: list
    @ivar oslist: list of OSes as diagnosed by DiagnoseOS
    @type vm_capable: boolean
    @ivar vm_capable: whether the node can host instances
    @type pv_min: float
    @ivar pv_min: size in MiB of the smallest PVs
    @type pv_max: float
    @ivar pv_max: size in MiB of the biggest PVs

    """
Thomas Thrainer's avatar
Thomas Thrainer committed
1446
1447
    def __init__(self, offline=False, uuid=None, vm_capable=True):
      self.uuid = uuid
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
      self.volumes = {}
      self.instances = []
      self.pinst = []
      self.sinst = []
      self.sbp = {}
      self.mfree = 0
      self.dfree = 0
      self.offline = offline
      self.vm_capable = vm_capable
      self.rpc_fail = False
      self.lvm_fail = False
      self.hyp_fail = False
      self.ghost = False
      self.os_fail = False
      self.oslist = {}
      self.pv_min = None
      self.pv_max = None

  def ExpandNames(self):
    # This raises errors.OpPrereqError on its own:
    self.group_uuid = self.cfg.LookupNodeGroup(self.op.group_name)

    # Get instances in node group; this is unsafe and needs verification later
    inst_names = \
      self.cfg.GetNodeGroupInstances(self.group_uuid, primary_only=True)

    self.needed_locks = {
      locking.LEVEL_INSTANCE: inst_names,
      locking.LEVEL_NODEGROUP: [self.group_uuid],
      locking.LEVEL_NODE: [],

      # This opcode is run by watcher every five minutes and acquires all nodes
      # for a group. It doesn't run for a long time, so it's better to acquire
      # the node allocation lock as well.
      locking.LEVEL_NODE_ALLOC: locking.ALL_SET,
      }

1485
    self.share_locks = ShareAll()
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509

  def DeclareLocks(self, level):
    if level == locking.LEVEL_NODE:
      # Get members of node group; this is unsafe and needs verification later
      nodes = set(self.cfg.GetNodeGroup(self.group_uuid).members)

      all_inst_info = self.cfg.GetAllInstancesInfo()

      # In Exec(), we warn about mirrored instances that have primary and
      # secondary living in separate node groups. To fully verify that
      # volumes for these instances are healthy, we will need to do an
      # extra call to their secondaries. We ensure here those nodes will
      # be locked.
      for inst in self.owned_locks(locking.LEVEL_INSTANCE):
        # Important: access only the instances whose lock is owned
        if all_inst_info[inst].disk_template in constants.DTS_INT_MIRROR:
          nodes.update(all_inst_info[inst].secondary_nodes)

      self.needed_locks[locking.LEVEL_NODE] = nodes

  def CheckPrereq(self):
    assert self.group_uuid in self.owned_locks(locking.LEVEL_NODEGROUP)
    self.group_info = self.cfg.GetNodeGroup(self.group_uuid)

Thomas Thrainer's avatar
Thomas Thrainer committed
1510
    group_node_uuids = set(self.group_info.members)
1511
1512
1513
    group_instances = \
      self.cfg.GetNodeGroupInstances(self.group_uuid, primary_only=True)

Thomas Thrainer's avatar
Thomas Thrainer committed
1514
1515
    unlocked_node_uuids = \
        group_node_uuids.difference(self.owned_locks(locking.LEVEL_NODE))
1516
1517
1518
1519

    unlocked_instances = \
        group_instances.difference(self.owned_locks(locking.LEVEL_INSTANCE))

Thomas Thrainer's avatar
Thomas Thrainer committed
1520
1521
1522
1523
1524
    if unlocked_node_uuids:
      raise errors.OpPrereqError(
        "Missing lock for nodes: %s" %
        utils.CommaJoin(self.cfg.GetNodeNames(unlocked_node_uuids)),
        errors.ECODE_STATE)
1525
1526
1527
1528
1529
1530
1531
1532
1533

    if unlocked_instances:
      raise errors.OpPrereqError("Missing lock for instances: %s" %
                                 utils.CommaJoin(unlocked_instances),
                                 errors.ECODE_STATE)

    self.all_node_info = self.cfg.GetAllNodesInfo()
    self.all_inst_info = self.cfg.GetAllInstancesInfo()

Thomas Thrainer's avatar
Thomas Thrainer committed
1534
1535
1536
    self.my_node_uuids = group_node_uuids
    self.my_node_info = dict((node_uuid, self.all_node_info[node_uuid])
                             for node_uuid in group_node_uuids)
1537

Thomas Thrainer's avatar
Thomas Thrainer committed
1538
    self.my_inst_names = utils.NiceSort(group_instances)
1539
1540
1541
1542
1543
1544
1545
1546
1547
    self.my_inst_info = dict((name, self.all_inst_info[name])
                             for name in self.my_inst_names)

    # We detect here the nodes that will need the extra RPC calls for verifying
    # split LV volumes; they should be locked.
    extra_lv_nodes = set()

    for inst in self.my_inst_info.values():
      if inst.disk_template in constants.DTS_INT_MIRROR:
Thomas Thrainer's avatar
Thomas Thrainer committed
1548
1549
1550
        for nuuid in inst.all_nodes:
          if self.all_node_info[nuuid].group != self.group_uuid:
            extra_lv_nodes.add(nuuid)
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577

    unlocked_lv_nodes = \
        extra_lv_nodes.difference(self.owned_locks(locking.LEVEL_NODE))

    if unlocked_lv_nodes:
      raise errors.OpPrereqError("Missing node locks for LV check: %s" %
                                 utils.CommaJoin(unlocked_lv_nodes),
                                 errors.ECODE_STATE)
    self.extra_lv_nodes = list(extra_lv_nodes)

  def _VerifyNode(self, ninfo, nresult):
    """Perform some basic validation on data returned from a node.

      - check the result data structure is well formed and has all the
        mandatory fields
      - check ganeti version

    @type ninfo: L{objects.Node}
    @param ninfo: the node to check
    @param nresult: the results from the node
    @rtype: boolean
    @return: whether overall this call was successful (and we can expect
         reasonable values in the respose)

    """
    # main result, nresult should be a non-empty dict
    test = not nresult or not isinstance(nresult, dict)
1578
    self._ErrorIf(test, constants.CV_ENODERPC, ninfo.name,
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
                  "unable to verify node: no data returned")
    if test:
      return False

    # compares ganeti version
    local_version = constants.PROTOCOL_VERSION
    remote_version = nresult.get("version", None)
    test = not (remote_version and
                isinstance(remote_version, (list, tuple)) and
                len(remote_version) == 2)
1589
1590
    self._ErrorIf(test, constants.CV_ENODERPC, ninfo.name,
                  "connection to node returned invalid data")
1591
1592
1593
1594
    if test:
      return False

    test = local_version != remote_version[0]
1595
1596
1597
    self._ErrorIf(test, constants.CV_ENODEVERSION, ninfo.name,
                  "incompatible protocol versions: master %s,"
                  " node %s", local_version, remote_version[0])
1598
1599
1600
1601
1602
1603
1604
    if test:
      return False

    # node seems compatible, we can actually try to look into its results

    # full package version
    self._ErrorIf(constants.RELEASE_VERSION != remote_version[1],
1605
                  constants.CV_ENODEVERSION, ninfo.name,
1606
1607
1608
1609
1610
1611
1612
1613
                  "software version mismatch: master %s, node %s",
                  constants.RELEASE_VERSION, remote_version[1],
                  code=self.ETYPE_WARNING)

    hyp_result = nresult.get(constants.NV_HYPERVISOR, None)
    if ninfo.vm_capable and isinstance(hyp_result, dict):
      for hv_name, hv_result in hyp_result.iteritems():
        test = hv_result is not None
1614
1615
        self._ErrorIf(test, constants.CV_ENODEHV, ninfo.name,
                      "hypervisor %s verify failure: '%s'", hv_name, hv_result)
1616
1617
1618
1619

    hvp_result = nresult.get(constants.NV_HVPARAMS, None)
    if ninfo.vm_capable and isinstance(hvp_result, list):
      for item, hv_name, hv_result in hvp_result:
1620
1621
1622
        self._ErrorIf(True, constants.CV_ENODEHV, ninfo.name,
                      "hypervisor %s parameter verify failure (source %s): %s",
                      hv_name, item, hv_result)
1623
1624
1625

    test = nresult.get(constants.NV_NODESETUP,
                       ["Missing NODESETUP results"])
1626
1627
    self._ErrorIf(test, constants.CV_ENODESETUP, ninfo.name,
                  "node setup error: %s", "; ".join(test))
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645

    return True

  def _VerifyNodeTime(self, ninfo, nresult,
                      nvinfo_starttime, nvinfo_endtime):
    """Check the node time.

    @type ninfo: L{objects.Node}
    @param ninfo: the node to check
    @param nresult: the remote results for the node
    @param nvinfo_starttime: the start time of the RPC call
    @param nvinfo_endtime: the end time of the RPC call

    """
    ntime = nresult.get(constants.NV_TIME, None)
    try:
      ntime_merged = utils.MergeTime(ntime)
    except (ValueError, TypeError):
1646
1647
      self._ErrorIf(True, constants.CV_ENODETIME, ninfo.name,
                    "Node returned invalid time")
1648
1649
1650
1651
1652
1653
1654
1655
1656
      return

    if ntime_merged < (nvinfo_starttime - constants.NODE_MAX_CLOCK_SKEW):
      ntime_diff = "%.01fs" % abs(nvinfo_starttime - ntime_merged)
    elif ntime_merged > (nvinfo_endtime + constants.NODE_MAX_CLOCK_SKEW):
      ntime_diff = "%.01fs" % abs(ntime_merged - nvinfo_endtime)
    else:
      ntime_diff = None

1657
1658
1659
    self._ErrorIf(ntime_diff is not None, constants.CV_ENODETIME, ninfo.name,
                  "Node time diverges by at least %s from master node time",
                  ntime_diff)
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677

  def _UpdateVerifyNodeLVM(self, ninfo, nresult, vg_name, nimg):
    """Check the node LVM results and update info for cross-node checks.

    @type ninfo: L{objects.Node}
    @param ninfo: the node to check
    @param nresult: the remote results for the node
    @param vg_name: the configured VG name
    @type nimg: L{NodeImage}
    @param nimg: node image

    """
    if vg_name is None:
      return

    # checks vg existence and size > 20G
    vglist = nresult.get(constants.NV_VGLIST, None)
    test = not vglist
1678
1679
    self._ErrorIf(test, constants.CV_ENODELVM, ninfo.name,
                  "unable to check volume groups")
1680
1681
1682
    if not test:
      vgstatus = utils.CheckVolumeGroupSize(vglist, vg_name,
                                            constants.MIN_VG_SIZE)
1683
      self._ErrorIf(vgstatus, constants.CV_ENODELVM, ninfo.name, vgstatus)
1684
1685

    # Check PVs
1686
    (errmsgs, pvminmax) = CheckNodePVs(nresult, self._exclusive_storage)
1687
    for em in errmsgs:
1688
      self._Error(constants.CV_ENODELVM, ninfo.name, em)
1689
1690
1691
    if pvminmax is not None:
      (nimg.pv_min, nimg.pv_max) = pvminmax

1692
1693
1694
1695
1696
1697
1698
1699
1700
  def _VerifyGroupDRBDVersion(self, node_verify_infos):
    """Check cross-node DRBD version consistency.

    @type node_verify_infos: dict
    @param node_verify_infos: infos about nodes as returned from the
      node_verify call.

    """
    node_versions = {}
Thomas Thrainer's avatar
Thomas Thrainer committed
1701
    for node_uuid, ndata in node_verify_infos.items():
1702
1703
      nresult = ndata.payload
      version = nresult.get(constants.NV_DRBDVERSION, "Missing DRBD version")
Thomas Thrainer's avatar
Thomas Thrainer committed
1704
      node_versions[node_uuid] = version
1705
1706

    if len(set(node_versions.values())) > 1:
Thomas Thrainer's avatar
Thomas Thrainer committed
1707
      for node_uuid, version in sorted(node_versions.items()):
1708
        msg = "DRBD version mismatch: %s" % version
Thomas Thrainer's avatar
Thomas Thrainer committed
1709
        self._Error(constants.CV_ENODEDRBDHELPER, node_uuid, msg,
1710
1711
                    code=self.ETYPE_WARNING)

1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
  def _VerifyGroupLVM(self, node_image, vg_name):
    """Check cross-node consistency in LVM.

    @type node_image: dict
    @param node_image: info about nodes, mapping from node to names to
      L{NodeImage} objects
    @param vg_name: the configured VG name

    """
    if vg_name is None:
      return

1724
    # Only exclusive storage needs this kind of checks
1725
1726
1727
1728
1729
1730
1731
1732
1733
    if not self._exclusive_storage:
      return

    # exclusive_storage wants all PVs to have the same size (approximately),
    # if the smallest and the biggest ones are okay, everything is fine.
    # pv_min is None iff pv_max is None
    vals = filter((lambda ni: ni.pv_min is not None), node_image.values())
    if not vals:
      return
1734
1735
    (pvmin, minnode_uuid) = min((ni.pv_min, ni.uuid) for ni in vals)
    (pvmax, maxnode_uuid) = max((ni.pv_max, ni.uuid) for ni in vals)
1736
1737
1738
1739
    bad = utils.LvmExclusiveTestBadPvSizes(pvmin, pvmax)
    self._ErrorIf(bad, constants.CV_EGROUPDIFFERENTPVSIZE, self.group_info.name,
                  "PV sizes differ too much in the group; smallest (%s MB) is"
                  " on %s, biggest (%s MB) is on %s",
1740
1741
                  pvmin, self.cfg.GetNodeName(minnode_uuid),
                  pvmax, self.cfg.GetNodeName(maxnode_uuid))
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756

  def _VerifyNodeBridges(self, ninfo, nresult, bridges):
    """Check the node bridges.

    @type ninfo: L{objects.Node}
    @param ninfo: the node to check
    @param nresult: the remote results for the node
    @param bridges: the expected list of bridges

    """
    if not bridges:
      return

    missing = nresult.get(constants.NV_BRIDGES, None)
    test = not isinstance(missing, list)
1757
1758
    self._ErrorIf(test, constants.CV_ENODENET, ninfo.name,
                  "did not return valid bridge information")
1759
    if not test:
1760
1761
      self._ErrorIf(bool(missing), constants.CV_ENODENET, ninfo.name,
                    "missing bridges: %s" % utils.CommaJoin(sorted(missing)))
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771

  def _VerifyNodeUserScripts(self, ninfo, nresult):
    """Check the results of user scripts presence and executability on the node

    @type ninfo: L{objects.Node}
    @param ninfo: the node to check
    @param nresult: the remote results for the node

    """
    test = not constants.NV_USERSCRIPTS in nresult
1772
    self._ErrorIf(test, constants.CV_ENODEUSERSCRIPTS, ninfo.name,
1773
1774
1775
1776
                  "did not return user scripts information")

    broken_scripts = nresult.get(constants.NV_USERSCRIPTS, None)
    if not test:
1777
      self._ErrorIf(broken_scripts, constants.CV_ENODEUSERSCRIPTS, ninfo.name,
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
                    "user scripts not present or not executable: %s" %
                    utils.CommaJoin(sorted(broken_scripts)))

  def _VerifyNodeNetwork(self, ninfo, nresult):
    """Check the node network connectivity results.

    @type ninfo: L{objects.Node}
    @param ninfo: the node to check
    @param nresult: the remote results for the node

    """
    test = constants.NV_NODELIST not in nresult
1790
1791
    self._ErrorIf(test, constants.CV_ENODESSH, ninfo.name,
                  "node hasn't returned node ssh connectivity data")
1792
1793
1794
    if not test:
      if nresult[constants.NV_NODELIST]:
        for a_node, a_msg in nresult[constants.NV_NODELIST].items():
1795
1796
          self._ErrorIf(True, constants.CV_ENODESSH, ninfo.name,
                        "ssh communication with node '%s': %s", a_node, a_msg)
1797
1798