Synnefo » snf-ganeti » ganeti-local

#

#

# 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 nodes."""

import logging

import operator

from ganeti import constants

from ganeti import errors

from ganeti import locking

from ganeti import netutils

from ganeti import objects

from ganeti import opcodes

from ganeti import qlang

from ganeti import query

from ganeti import rpc

from ganeti import utils

from ganeti.masterd import iallocator

from ganeti.cmdlib.base import LogicalUnit, NoHooksLU, QueryBase, \

  ResultWithJobs

from ganeti.cmdlib.common import CheckParamsNotGlobal, \

  MergeAndVerifyHvState, MergeAndVerifyDiskState, \

  IsExclusiveStorageEnabledNode, CheckNodePVs, \

  RedistributeAncillaryFiles, ExpandNodeUuidAndName, ShareAll, SupportsOob, \

  CheckInstanceState, INSTANCE_DOWN, GetUpdatedParams, \

  AdjustCandidatePool, CheckIAllocatorOrNode, LoadNodeEvacResult, \

  GetWantedNodes, MapInstanceLvsToNodes, RunPostHook, \

  FindFaultyInstanceDisks, CheckStorageTypeEnabled

def _DecideSelfPromotion(lu, exceptions=None):

  """Decide whether I should promote myself as a master candidate.

  """

  cp_size = lu.cfg.GetClusterInfo().candidate_pool_size

  mc_now, mc_should, _ = lu.cfg.GetMasterCandidateStats(exceptions)

  # the new node will increase mc_max with one, so:

  mc_should = min(mc_should + 1, cp_size)

  return mc_now < mc_should

def _CheckNodeHasSecondaryIP(lu, node, secondary_ip, prereq):

  """Ensure that a node has the given secondary ip.

  @type lu: L{LogicalUnit}

  @param lu: the LU on behalf of which we make the check

  @type node: L{objects.Node}

  @param node: the node to check

  @type secondary_ip: string

  @param secondary_ip: the ip to check

  @type prereq: boolean

  @param prereq: whether to throw a prerequisite or an execute error

  @raise errors.OpPrereqError: if the node doesn't have the ip,

  and prereq=True

  @raise errors.OpExecError: if the node doesn't have the ip, and prereq=False

  """

  # this can be called with a new node, which has no UUID yet, so perform the

  # RPC call using its name

  result = lu.rpc.call_node_has_ip_address(node.name, secondary_ip)

  result.Raise("Failure checking secondary ip on node %s" % node.name,

               prereq=prereq, ecode=errors.ECODE_ENVIRON)

  if not result.payload:

    msg = ("Node claims it doesn't have the secondary ip you gave (%s),"

           " please fix and re-run this command" % secondary_ip)

    if prereq:

      raise errors.OpPrereqError(msg, errors.ECODE_ENVIRON)

    else:

      raise errors.OpExecError(msg)

class LUNodeAdd(LogicalUnit):

  """Logical unit for adding node to the cluster.

  """

  HPATH = "node-add"

  HTYPE = constants.HTYPE_NODE

  _NFLAGS = ["master_capable", "vm_capable"]

  def CheckArguments(self):

    self.primary_ip_family = self.cfg.GetPrimaryIPFamily()

    # validate/normalize the node name

    self.hostname = netutils.GetHostname(name=self.op.node_name,

                                         family=self.primary_ip_family)

    self.op.node_name = self.hostname.name

    if self.op.readd and self.op.node_name == self.cfg.GetMasterNodeName():

      raise errors.OpPrereqError("Cannot readd the master node",

                                 errors.ECODE_STATE)

    if self.op.readd and self.op.group:

      raise errors.OpPrereqError("Cannot pass a node group when a node is"

                                 " being readded", errors.ECODE_INVAL)

  def BuildHooksEnv(self):

    """Build hooks env.

    This will run on all nodes before, and on all nodes + the new node after.

    """

    return {

      "OP_TARGET": self.op.node_name,

      "NODE_NAME": self.op.node_name,

      "NODE_PIP": self.op.primary_ip,

      "NODE_SIP": self.op.secondary_ip,

      "MASTER_CAPABLE": str(self.op.master_capable),

      "VM_CAPABLE": str(self.op.vm_capable),

      }

  def BuildHooksNodes(self):

    """Build hooks nodes.

    """

    hook_nodes = self.cfg.GetNodeList()

    new_node_info = self.cfg.GetNodeInfoByName(self.op.node_name)

    if new_node_info is not None:

      # Exclude added node

      hook_nodes = list(set(hook_nodes) - set([new_node_info.uuid]))

    # add the new node as post hook node by name; it does not have an UUID yet

    return (hook_nodes, hook_nodes, [self.op.node_name, ])

  def CheckPrereq(self):

    """Check prerequisites.

    This checks:

     - the new node is not already in the config

     - it is resolvable

     - its parameters (single/dual homed) matches the cluster

    Any errors are signaled by raising errors.OpPrereqError.

    """

    node_name = self.hostname.name

    self.op.primary_ip = self.hostname.ip

    if self.op.secondary_ip is None:

      if self.primary_ip_family == netutils.IP6Address.family:

        raise errors.OpPrereqError("When using a IPv6 primary address, a valid"

                                   " IPv4 address must be given as secondary",

                                   errors.ECODE_INVAL)

      self.op.secondary_ip = self.op.primary_ip

    secondary_ip = self.op.secondary_ip

    if not netutils.IP4Address.IsValid(secondary_ip):

      raise errors.OpPrereqError("Secondary IP (%s) needs to be a valid IPv4"

                                 " address" % secondary_ip, errors.ECODE_INVAL)

    existing_node_info = self.cfg.GetNodeInfoByName(node_name)

    if not self.op.readd and existing_node_info is not None:

      raise errors.OpPrereqError("Node %s is already in the configuration" %

                                 node_name, errors.ECODE_EXISTS)

    elif self.op.readd and existing_node_info is None:

      raise errors.OpPrereqError("Node %s is not in the configuration" %

                                 node_name, errors.ECODE_NOENT)

    self.changed_primary_ip = False

    for existing_node in self.cfg.GetAllNodesInfo().values():

      if self.op.readd and node_name == existing_node.name:

        if existing_node.secondary_ip != secondary_ip:

          raise errors.OpPrereqError("Readded node doesn't have the same IP"

                                     " address configuration as before",

                                     errors.ECODE_INVAL)

        if existing_node.primary_ip != self.op.primary_ip:

          self.changed_primary_ip = True

        continue

      if (existing_node.primary_ip == self.op.primary_ip or

          existing_node.secondary_ip == self.op.primary_ip or

          existing_node.primary_ip == secondary_ip or

          existing_node.secondary_ip == secondary_ip):

        raise errors.OpPrereqError("New node ip address(es) conflict with"

                                   " existing node %s" % existing_node.name,

                                   errors.ECODE_NOTUNIQUE)

    # After this 'if' block, None is no longer a valid value for the

    # _capable op attributes

    if self.op.readd:

      assert existing_node_info is not None, \

        "Can't retrieve locked node %s" % node_name

      for attr in self._NFLAGS:

        if getattr(self.op, attr) is None:

          setattr(self.op, attr, getattr(existing_node_info, attr))

    else:

      for attr in self._NFLAGS:

        if getattr(self.op, attr) is None:

          setattr(self.op, attr, True)

    if self.op.readd and not self.op.vm_capable:

      pri, sec = self.cfg.GetNodeInstances(existing_node_info.uuid)

      if pri or sec:

        raise errors.OpPrereqError("Node %s being re-added with vm_capable"

                                   " flag set to false, but it already holds"

                                   " instances" % node_name,

                                   errors.ECODE_STATE)

    # check that the type of the node (single versus dual homed) is the

    # same as for the master

    myself = self.cfg.GetMasterNodeInfo()

    master_singlehomed = myself.secondary_ip == myself.primary_ip

    newbie_singlehomed = secondary_ip == self.op.primary_ip

    if master_singlehomed != newbie_singlehomed:

      if master_singlehomed:

        raise errors.OpPrereqError("The master has no secondary ip but the"

                                   " new node has one",

                                   errors.ECODE_INVAL)

      else:

        raise errors.OpPrereqError("The master has a secondary ip but the"

                                   " new node doesn't have one",

                                   errors.ECODE_INVAL)

    # checks reachability

    if not netutils.TcpPing(self.op.primary_ip, constants.DEFAULT_NODED_PORT):

      raise errors.OpPrereqError("Node not reachable by ping",

                                 errors.ECODE_ENVIRON)

    if not newbie_singlehomed:

      # check reachability from my secondary ip to newbie's secondary ip

      if not netutils.TcpPing(secondary_ip, constants.DEFAULT_NODED_PORT,

                              source=myself.secondary_ip):

        raise errors.OpPrereqError("Node secondary ip not reachable by TCP"

                                   " based ping to node daemon port",

                                   errors.ECODE_ENVIRON)

    if self.op.readd:

      exceptions = [existing_node_info.uuid]

    else:

      exceptions = []

    if self.op.master_capable:

      self.master_candidate = _DecideSelfPromotion(self, exceptions=exceptions)

    else:

      self.master_candidate = False

    if self.op.readd:

      self.new_node = existing_node_info

    else:

      node_group = self.cfg.LookupNodeGroup(self.op.group)

      self.new_node = objects.Node(name=node_name,

                                   primary_ip=self.op.primary_ip,

                                   secondary_ip=secondary_ip,

                                   master_candidate=self.master_candidate,

                                   offline=False, drained=False,

                                   group=node_group, ndparams={})

    if self.op.ndparams:

      utils.ForceDictType(self.op.ndparams, constants.NDS_PARAMETER_TYPES)

      CheckParamsNotGlobal(self.op.ndparams, constants.NDC_GLOBALS, "node",

                           "node", "cluster or group")

    if self.op.hv_state:

      self.new_hv_state = MergeAndVerifyHvState(self.op.hv_state, None)

    if self.op.disk_state:

      self.new_disk_state = MergeAndVerifyDiskState(self.op.disk_state, None)

    # TODO: If we need to have multiple DnsOnlyRunner we probably should make

    #       it a property on the base class.

    rpcrunner = rpc.DnsOnlyRunner()

    result = rpcrunner.call_version([node_name])[node_name]

    result.Raise("Can't get version information from node %s" % node_name)

    if constants.PROTOCOL_VERSION == result.payload:

      logging.info("Communication to node %s fine, sw version %s match",

                   node_name, result.payload)

    else:

      raise errors.OpPrereqError("Version mismatch master version %s,"

                                 " node version %s" %

                                 (constants.PROTOCOL_VERSION, result.payload),

                                 errors.ECODE_ENVIRON)

    vg_name = self.cfg.GetVGName()

    if vg_name is not None:

      vparams = {constants.NV_PVLIST: [vg_name]}

      excl_stor = IsExclusiveStorageEnabledNode(self.cfg, self.new_node)

      cname = self.cfg.GetClusterName()

      result = rpcrunner.call_node_verify_light(

          [node_name], vparams, cname,

          self.cfg.GetClusterInfo().hvparams)[node_name]

      (errmsgs, _) = CheckNodePVs(result.payload, excl_stor)

      if errmsgs:

        raise errors.OpPrereqError("Checks on node PVs failed: %s" %

                                   "; ".join(errmsgs), errors.ECODE_ENVIRON)

  def _InitOpenVSwitch(self):

    filled_ndparams = self.cfg.GetClusterInfo().FillND(

      self.new_node, self.cfg.GetNodeGroup(self.new_node.group))

    ovs = filled_ndparams.get(constants.ND_OVS, None)

    ovs_name = filled_ndparams.get(constants.ND_OVS_NAME, None)

    ovs_link = filled_ndparams.get(constants.ND_OVS_LINK, None)

    if ovs:

      if not ovs_link:

        self.LogInfo("No physical interface for OpenvSwitch was given."

                     " OpenvSwitch will not have an outside connection. This"

                     " might not be what you want.")

      result = self.rpc.call_node_configure_ovs(

                 self.new_node.name, ovs_name, ovs_link)

      result.Raise("Failed to initialize OpenVSwitch on new node")

  def Exec(self, feedback_fn):

    """Adds the new node to the cluster.

    """

    assert locking.BGL in self.owned_locks(locking.LEVEL_CLUSTER), \

      "Not owning BGL"

    # We adding a new node so we assume it's powered

    self.new_node.powered = True

    # for re-adds, reset the offline/drained/master-candidate flags;

    # we need to reset here, otherwise offline would prevent RPC calls

    # later in the procedure; this also means that if the re-add

    # fails, we are left with a non-offlined, broken node

    if self.op.readd:

      self.new_node.offline = False

      self.new_node.drained = False

      self.LogInfo("Readding a node, the offline/drained flags were reset")

      # if we demote the node, we do cleanup later in the procedure

      self.new_node.master_candidate = self.master_candidate

      if self.changed_primary_ip:

        self.new_node.primary_ip = self.op.primary_ip

    # copy the master/vm_capable flags

    for attr in self._NFLAGS:

      setattr(self.new_node, attr, getattr(self.op, attr))

    # notify the user about any possible mc promotion

    if self.new_node.master_candidate:

      self.LogInfo("Node will be a master candidate")

    if self.op.ndparams:

      self.new_node.ndparams = self.op.ndparams

    else:

      self.new_node.ndparams = {}

    if self.op.hv_state:

      self.new_node.hv_state_static = self.new_hv_state

    if self.op.disk_state:

      self.new_node.disk_state_static = self.new_disk_state

    # Add node to our /etc/hosts, and add key to known_hosts

    if self.cfg.GetClusterInfo().modify_etc_hosts:

      master_node = self.cfg.GetMasterNode()

      result = self.rpc.call_etc_hosts_modify(

                 master_node, constants.ETC_HOSTS_ADD, self.hostname.name,

                 self.hostname.ip)

      result.Raise("Can't update hosts file with new host data")

    if self.new_node.secondary_ip != self.new_node.primary_ip:

      _CheckNodeHasSecondaryIP(self, self.new_node, self.new_node.secondary_ip,

                               False)

    node_verifier_uuids = [self.cfg.GetMasterNode()]

    node_verify_param = {

      constants.NV_NODELIST: ([self.new_node.name], {}),

      # TODO: do a node-net-test as well?

    }

    result = self.rpc.call_node_verify(

               node_verifier_uuids, node_verify_param,

               self.cfg.GetClusterName(),

               self.cfg.GetClusterInfo().hvparams)

    for verifier in node_verifier_uuids:

      result[verifier].Raise("Cannot communicate with node %s" % verifier)

      nl_payload = result[verifier].payload[constants.NV_NODELIST]

      if nl_payload:

        for failed in nl_payload:

          feedback_fn("ssh/hostname verification failed"

                      " (checking from %s): %s" %

                      (verifier, nl_payload[failed]))

        raise errors.OpExecError("ssh/hostname verification failed")

    self._InitOpenVSwitch()

    if self.op.readd:

      self.context.ReaddNode(self.new_node)

      RedistributeAncillaryFiles(self)

      # make sure we redistribute the config

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

      # and make sure the new node will not have old files around

      if not self.new_node.master_candidate:

        result = self.rpc.call_node_demote_from_mc(self.new_node.uuid)

        result.Warn("Node failed to demote itself from master candidate status",

                    self.LogWarning)

    else:

      self.context.AddNode(self.new_node, self.proc.GetECId())

      RedistributeAncillaryFiles(self)

class LUNodeSetParams(LogicalUnit):

  """Modifies the parameters of a node.

  @cvar _F2R: a dictionary from tuples of flags (mc, drained, offline)

      to the node role (as _ROLE_*)

  @cvar _R2F: a dictionary from node role to tuples of flags

  @cvar _FLAGS: a list of attribute names corresponding to the flags

  """

  HPATH = "node-modify"

  HTYPE = constants.HTYPE_NODE

  REQ_BGL = False

  (_ROLE_CANDIDATE, _ROLE_DRAINED, _ROLE_OFFLINE, _ROLE_REGULAR) = range(4)

  _F2R = {

    (True, False, False): _ROLE_CANDIDATE,

    (False, True, False): _ROLE_DRAINED,

    (False, False, True): _ROLE_OFFLINE,

    (False, False, False): _ROLE_REGULAR,

    }

  _R2F = dict((v, k) for k, v in _F2R.items())

  _FLAGS = ["master_candidate", "drained", "offline"]

  def CheckArguments(self):

    (self.op.node_uuid, self.op.node_name) = \

      ExpandNodeUuidAndName(self.cfg, self.op.node_uuid, self.op.node_name)

    all_mods = [self.op.offline, self.op.master_candidate, self.op.drained,

                self.op.master_capable, self.op.vm_capable,

                self.op.secondary_ip, self.op.ndparams, self.op.hv_state,

                self.op.disk_state]

    if all_mods.count(None) == len(all_mods):

      raise errors.OpPrereqError("Please pass at least one modification",

                                 errors.ECODE_INVAL)

    if all_mods.count(True) > 1:

      raise errors.OpPrereqError("Can't set the node into more than one"

                                 " state at the same time",

                                 errors.ECODE_INVAL)

    # Boolean value that tells us whether we might be demoting from MC

    self.might_demote = (self.op.master_candidate is False or

                         self.op.offline is True or

                         self.op.drained is True or

                         self.op.master_capable is False)

    if self.op.secondary_ip:

      if not netutils.IP4Address.IsValid(self.op.secondary_ip):

        raise errors.OpPrereqError("Secondary IP (%s) needs to be a valid IPv4"

                                   " address" % self.op.secondary_ip,

                                   errors.ECODE_INVAL)

    self.lock_all = self.op.auto_promote and self.might_demote

    self.lock_instances = self.op.secondary_ip is not None

  def _InstanceFilter(self, instance):

    """Filter for getting affected instances.

    """

    return (instance.disk_template in constants.DTS_INT_MIRROR and

            self.op.node_uuid in instance.all_nodes)

  def ExpandNames(self):

    if self.lock_all:

      self.needed_locks = {

        locking.LEVEL_NODE: locking.ALL_SET,

        # Block allocations when all nodes are locked

        locking.LEVEL_NODE_ALLOC: locking.ALL_SET,

        }

    else:

      self.needed_locks = {

        locking.LEVEL_NODE: self.op.node_uuid,

        }

    # Since modifying a node can have severe effects on currently running

    # operations the resource lock is at least acquired in shared mode

    self.needed_locks[locking.LEVEL_NODE_RES] = \

      self.needed_locks[locking.LEVEL_NODE]

    # Get all locks except nodes in shared mode; they are not used for anything

    # but read-only access

    self.share_locks = ShareAll()

    self.share_locks[locking.LEVEL_NODE] = 0

    self.share_locks[locking.LEVEL_NODE_RES] = 0

    self.share_locks[locking.LEVEL_NODE_ALLOC] = 0

    if self.lock_instances:

      self.needed_locks[locking.LEVEL_INSTANCE] = \

        self.cfg.GetInstanceNames(

          self.cfg.GetInstancesInfoByFilter(self._InstanceFilter).keys())

  def BuildHooksEnv(self):

    """Build hooks env.

    This runs on the master node.

    """

    return {

      "OP_TARGET": self.op.node_name,

      "MASTER_CANDIDATE": str(self.op.master_candidate),

      "OFFLINE": str(self.op.offline),

      "DRAINED": str(self.op.drained),

      "MASTER_CAPABLE": str(self.op.master_capable),

      "VM_CAPABLE": str(self.op.vm_capable),

      }

  def BuildHooksNodes(self):

    """Build hooks nodes.

    """

    nl = [self.cfg.GetMasterNode(), self.op.node_uuid]

    return (nl, nl)

  def CheckPrereq(self):

    """Check prerequisites.

    This only checks the instance list against the existing names.

    """

    node = self.cfg.GetNodeInfo(self.op.node_uuid)

    if self.lock_instances:

      affected_instances = \

        self.cfg.GetInstancesInfoByFilter(self._InstanceFilter)

      # Verify instance locks

      owned_instance_names = self.owned_locks(locking.LEVEL_INSTANCE)

      wanted_instance_names = frozenset([inst.name for inst in

                                         affected_instances.values()])

      if wanted_instance_names - owned_instance_names:

        raise errors.OpPrereqError("Instances affected by changing node %s's"

                                   " secondary IP address have changed since"

                                   " locks were acquired, wanted '%s', have"

                                   " '%s'; retry the operation" %

                                   (node.name,

                                    utils.CommaJoin(wanted_instance_names),

                                    utils.CommaJoin(owned_instance_names)),

                                   errors.ECODE_STATE)

    else:

      affected_instances = None

    if (self.op.master_candidate is not None or

        self.op.drained is not None or

        self.op.offline is not None):

      # we can't change the master's node flags

      if node.uuid == self.cfg.GetMasterNode():

        raise errors.OpPrereqError("The master role can be changed"

                                   " only via master-failover",

                                   errors.ECODE_INVAL)

    if self.op.master_candidate and not node.master_capable:

      raise errors.OpPrereqError("Node %s is not master capable, cannot make"

                                 " it a master candidate" % node.name,

                                 errors.ECODE_STATE)

    if self.op.vm_capable is False:

      (ipri, isec) = self.cfg.GetNodeInstances(node.uuid)

      if ipri or isec:

        raise errors.OpPrereqError("Node %s hosts instances, cannot unset"

                                   " the vm_capable flag" % node.name,

                                   errors.ECODE_STATE)

    if node.master_candidate and self.might_demote and not self.lock_all:

      assert not self.op.auto_promote, "auto_promote set but lock_all not"

      # check if after removing the current node, we're missing master

      # candidates

      (mc_remaining, mc_should, _) = \

          self.cfg.GetMasterCandidateStats(exceptions=[node.uuid])

      if mc_remaining < mc_should:

        raise errors.OpPrereqError("Not enough master candidates, please"

                                   " pass auto promote option to allow"

                                   " promotion (--auto-promote or RAPI"

                                   " auto_promote=True)", errors.ECODE_STATE)

    self.old_flags = old_flags = (node.master_candidate,

                                  node.drained, node.offline)

    assert old_flags in self._F2R, "Un-handled old flags %s" % str(old_flags)

    self.old_role = old_role = self._F2R[old_flags]

    # Check for ineffective changes

    for attr in self._FLAGS:

      if getattr(self.op, attr) is False and getattr(node, attr) is False:

        self.LogInfo("Ignoring request to unset flag %s, already unset", attr)

        setattr(self.op, attr, None)

    # Past this point, any flag change to False means a transition

    # away from the respective state, as only real changes are kept

    # TODO: We might query the real power state if it supports OOB

    if SupportsOob(self.cfg, node):

      if self.op.offline is False and not (node.powered or

                                           self.op.powered is True):

        raise errors.OpPrereqError(("Node %s needs to be turned on before its"

                                    " offline status can be reset") %

                                   self.op.node_name, errors.ECODE_STATE)

    elif self.op.powered is not None:

      raise errors.OpPrereqError(("Unable to change powered state for node %s"

                                  " as it does not support out-of-band"

                                  " handling") % self.op.node_name,

                                 errors.ECODE_STATE)

    # If we're being deofflined/drained, we'll MC ourself if needed

    if (self.op.drained is False or self.op.offline is False or

        (self.op.master_capable and not node.master_capable)):

      if _DecideSelfPromotion(self):

        self.op.master_candidate = True

        self.LogInfo("Auto-promoting node to master candidate")

    # If we're no longer master capable, we'll demote ourselves from MC

    if self.op.master_capable is False and node.master_candidate:

      self.LogInfo("Demoting from master candidate")

      self.op.master_candidate = False

    # Compute new role

    assert [getattr(self.op, attr) for attr in self._FLAGS].count(True) <= 1

    if self.op.master_candidate:

      new_role = self._ROLE_CANDIDATE

    elif self.op.drained:

      new_role = self._ROLE_DRAINED

    elif self.op.offline:

      new_role = self._ROLE_OFFLINE

    elif False in [self.op.master_candidate, self.op.drained, self.op.offline]:

      # False is still in new flags, which means we're un-setting (the

      # only) True flag

      new_role = self._ROLE_REGULAR

    else: # no new flags, nothing, keep old role

      new_role = old_role

    self.new_role = new_role

    if old_role == self._ROLE_OFFLINE and new_role != old_role:

      # Trying to transition out of offline status

      result = self.rpc.call_version([node.uuid])[node.uuid]

      if result.fail_msg:

        raise errors.OpPrereqError("Node %s is being de-offlined but fails"

                                   " to report its version: %s" %

                                   (node.name, result.fail_msg),

                                   errors.ECODE_STATE)

      else:

        self.LogWarning("Transitioning node from offline to online state"

                        " without using re-add. Please make sure the node"

                        " is healthy!")

    # When changing the secondary ip, verify if this is a single-homed to

    # multi-homed transition or vice versa, and apply the relevant

    # restrictions.

    if self.op.secondary_ip:

      # Ok even without locking, because this can't be changed by any LU

      master = self.cfg.GetMasterNodeInfo()

      master_singlehomed = master.secondary_ip == master.primary_ip

      if master_singlehomed and self.op.secondary_ip != node.primary_ip:

        if self.op.force and node.uuid == master.uuid:

          self.LogWarning("Transitioning from single-homed to multi-homed"

                          " cluster; all nodes will require a secondary IP"

                          " address")

        else:

          raise errors.OpPrereqError("Changing the secondary ip on a"

                                     " single-homed cluster requires the"

                                     " --force option to be passed, and the"

                                     " target node to be the master",

                                     errors.ECODE_INVAL)

      elif not master_singlehomed and self.op.secondary_ip == node.primary_ip:

        if self.op.force and node.uuid == master.uuid:

          self.LogWarning("Transitioning from multi-homed to single-homed"

                          " cluster; secondary IP addresses will have to be"

                          " removed")

        else:

          raise errors.OpPrereqError("Cannot set the secondary IP to be the"

                                     " same as the primary IP on a multi-homed"

                                     " cluster, unless the --force option is"

                                     " passed, and the target node is the"

                                     " master", errors.ECODE_INVAL)

      assert not (set([inst.name for inst in affected_instances.values()]) -

                  self.owned_locks(locking.LEVEL_INSTANCE))

      if node.offline:

        if affected_instances:

          msg = ("Cannot change secondary IP address: offline node has"

                 " instances (%s) configured to use it" %

                 utils.CommaJoin(

                   [inst.name for inst in affected_instances.values()]))

          raise errors.OpPrereqError(msg, errors.ECODE_STATE)

      else:

        # On online nodes, check that no instances are running, and that

        # the node has the new ip and we can reach it.

        for instance in affected_instances.values():

          CheckInstanceState(self, instance, INSTANCE_DOWN,

                             msg="cannot change secondary ip")

        _CheckNodeHasSecondaryIP(self, node, self.op.secondary_ip, True)

        if master.uuid != node.uuid:

          # check reachability from master secondary ip to new secondary ip

          if not netutils.TcpPing(self.op.secondary_ip,

                                  constants.DEFAULT_NODED_PORT,

                                  source=master.secondary_ip):

            raise errors.OpPrereqError("Node secondary ip not reachable by TCP"

                                       " based ping to node daemon port",

                                       errors.ECODE_ENVIRON)

    if self.op.ndparams:

      new_ndparams = GetUpdatedParams(node.ndparams, self.op.ndparams)

      utils.ForceDictType(new_ndparams, constants.NDS_PARAMETER_TYPES)

      CheckParamsNotGlobal(self.op.ndparams, constants.NDC_GLOBALS, "node",

                           "node", "cluster or group")

      self.new_ndparams = new_ndparams

    if self.op.hv_state:

      self.new_hv_state = MergeAndVerifyHvState(self.op.hv_state,

                                                node.hv_state_static)

    if self.op.disk_state:

      self.new_disk_state = \

        MergeAndVerifyDiskState(self.op.disk_state, node.disk_state_static)

  def Exec(self, feedback_fn):

    """Modifies a node.

    """

    node = self.cfg.GetNodeInfo(self.op.node_uuid)

    result = []

    if self.op.ndparams:

      node.ndparams = self.new_ndparams

    if self.op.powered is not None:

      node.powered = self.op.powered

    if self.op.hv_state:

      node.hv_state_static = self.new_hv_state

    if self.op.disk_state:

      node.disk_state_static = self.new_disk_state

    for attr in ["master_capable", "vm_capable"]:

      val = getattr(self.op, attr)

      if val is not None:

        setattr(node, attr, val)

        result.append((attr, str(val)))

    if self.new_role != self.old_role:

      # Tell the node to demote itself, if no longer MC and not offline

      if self.old_role == self._ROLE_CANDIDATE and \

          self.new_role != self._ROLE_OFFLINE:

        msg = self.rpc.call_node_demote_from_mc(node.name).fail_msg

        if msg:

          self.LogWarning("Node failed to demote itself: %s", msg)

      new_flags = self._R2F[self.new_role]

      for of, nf, desc in zip(self.old_flags, new_flags, self._FLAGS):

        if of != nf:

          result.append((desc, str(nf)))

      (node.master_candidate, node.drained, node.offline) = new_flags

      # we locked all nodes, we adjust the CP before updating this node

      if self.lock_all:

        AdjustCandidatePool(self, [node.uuid])

    if self.op.secondary_ip:

      node.secondary_ip = self.op.secondary_ip

      result.append(("secondary_ip", self.op.secondary_ip))

    # this will trigger configuration file update, if needed

    self.cfg.Update(node, feedback_fn)

    # this will trigger job queue propagation or cleanup if the mc

    # flag changed

    if [self.old_role, self.new_role].count(self._ROLE_CANDIDATE) == 1:

      self.context.ReaddNode(node)

    return result

class LUNodePowercycle(NoHooksLU):

  """Powercycles a node.

  """

  REQ_BGL = False

  def CheckArguments(self):

    (self.op.node_uuid, self.op.node_name) = \

      ExpandNodeUuidAndName(self.cfg, self.op.node_uuid, self.op.node_name)

    if self.op.node_uuid == self.cfg.GetMasterNode() and not self.op.force:

      raise errors.OpPrereqError("The node is the master and the force"

                                 " parameter was not set",

                                 errors.ECODE_INVAL)

  def ExpandNames(self):

    """Locking for PowercycleNode.

    This is a last-resort option and shouldn't block on other

    jobs. Therefore, we grab no locks.

    """

    self.needed_locks = {}

  def Exec(self, feedback_fn):

    """Reboots a node.

    """

    default_hypervisor = self.cfg.GetHypervisorType()

    hvparams = self.cfg.GetClusterInfo().hvparams[default_hypervisor]

    result = self.rpc.call_node_powercycle(self.op.node_uuid,

                                           default_hypervisor,

                                           hvparams)

    result.Raise("Failed to schedule the reboot")

    return result.payload

def _GetNodeInstancesInner(cfg, fn):

  return [i for i in cfg.GetAllInstancesInfo().values() if fn(i)]

def _GetNodePrimaryInstances(cfg, node_uuid):

  """Returns primary instances on a node.

  """

  return _GetNodeInstancesInner(cfg,

                                lambda inst: node_uuid == inst.primary_node)

def _GetNodeSecondaryInstances(cfg, node_uuid):

  """Returns secondary instances on a node.

  """

  return _GetNodeInstancesInner(cfg,

                                lambda inst: node_uuid in inst.secondary_nodes)

def _GetNodeInstances(cfg, node_uuid):

  """Returns a list of all primary and secondary instances on a node.

  """

  return _GetNodeInstancesInner(cfg, lambda inst: node_uuid in inst.all_nodes)

class LUNodeEvacuate(NoHooksLU):

  """Evacuates instances off a list of nodes.

  """

  REQ_BGL = False

  def CheckArguments(self):

    CheckIAllocatorOrNode(self, "iallocator", "remote_node")

  def ExpandNames(self):

    (self.op.node_uuid, self.op.node_name) = \

      ExpandNodeUuidAndName(self.cfg, self.op.node_uuid, self.op.node_name)

    if self.op.remote_node is not None:

      (self.op.remote_node_uuid, self.op.remote_node) = \

        ExpandNodeUuidAndName(self.cfg, self.op.remote_node_uuid,

                              self.op.remote_node)

      assert self.op.remote_node

      if self.op.node_uuid == self.op.remote_node_uuid:

        raise errors.OpPrereqError("Can not use evacuated node as a new"

                                   " secondary node", errors.ECODE_INVAL)

      if self.op.mode != constants.NODE_EVAC_SEC:

        raise errors.OpPrereqError("Without the use of an iallocator only"

                                   " secondary instances can be evacuated",

                                   errors.ECODE_INVAL)

    # Declare locks

    self.share_locks = ShareAll()

    self.needed_locks = {

      locking.LEVEL_INSTANCE: [],

      locking.LEVEL_NODEGROUP: [],

      locking.LEVEL_NODE: [],

      }

    # Determine nodes (via group) optimistically, needs verification once locks

    # have been acquired

    self.lock_nodes = self._DetermineNodes()

  def _DetermineNodes(self):

    """Gets the list of node UUIDs to operate on.

    """

    if self.op.remote_node is None:

      # Iallocator will choose any node(s) in the same group

      group_nodes = self.cfg.GetNodeGroupMembersByNodes([self.op.node_uuid])

    else:

      group_nodes = frozenset([self.op.remote_node_uuid])

    # Determine nodes to be locked

    return set([self.op.node_uuid]) | group_nodes

  def _DetermineInstances(self):

    """Builds list of instances to operate on.

    """

    assert self.op.mode in constants.NODE_EVAC_MODES

    if self.op.mode == constants.NODE_EVAC_PRI:

      # Primary instances only

      inst_fn = _GetNodePrimaryInstances

      assert self.op.remote_node is None, \

        "Evacuating primary instances requires iallocator"

    elif self.op.mode == constants.NODE_EVAC_SEC:

      # Secondary instances only

      inst_fn = _GetNodeSecondaryInstances

    else:

      # All instances

      assert self.op.mode == constants.NODE_EVAC_ALL

      inst_fn = _GetNodeInstances

      # TODO: In 2.6, change the iallocator interface to take an evacuation mode

      # per instance

      raise errors.OpPrereqError("Due to an issue with the iallocator"

                                 " interface it is not possible to evacuate"

                                 " all instances at once; specify explicitly"

                                 " whether to evacuate primary or secondary"

                                 " instances",

                                 errors.ECODE_INVAL)

    return inst_fn(self.cfg, self.op.node_uuid)

  def DeclareLocks(self, level):

    if level == locking.LEVEL_INSTANCE:

      # Lock instances optimistically, needs verification once node and group

      # locks have been acquired

      self.needed_locks[locking.LEVEL_INSTANCE] = \

        set(i.name for i in self._DetermineInstances())

    elif level == locking.LEVEL_NODEGROUP:

      # Lock node groups for all potential target nodes optimistically, needs

      # verification once nodes have been acquired

      self.needed_locks[locking.LEVEL_NODEGROUP] = \

        self.cfg.GetNodeGroupsFromNodes(self.lock_nodes)

    elif level == locking.LEVEL_NODE:

      self.needed_locks[locking.LEVEL_NODE] = self.lock_nodes

  def CheckPrereq(self):

    # Verify locks

    owned_instance_names = self.owned_locks(locking.LEVEL_INSTANCE)

    owned_nodes = self.owned_locks(locking.LEVEL_NODE)

    owned_groups = self.owned_locks(locking.LEVEL_NODEGROUP)

    need_nodes = self._DetermineNodes()

    if not owned_nodes.issuperset(need_nodes):

      raise errors.OpPrereqError("Nodes in same group as '%s' changed since"

                                 " locks were acquired, current nodes are"

                                 " are '%s', used to be '%s'; retry the"

                                 " operation" %

                                 (self.op.node_name,

                                  utils.CommaJoin(need_nodes),

                                  utils.CommaJoin(owned_nodes)),

                                 errors.ECODE_STATE)

    wanted_groups = self.cfg.GetNodeGroupsFromNodes(owned_nodes)

    if owned_groups != wanted_groups:

      raise errors.OpExecError("Node groups changed since locks were acquired,"

                               " current groups are '%s', used to be '%s';"

                               " retry the operation" %

                               (utils.CommaJoin(wanted_groups),

                                utils.CommaJoin(owned_groups)))

    # Determine affected instances

    self.instances = self._DetermineInstances()

    self.instance_names = [i.name for i in self.instances]

    if set(self.instance_names) != owned_instance_names:

      raise errors.OpExecError("Instances on node '%s' changed since locks"

                               " were acquired, current instances are '%s',"

                               " used to be '%s'; retry the operation" %

                               (self.op.node_name,

                                utils.CommaJoin(self.instance_names),

                                utils.CommaJoin(owned_instance_names)))

    if self.instance_names:

      self.LogInfo("Evacuating instances from node '%s': %s",

                   self.op.node_name,

                   utils.CommaJoin(utils.NiceSort(self.instance_names)))

    else:

      self.LogInfo("No instances to evacuate from node '%s'",

                   self.op.node_name)

    if self.op.remote_node is not None:

      for i in self.instances:

        if i.primary_node == self.op.remote_node_uuid:

          raise errors.OpPrereqError("Node %s is the primary node of"

                                     " instance %s, cannot use it as"

                                     " secondary" %

                                     (self.op.remote_node, i.name),

                                     errors.ECODE_INVAL)

  def Exec(self, feedback_fn):

    assert (self.op.iallocator is not None) ^ (self.op.remote_node is not None)

    if not self.instance_names:

      # No instances to evacuate

      jobs = []

    elif self.op.iallocator is not None:

      # TODO: Implement relocation to other group

      req = iallocator.IAReqNodeEvac(evac_mode=self.op.mode,

                                     instances=list(self.instance_names))

      ial = iallocator.IAllocator(self.cfg, self.rpc, req)

      ial.Run(self.op.iallocator)

      if not ial.success:

        raise errors.OpPrereqError("Can't compute node evacuation using"

                                   " iallocator '%s': %s" %

                                   (self.op.iallocator, ial.info),

                                   errors.ECODE_NORES)

      jobs = LoadNodeEvacResult(self, ial.result, self.op.early_release, True)

    elif self.op.remote_node is not None:

      assert self.op.mode == constants.NODE_EVAC_SEC

      jobs = [

        [opcodes.OpInstanceReplaceDisks(instance_name=instance_name,

                                        remote_node=self.op.remote_node,

                                        disks=[],

                                        mode=constants.REPLACE_DISK_CHG,

                                        early_release=self.op.early_release)]

        for instance_name in self.instance_names]

    else:

      raise errors.ProgrammerError("No iallocator or remote node")

    return ResultWithJobs(jobs)

class LUNodeMigrate(LogicalUnit):

  """Migrate all instances from a node.

  """

  HPATH = "node-migrate"

  HTYPE = constants.HTYPE_NODE

  REQ_BGL = False

  def CheckArguments(self):

    pass

  def ExpandNames(self):

    (self.op.node_uuid, self.op.node_name) = \

      ExpandNodeUuidAndName(self.cfg, self.op.node_uuid, self.op.node_name)

    self.share_locks = ShareAll()

    self.needed_locks = {

      locking.LEVEL_NODE: [self.op.node_uuid],

      }

  def BuildHooksEnv(self):

    """Build hooks env.

    This runs on the master, the primary and all the secondaries.

    """

    return {

      "NODE_NAME": self.op.node_name,

      "ALLOW_RUNTIME_CHANGES": self.op.allow_runtime_changes,

      }

  def BuildHooksNodes(self):

    """Build hooks nodes.

    """

    nl = [self.cfg.GetMasterNode()]

    return (nl, nl)

  def CheckPrereq(self):

    pass

  def Exec(self, feedback_fn):

    # Prepare jobs for migration instances

    jobs = [

      [opcodes.OpInstanceMigrate(

        instance_name=inst.name,

        mode=self.op.mode,

        live=self.op.live,

        iallocator=self.op.iallocator,

        target_node=self.op.target_node,

        allow_runtime_changes=self.op.allow_runtime_changes,

        ignore_ipolicy=self.op.ignore_ipolicy)]

      for inst in _GetNodePrimaryInstances(self.cfg, self.op.node_uuid)]

    # TODO: Run iallocator in this opcode and pass correct placement options to

    # OpInstanceMigrate. Since other jobs can modify the cluster between

    # running the iallocator and the actual migration, a good consistency model

    # will have to be found.