Synnefo » snf-ganeti

#

#

# Copyright (C) 2006, 2007, 2008 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.

"""Module implementing the master-side code."""

# pylint: disable-msg=W0613,W0201

import os

import os.path

import sha

import time

import tempfile

import re

import platform

import logging

from ganeti import rpc

from ganeti import ssh

from ganeti import logger

from ganeti import utils

from ganeti import errors

from ganeti import hypervisor

from ganeti import locking

from ganeti import constants

from ganeti import objects

from ganeti import opcodes

from ganeti import serializer

class LogicalUnit(object):

  """Logical Unit base class.

  Subclasses must follow these rules:

    - implement ExpandNames

    - implement CheckPrereq

    - implement Exec

    - implement BuildHooksEnv

    - redefine HPATH and HTYPE

    - optionally redefine their run requirements:

        REQ_MASTER: the LU needs to run on the master node

        REQ_BGL: the LU needs to hold the Big Ganeti Lock exclusively

  Note that all commands require root permissions.

  """

  HPATH = None

  HTYPE = None

  _OP_REQP = []

  REQ_MASTER = True

  REQ_BGL = True

  def __init__(self, processor, op, context):

    """Constructor for LogicalUnit.

    This needs to be overriden in derived classes in order to check op

    validity.

    """

    self.proc = processor

    self.op = op

    self.cfg = context.cfg

    self.context = context

    # Dicts used to declare locking needs to mcpu

    self.needed_locks = None

    self.acquired_locks = {}

    self.share_locks = dict(((i, 0) for i in locking.LEVELS))

    self.add_locks = {}

    self.remove_locks = {}

    # Used to force good behavior when calling helper functions

    self.recalculate_locks = {}

    self.__ssh = None

    for attr_name in self._OP_REQP:

      attr_val = getattr(op, attr_name, None)

      if attr_val is None:

        raise errors.OpPrereqError("Required parameter '%s' missing" %

                                   attr_name)

    if not self.cfg.IsCluster():

      raise errors.OpPrereqError("Cluster not initialized yet,"

                                 " use 'gnt-cluster init' first.")

    if self.REQ_MASTER:

      master = self.cfg.GetMasterNode()

      if master != utils.HostInfo().name:

        raise errors.OpPrereqError("Commands must be run on the master"

                                   " node %s" % master)

  def __GetSSH(self):

    """Returns the SshRunner object

    """

    if not self.__ssh:

      self.__ssh = ssh.SshRunner(self.cfg.GetClusterName())

    return self.__ssh

  ssh = property(fget=__GetSSH)

  def ExpandNames(self):

    """Expand names for this LU.

    This method is called before starting to execute the opcode, and it should

    update all the parameters of the opcode to their canonical form (e.g. a

    short node name must be fully expanded after this method has successfully

    completed). This way locking, hooks, logging, ecc. can work correctly.

    LUs which implement this method must also populate the self.needed_locks

    member, as a dict with lock levels as keys, and a list of needed lock names

    as values. Rules:

      - Use an empty dict if you don't need any lock

      - If you don't need any lock at a particular level omit that level

      - Don't put anything for the BGL level

      - If you want all locks at a level use locking.ALL_SET as a value

    If you need to share locks (rather than acquire them exclusively) at one

    level you can modify self.share_locks, setting a true value (usually 1) for

    that level. By default locks are not shared.

    Examples:

    # Acquire all nodes and one instance

    self.needed_locks = {

      locking.LEVEL_NODE: locking.ALL_SET,

      locking.LEVEL_INSTANCE: ['instance1.example.tld'],

    }

    # Acquire just two nodes

    self.needed_locks = {

      locking.LEVEL_NODE: ['node1.example.tld', 'node2.example.tld'],

    }

    # Acquire no locks

    self.needed_locks = {} # No, you can't leave it to the default value None

    """

    # The implementation of this method is mandatory only if the new LU is

    # concurrent, so that old LUs don't need to be changed all at the same

    # time.

    if self.REQ_BGL:

      self.needed_locks = {} # Exclusive LUs don't need locks.

    else:

      raise NotImplementedError

  def DeclareLocks(self, level):

    """Declare LU locking needs for a level

    While most LUs can just declare their locking needs at ExpandNames time,

    sometimes there's the need to calculate some locks after having acquired

    the ones before. This function is called just before acquiring locks at a

    particular level, but after acquiring the ones at lower levels, and permits

    such calculations. It can be used to modify self.needed_locks, and by

    default it does nothing.

    This function is only called if you have something already set in

    self.needed_locks for the level.

    @param level: Locking level which is going to be locked

    @type level: member of ganeti.locking.LEVELS

    """

  def CheckPrereq(self):

    """Check prerequisites for this LU.

    This method should check that the prerequisites for the execution

    of this LU are fulfilled. It can do internode communication, but

    it should be idempotent - no cluster or system changes are

    allowed.

    The method should raise errors.OpPrereqError in case something is

    not fulfilled. Its return value is ignored.

    This method should also update all the parameters of the opcode to

    their canonical form if it hasn't been done by ExpandNames before.

    """

    raise NotImplementedError

  def Exec(self, feedback_fn):

    """Execute the LU.

    This method should implement the actual work. It should raise

    errors.OpExecError for failures that are somewhat dealt with in

    code, or expected.

    """

    raise NotImplementedError

  def BuildHooksEnv(self):

    """Build hooks environment for this LU.

    This method should return a three-node tuple consisting of: a dict

    containing the environment that will be used for running the

    specific hook for this LU, a list of node names on which the hook

    should run before the execution, and a list of node names on which

    the hook should run after the execution.

    The keys of the dict must not have 'GANETI_' prefixed as this will

    be handled in the hooks runner. Also note additional keys will be

    added by the hooks runner. If the LU doesn't define any

    environment, an empty dict (and not None) should be returned.

    No nodes should be returned as an empty list (and not None).

    Note that if the HPATH for a LU class is None, this function will

    not be called.

    """

    raise NotImplementedError

  def HooksCallBack(self, phase, hook_results, feedback_fn, lu_result):

    """Notify the LU about the results of its hooks.

    This method is called every time a hooks phase is executed, and notifies

    the Logical Unit about the hooks' result. The LU can then use it to alter

    its result based on the hooks.  By default the method does nothing and the

    previous result is passed back unchanged but any LU can define it if it

    wants to use the local cluster hook-scripts somehow.

    Args:

      phase: the hooks phase that has just been run

      hooks_results: the results of the multi-node hooks rpc call

      feedback_fn: function to send feedback back to the caller

      lu_result: the previous result this LU had, or None in the PRE phase.

    """

    return lu_result

  def _ExpandAndLockInstance(self):

    """Helper function to expand and lock an instance.

    Many LUs that work on an instance take its name in self.op.instance_name

    and need to expand it and then declare the expanded name for locking. This

    function does it, and then updates self.op.instance_name to the expanded

    name. It also initializes needed_locks as a dict, if this hasn't been done

    before.

    """

    if self.needed_locks is None:

      self.needed_locks = {}

    else:

      assert locking.LEVEL_INSTANCE not in self.needed_locks, \

        "_ExpandAndLockInstance called with instance-level locks set"

    expanded_name = self.cfg.ExpandInstanceName(self.op.instance_name)

    if expanded_name is None:

      raise errors.OpPrereqError("Instance '%s' not known" %

                                  self.op.instance_name)

    self.needed_locks[locking.LEVEL_INSTANCE] = expanded_name

    self.op.instance_name = expanded_name

  def _LockInstancesNodes(self, primary_only=False):

    """Helper function to declare instances' nodes for locking.

    This function should be called after locking one or more instances to lock

    their nodes. Its effect is populating self.needed_locks[locking.LEVEL_NODE]

    with all primary or secondary nodes for instances already locked and

    present in self.needed_locks[locking.LEVEL_INSTANCE].

    It should be called from DeclareLocks, and for safety only works if

    self.recalculate_locks[locking.LEVEL_NODE] is set.

    In the future it may grow parameters to just lock some instance's nodes, or

    to just lock primaries or secondary nodes, if needed.

    If should be called in DeclareLocks in a way similar to:

    if level == locking.LEVEL_NODE:

      self._LockInstancesNodes()

    @type primary_only: boolean

    @param primary_only: only lock primary nodes of locked instances

    """

    assert locking.LEVEL_NODE in self.recalculate_locks, \

      "_LockInstancesNodes helper function called with no nodes to recalculate"

    # TODO: check if we're really been called with the instance locks held

    # For now we'll replace self.needed_locks[locking.LEVEL_NODE], but in the

    # future we might want to have different behaviors depending on the value

    # of self.recalculate_locks[locking.LEVEL_NODE]

    wanted_nodes = []

    for instance_name in self.acquired_locks[locking.LEVEL_INSTANCE]:

      instance = self.context.cfg.GetInstanceInfo(instance_name)

      wanted_nodes.append(instance.primary_node)

      if not primary_only:

        wanted_nodes.extend(instance.secondary_nodes)

    if self.recalculate_locks[locking.LEVEL_NODE] == constants.LOCKS_REPLACE:

      self.needed_locks[locking.LEVEL_NODE] = wanted_nodes

    elif self.recalculate_locks[locking.LEVEL_NODE] == constants.LOCKS_APPEND:

      self.needed_locks[locking.LEVEL_NODE].extend(wanted_nodes)

    del self.recalculate_locks[locking.LEVEL_NODE]

class NoHooksLU(LogicalUnit):

  """Simple LU which runs no hooks.

  This LU is intended as a parent for other LogicalUnits which will

  run no hooks, in order to reduce duplicate code.

  """

  HPATH = None

  HTYPE = None

def _GetWantedNodes(lu, nodes):

  """Returns list of checked and expanded node names.

  Args:

    nodes: List of nodes (strings) or None for all

  """

  if not isinstance(nodes, list):

    raise errors.OpPrereqError("Invalid argument type 'nodes'")

  if not nodes:

    raise errors.ProgrammerError("_GetWantedNodes should only be called with a"

      " non-empty list of nodes whose name is to be expanded.")

  wanted = []

  for name in nodes:

    node = lu.cfg.ExpandNodeName(name)

    if node is None:

      raise errors.OpPrereqError("No such node name '%s'" % name)

    wanted.append(node)

  return utils.NiceSort(wanted)

def _GetWantedInstances(lu, instances):

  """Returns list of checked and expanded instance names.

  Args:

    instances: List of instances (strings) or None for all

  """

  if not isinstance(instances, list):

    raise errors.OpPrereqError("Invalid argument type 'instances'")

  if instances:

    wanted = []

    for name in instances:

      instance = lu.cfg.ExpandInstanceName(name)

      if instance is None:

        raise errors.OpPrereqError("No such instance name '%s'" % name)

      wanted.append(instance)

  else:

    wanted = lu.cfg.GetInstanceList()

  return utils.NiceSort(wanted)

def _CheckOutputFields(static, dynamic, selected):

  """Checks whether all selected fields are valid.

  Args:

    static: Static fields

    dynamic: Dynamic fields

  """

  static_fields = frozenset(static)

  dynamic_fields = frozenset(dynamic)

  all_fields = static_fields | dynamic_fields

  if not all_fields.issuperset(selected):

    raise errors.OpPrereqError("Unknown output fields selected: %s"

                               % ",".join(frozenset(selected).

                                          difference(all_fields)))

def _BuildInstanceHookEnv(name, primary_node, secondary_nodes, os_type, status,

                          memory, vcpus, nics):

  """Builds instance related env variables for hooks from single variables.

  Args:

    secondary_nodes: List of secondary nodes as strings

  """

  env = {

    "OP_TARGET": name,

    "INSTANCE_NAME": name,

    "INSTANCE_PRIMARY": primary_node,

    "INSTANCE_SECONDARIES": " ".join(secondary_nodes),

    "INSTANCE_OS_TYPE": os_type,

    "INSTANCE_STATUS": status,

    "INSTANCE_MEMORY": memory,

    "INSTANCE_VCPUS": vcpus,

  }

  if nics:

    nic_count = len(nics)

    for idx, (ip, bridge, mac) in enumerate(nics):

      if ip is None:

        ip = ""

      env["INSTANCE_NIC%d_IP" % idx] = ip

      env["INSTANCE_NIC%d_BRIDGE" % idx] = bridge

      env["INSTANCE_NIC%d_HWADDR" % idx] = mac

  else:

    nic_count = 0

  env["INSTANCE_NIC_COUNT"] = nic_count

  return env

def _BuildInstanceHookEnvByObject(instance, override=None):

  """Builds instance related env variables for hooks from an object.

  Args:

    instance: objects.Instance object of instance

    override: dict of values to override

  """

  args = {

    'name': instance.name,

    'primary_node': instance.primary_node,

    'secondary_nodes': instance.secondary_nodes,

    'os_type': instance.os,

    'status': instance.os,

    'memory': instance.memory,

    'vcpus': instance.vcpus,

    'nics': [(nic.ip, nic.bridge, nic.mac) for nic in instance.nics],

  }

  if override:

    args.update(override)

  return _BuildInstanceHookEnv(**args)

def _CheckInstanceBridgesExist(instance):

  """Check that the brigdes needed by an instance exist.

  """

  # check bridges existance

  brlist = [nic.bridge for nic in instance.nics]

  if not rpc.call_bridges_exist(instance.primary_node, brlist):

    raise errors.OpPrereqError("one or more target bridges %s does not"

                               " exist on destination node '%s'" %

                               (brlist, instance.primary_node))

class LUDestroyCluster(NoHooksLU):

  """Logical unit for destroying the cluster.

  """

  _OP_REQP = []

  def CheckPrereq(self):

    """Check prerequisites.

    This checks whether the cluster is empty.

    Any errors are signalled 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))

    instancelist = self.cfg.GetInstanceList()

    if instancelist:

      raise errors.OpPrereqError("There are still %d instance(s) in"

                                 " this cluster." % len(instancelist))

  def Exec(self, feedback_fn):

    """Destroys the cluster.

    """

    master = self.cfg.GetMasterNode()

    if not rpc.call_node_stop_master(master, False):

      raise errors.OpExecError("Could not disable the master role")

    priv_key, pub_key, _ = ssh.GetUserFiles(constants.GANETI_RUNAS)

    utils.CreateBackup(priv_key)

    utils.CreateBackup(pub_key)

    return master

class LUVerifyCluster(LogicalUnit):

  """Verifies the cluster status.

  """

  HPATH = "cluster-verify"

  HTYPE = constants.HTYPE_CLUSTER

  _OP_REQP = ["skip_checks"]

  REQ_BGL = False

  def ExpandNames(self):

    self.needed_locks = {

      locking.LEVEL_NODE: locking.ALL_SET,

      locking.LEVEL_INSTANCE: locking.ALL_SET,

    }

    self.share_locks = dict(((i, 1) for i in locking.LEVELS))

  def _VerifyNode(self, node, file_list, local_cksum, vglist, node_result,

                  remote_version, feedback_fn):

    """Run multiple tests against a node.

    Test list:

      - compares ganeti version

      - checks vg existance and size > 20G

      - checks config file checksum

      - checks ssh to other nodes

    Args:

      node: name of the node to check

      file_list: required list of files

      local_cksum: dictionary of local files and their checksums

    """

    # compares ganeti version

    local_version = constants.PROTOCOL_VERSION

    if not remote_version:

      feedback_fn("  - ERROR: connection to %s failed" % (node))

      return True

    if local_version != remote_version:

      feedback_fn("  - ERROR: sw version mismatch: master %s, node(%s) %s" %

                      (local_version, node, remote_version))

      return True

    # checks vg existance and size > 20G

    bad = False

    if not vglist:

      feedback_fn("  - ERROR: unable to check volume groups on node %s." %

                      (node,))

      bad = True

    else:

      vgstatus = utils.CheckVolumeGroupSize(vglist, self.cfg.GetVGName(),

                                            constants.MIN_VG_SIZE)

      if vgstatus:

        feedback_fn("  - ERROR: %s on node %s" % (vgstatus, node))

        bad = True

    # checks config file checksum

    # checks ssh to any

    if 'filelist' not in node_result:

      bad = True

      feedback_fn("  - ERROR: node hasn't returned file checksum data")

    else:

      remote_cksum = node_result['filelist']

      for file_name in file_list:

        if file_name not in remote_cksum:

          bad = True

          feedback_fn("  - ERROR: file '%s' missing" % file_name)

        elif remote_cksum[file_name] != local_cksum[file_name]:

          bad = True

          feedback_fn("  - ERROR: file '%s' has wrong checksum" % file_name)

    if 'nodelist' not in node_result:

      bad = True

      feedback_fn("  - ERROR: node hasn't returned node ssh connectivity data")

    else:

      if node_result['nodelist']:

        bad = True

        for node in node_result['nodelist']:

          feedback_fn("  - ERROR: ssh communication with node '%s': %s" %

                          (node, node_result['nodelist'][node]))

    if 'node-net-test' not in node_result:

      bad = True

      feedback_fn("  - ERROR: node hasn't returned node tcp connectivity data")

    else:

      if node_result['node-net-test']:

        bad = True

        nlist = utils.NiceSort(node_result['node-net-test'].keys())

        for node in nlist:

          feedback_fn("  - ERROR: tcp communication with node '%s': %s" %

                          (node, node_result['node-net-test'][node]))

    hyp_result = node_result.get('hypervisor', None)

    if isinstance(hyp_result, dict):

      for hv_name, hv_result in hyp_result.iteritems():

        if hv_result is not None:

          feedback_fn("  - ERROR: hypervisor %s verify failure: '%s'" %

                      (hv_name, hv_result))

    return bad

  def _VerifyInstance(self, instance, instanceconfig, node_vol_is,

                      node_instance, feedback_fn):

    """Verify an instance.

    This function checks to see if the required block devices are

    available on the instance's node.

    """

    bad = False

    node_current = instanceconfig.primary_node

    node_vol_should = {}

    instanceconfig.MapLVsByNode(node_vol_should)

    for node in node_vol_should:

      for volume in node_vol_should[node]:

        if node not in node_vol_is or volume not in node_vol_is[node]:

          feedback_fn("  - ERROR: volume %s missing on node %s" %

                          (volume, node))

          bad = True

    if not instanceconfig.status == 'down':

      if (node_current not in node_instance or

          not instance in node_instance[node_current]):

        feedback_fn("  - ERROR: instance %s not running on node %s" %

                        (instance, node_current))

        bad = True

    for node in node_instance:

      if (not node == node_current):

        if instance in node_instance[node]:

          feedback_fn("  - ERROR: instance %s should not run on node %s" %

                          (instance, node))

          bad = True

    return bad

  def _VerifyOrphanVolumes(self, node_vol_should, node_vol_is, feedback_fn):

    """Verify if there are any unknown volumes in the cluster.

    The .os, .swap and backup volumes are ignored. All other volumes are

    reported as unknown.

    """

    bad = False

    for node in node_vol_is:

      for volume in node_vol_is[node]:

        if node not in node_vol_should or volume not in node_vol_should[node]:

          feedback_fn("  - ERROR: volume %s on node %s should not exist" %

                      (volume, node))

          bad = True

    return bad

  def _VerifyOrphanInstances(self, instancelist, node_instance, feedback_fn):

    """Verify the list of running instances.

    This checks what instances are running but unknown to the cluster.

    """

    bad = False

    for node in node_instance:

      for runninginstance in node_instance[node]:

        if runninginstance not in instancelist:

          feedback_fn("  - ERROR: instance %s on node %s should not exist" %

                          (runninginstance, node))

          bad = True

    return bad

  def _VerifyNPlusOneMemory(self, node_info, instance_cfg, feedback_fn):

    """Verify N+1 Memory Resilience.

    Check that if one single node dies we can still start all the instances it

    was primary for.

    """

    bad = False

    for node, nodeinfo in node_info.iteritems():

      # This code checks that every node which is now listed as secondary has

      # enough memory to host all instances it is supposed to should a single

      # other node in the cluster fail.

      # FIXME: not ready for failover to an arbitrary node

      # FIXME: does not support file-backed instances

      # WARNING: we currently take into account down instances as well as up

      # ones, considering that even if they're down someone might want to start

      # them even in the event of a node failure.

      for prinode, instances in nodeinfo['sinst-by-pnode'].iteritems():

        needed_mem = 0

        for instance in instances:

          needed_mem += instance_cfg[instance].memory

        if nodeinfo['mfree'] < needed_mem:

          feedback_fn("  - ERROR: not enough memory on node %s to accomodate"

                      " failovers should node %s fail" % (node, prinode))

          bad = True

    return bad

  def CheckPrereq(self):

    """Check prerequisites.

    Transform the list of checks we're going to skip into a set and check that

    all its members are valid.

    """

    self.skip_set = frozenset(self.op.skip_checks)

    if not constants.VERIFY_OPTIONAL_CHECKS.issuperset(self.skip_set):

      raise errors.OpPrereqError("Invalid checks to be skipped specified")

  def BuildHooksEnv(self):

    """Build hooks env.

    Cluster-Verify hooks just rone in the post phase and their failure makes

    the output be logged in the verify output and the verification to fail.

    """

    all_nodes = self.cfg.GetNodeList()

    # TODO: populate the environment with useful information for verify hooks

    env = {}

    return env, [], all_nodes

  def Exec(self, feedback_fn):

    """Verify integrity of cluster, performing various test on nodes.

    """

    bad = False

    feedback_fn("* Verifying global settings")

    for msg in self.cfg.VerifyConfig():

      feedback_fn("  - ERROR: %s" % msg)

    vg_name = self.cfg.GetVGName()

    hypervisors = self.cfg.GetClusterInfo().enabled_hypervisors

    nodelist = utils.NiceSort(self.cfg.GetNodeList())

    nodeinfo = [self.cfg.GetNodeInfo(nname) for nname in nodelist]

    instancelist = utils.NiceSort(self.cfg.GetInstanceList())

    i_non_redundant = [] # Non redundant instances

    node_volume = {}

    node_instance = {}

    node_info = {}