nfs.py

# Copyright (c) 2012 NetApp, Inc.
# Copyright (c) 2016 Red Hat, Inc.
# All Rights Reserved.
#
#    Licensed under the Apache License, Version 2.0 (the "License"); you may
#    not use this file except in compliance with the License. You may obtain
#    a copy of the License at
#
#         http://www.apache.org/licenses/LICENSE-2.0
#
#    Unless required by applicable law or agreed to in writing, software
#    distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
#    WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
#    License for the specific language governing permissions and limitations
#    under the License.

import binascii
import errno
import os
import tempfile
import time

from castellan import key_manager
from os_brick.remotefs import remotefs as remotefs_brick
from oslo_concurrency import processutils as putils
from oslo_config import cfg
from oslo_log import log as logging
from oslo_utils import units
import six

from cinder import context
from cinder import coordination
from cinder import exception
from cinder.i18n import _
from cinder.image import image_utils
from cinder import interface
from cinder import objects
from cinder import utils
from cinder.volume import configuration
from cinder.volume.drivers import remotefs
from cinder.volume import volume_utils

VERSION = '1.4.0'

LOG = logging.getLogger(__name__)


nfs_opts = [
    cfg.StrOpt('nfs_shares_config',
               default='/etc/cinder/nfs_shares',
               help='File with the list of available NFS shares.'),
    cfg.BoolOpt('nfs_sparsed_volumes',
                default=True,
                help='Create volumes as sparsed files which take no space. '
                     'If set to False volume is created as regular file. '
                     'In such case volume creation takes a lot of time.'),
    cfg.BoolOpt('nfs_qcow2_volumes',
                default=False,
                help='Create volumes as QCOW2 files rather than raw files.'),
    cfg.StrOpt('nfs_mount_point_base',
               default='$state_path/mnt',
               help='Base dir containing mount points for NFS shares.'),
    cfg.StrOpt('nfs_mount_options',
               help='Mount options passed to the NFS client. See the NFS(5) '
                    'man page for details.'),
    cfg.IntOpt('nfs_mount_attempts',
               default=3,
               help='The number of attempts to mount NFS shares before '
                    'raising an error.  At least one attempt will be '
                    'made to mount an NFS share, regardless of the '
                    'value specified.'),
    cfg.BoolOpt('nfs_snapshot_support',
                default=False,
                help='Enable support for snapshots on the NFS driver. '
                     'Platforms using libvirt <1.2.7 will encounter issues '
                     'with this feature.'),
]

CONF = cfg.CONF
CONF.register_opts(nfs_opts, group=configuration.SHARED_CONF_GROUP)


@interface.volumedriver
class NfsDriver(remotefs.RemoteFSSnapDriverDistributed):
    """NFS based cinder driver.

    Creates file on NFS share for using it as block device on hypervisor.
    """

    driver_volume_type = 'nfs'
    driver_prefix = 'nfs'
    volume_backend_name = 'Generic_NFS'
    VERSION = VERSION

    # ThirdPartySystems wiki page
    CI_WIKI_NAME = "Cinder_Jenkins"

    def __init__(self, execute=putils.execute, *args, **kwargs):
        self._remotefsclient = None
        super(NfsDriver, self).__init__(*args, **kwargs)
        self.configuration.append_config_values(nfs_opts)
        root_helper = utils.get_root_helper()
        # base bound to instance is used in RemoteFsConnector.
        self.base = getattr(self.configuration,
                            'nfs_mount_point_base')
        self.base = os.path.realpath(self.base)
        opts = getattr(self.configuration,
                       'nfs_mount_options')

        nas_mount_options = getattr(self.configuration,
                                    'nas_mount_options',
                                    None)
        if nas_mount_options is not None:
            LOG.debug('overriding nfs_mount_options with nas_mount_options')
            opts = nas_mount_options

        self._remotefsclient = remotefs_brick.RemoteFsClient(
            'nfs', root_helper, execute=execute,
            nfs_mount_point_base=self.base,
            nfs_mount_options=opts)

        supports_auto_mosr = kwargs.get('supports_auto_mosr', False)
        self._sparse_copy_volume_data = True
        self.reserved_percentage = self.configuration.reserved_percentage
        self.max_over_subscription_ratio = (
            volume_utils.get_max_over_subscription_ratio(
                self.configuration.max_over_subscription_ratio,
                supports_auto=supports_auto_mosr))

        self._supports_encryption = True

    @staticmethod
    def get_driver_options():
        return nfs_opts + remotefs.nas_opts

    def initialize_connection(self, volume, connector):

        LOG.debug('Initializing connection to volume %(vol)s. '
                  'Connector: %(con)s', {'vol': volume.id, 'con': connector})

        active_vol = self.get_active_image_from_info(volume)
        volume_dir = self._local_volume_dir(volume)
        path_to_vol = os.path.join(volume_dir, active_vol)
        info = self._qemu_img_info(path_to_vol,
                                   volume['name'])

        data = {'export': volume.provider_location,
                'name': active_vol}
        if volume.provider_location in self.shares:
            data['options'] = self.shares[volume.provider_location]

        conn_info = {
            'driver_volume_type': self.driver_volume_type,
            'data': data,
            'mount_point_base': self._get_mount_point_base()
        }

        # Test file for raw vs. qcow2 format
        if info.file_format not in ['raw', 'qcow2']:
            msg = _('nfs volume must be a valid raw or qcow2 image.')
            raise exception.InvalidVolume(reason=msg)

        conn_info['data']['format'] = info.file_format
        LOG.debug('NfsDriver: conn_info: %s', conn_info)
        return conn_info

    def do_setup(self, context):
        """Any initialization the volume driver does while starting."""
        super(NfsDriver, self).do_setup(context)

        nas_host = getattr(self.configuration,
                           'nas_host',
                           None)
        nas_share_path = getattr(self.configuration,
                                 'nas_share_path',
                                 None)

        # If both nas_host and nas_share_path are set we are not
        # going to use the nfs_shares_config file.  So, only check
        # for its existence if it is going to be used.
        if((not nas_host) or (not nas_share_path)):
            config = self.configuration.nfs_shares_config
            if not config:
                msg = (_("There's no NFS config file configured (%s)") %
                       'nfs_shares_config')
                LOG.warning(msg)
                raise exception.NfsException(msg)
            if not os.path.exists(config):
                msg = (_("NFS config file at %(config)s doesn't exist") %
                       {'config': config})
                LOG.warning(msg)
                raise exception.NfsException(msg)

        self.shares = {}  # address : options

        # Check if mount.nfs is installed on this system; note that we
        # need to be root, to also find mount.nfs on distributions, where
        # it is not located in an unprivileged users PATH (e.g. /sbin).
        package = 'mount.nfs'
        try:
            self._execute(package, check_exit_code=False,
                          run_as_root=True)
        except OSError as exc:
            if exc.errno == errno.ENOENT:
                msg = _('%s is not installed') % package
                raise exception.NfsException(msg)
            else:
                raise

        # Now that all configuration data has been loaded (shares),
        # we can "set" our final NAS file security options.
        self.set_nas_security_options(self._is_voldb_empty_at_startup)
        self._check_snapshot_support(setup_checking=True)

    def _ensure_share_mounted(self, nfs_share):
        mnt_flags = []
        if self.shares.get(nfs_share) is not None:
            mnt_flags = self.shares[nfs_share].split()
        num_attempts = max(1, self.configuration.nfs_mount_attempts)
        for attempt in range(num_attempts):
            try:
                self._remotefsclient.mount(nfs_share, mnt_flags)
                return
            except Exception as e:
                if attempt == (num_attempts - 1):
                    LOG.error('Mount failure for %(share)s after '
                              '%(count)d attempts.',
                              {'share': nfs_share,
                               'count': num_attempts})
                    raise exception.NfsException(six.text_type(e))
                LOG.debug('Mount attempt %(attempt)d failed: %(exc)s.\n'
                          'Retrying mount ...',
                          {'attempt': attempt, 'exc': e})
                time.sleep(1)

    def _find_share(self, volume):
        """Choose NFS share among available ones for given volume size.

        For instances with more than one share that meets the criteria, the
        share with the least "allocated" space will be selected.

        :param volume: the volume to be created.
        """

        if not self._mounted_shares:
            raise exception.NfsNoSharesMounted()

        target_share = None
        target_share_reserved = 0

        for nfs_share in self._mounted_shares:
            total_size, total_available, total_allocated = (
                self._get_capacity_info(nfs_share))
            share_info = {'total_size': total_size,
                          'total_available': total_available,
                          'total_allocated': total_allocated,
                          }
            if not self._is_share_eligible(nfs_share,
                                           volume.size,
                                           share_info):
                continue
            if target_share is not None:
                if target_share_reserved > total_allocated:
                    target_share = nfs_share
                    target_share_reserved = total_allocated
            else:
                target_share = nfs_share
                target_share_reserved = total_allocated

        if target_share is None:
            raise exception.NfsNoSuitableShareFound(
                volume_size=volume.size)

        LOG.debug('Selected %s as target NFS share.', target_share)

        return target_share

    def _is_share_eligible(self, nfs_share, volume_size_in_gib,
                           share_info=None):
        """Verifies NFS share is eligible to host volume with given size.

        First validation step: ratio of actual space (used_space / total_space)
        is less than used_ratio. Second validation step: apparent space
        allocated (differs from actual space used when using sparse files)
        and compares the apparent available
        space (total_available * oversub_ratio) to ensure enough space is
        available for the new volume.

        :param nfs_share: NFS share
        :param volume_size_in_gib: int size in GB
        """
        # Because the generic NFS driver aggregates over all shares
        # when reporting capacity and usage stats to the scheduler,
        # we still have to perform some scheduler-like capacity
        # checks here, and these have to take into account
        # configuration for reserved space and oversubscription.
        # It would be better to do all this in the scheduler, but
        # this requires either pool support for the generic NFS
        # driver or limiting each NFS backend driver to a single share.

        # derive used_ratio from reserved percentage
        if share_info is None:
            total_size, total_available, total_allocated = (
                self._get_capacity_info(nfs_share))
            share_info = {'total_size': total_size,
                          'total_available': total_available,
                          'total_allocated': total_allocated,
                          }
        used_percentage = 100 - self.reserved_percentage
        used_ratio = used_percentage / 100.0

        requested_volume_size = volume_size_in_gib * units.Gi

        apparent_size = max(0, share_info['total_size'] *
                            self.max_over_subscription_ratio)

        apparent_available = max(0, apparent_size -
                                 share_info['total_allocated'])

        actual_used_ratio = ((share_info['total_size'] -
                              share_info['total_available']) /
                             float(share_info['total_size']))
        if actual_used_ratio > used_ratio:
            # NOTE(morganfainberg): We check the used_ratio first since
            # with oversubscription it is possible to not have the actual
            # available space but be within our oversubscription limit
            # therefore allowing this share to still be selected as a valid
            # target.
            LOG.debug('%s is not eligible - used ratio exceeded.',
                      nfs_share)
            return False
        if apparent_available <= requested_volume_size:
            LOG.debug('%s is not eligible - insufficient (apparent) available '
                      'space.',
                      nfs_share)
            return False
        if share_info['total_allocated'] / share_info['total_size'] >= (
                self.max_over_subscription_ratio):
            LOG.debug('%s is not eligible - utilization exceeds max '
                      'over subscription ratio.',
                      nfs_share)
            return False
        return True

    def _get_mount_point_for_share(self, nfs_share):
        """Needed by parent class."""
        return self._remotefsclient.get_mount_point(nfs_share)

    def _get_capacity_info(self, nfs_share):
        """Calculate available space on the NFS share.

        :param nfs_share: example 172.18.194.100:/var/nfs
        """
        mount_point = self._get_mount_point_for_share(nfs_share)

        df, _ = self._execute('stat', '-f', '-c', '%S %b %a', mount_point,
                              run_as_root=self._execute_as_root)
        block_size, blocks_total, blocks_avail = map(float, df.split())
        total_available = block_size * blocks_avail
        total_size = block_size * blocks_total

        du, _ = self._execute('du', '-sb', '--apparent-size', '--exclude',
                              '*snapshot*', mount_point,
                              run_as_root=self._execute_as_root)
        total_allocated = float(du.split()[0])
        return total_size, total_available, total_allocated

    def _get_mount_point_base(self):
        return self.base

    def extend_volume(self, volume, new_size):
        """Extend an existing volume to the new size."""
        if self._is_volume_attached(volume):
            # NOTE(kaisers): no attached extensions until #1870367 is fixed
            msg = (_("Cannot extend volume %s while it is attached.")
                   % volume['id'])
            raise exception.ExtendVolumeError(msg)

        LOG.info('Extending volume %s.', volume.id)
        extend_by = int(new_size) - volume.size
        if not self._is_share_eligible(volume.provider_location,
                                       extend_by):
            raise exception.ExtendVolumeError(reason='Insufficient space to'
                                              ' extend volume %s to %sG'
                                              % (volume.id, new_size))
        path = self.local_path(volume)
        LOG.info('Resizing file to %sG...', new_size)
        file_format = None
        admin_metadata = objects.Volume.get_by_id(
            context.get_admin_context(), volume.id).admin_metadata
        if admin_metadata and 'format' in admin_metadata:
            file_format = admin_metadata['format']
        image_utils.resize_image(path, new_size,
                                 run_as_root=self._execute_as_root,
                                 file_format=file_format)
        if file_format == 'qcow2' and not self._is_file_size_equal(
                path, new_size):
            raise exception.ExtendVolumeError(
                reason='Resizing image file failed.')

    def _is_file_size_equal(self, path, size):
        """Checks if file size at path is equal to size."""
        data = image_utils.qemu_img_info(path,
                                         run_as_root=self._execute_as_root)
        virt_size = int(data.virtual_size / units.Gi)
        return virt_size == size

    def set_nas_security_options(self, is_new_cinder_install):
        """Determine the setting to use for Secure NAS options.

        Value of each NAS Security option is checked and updated. If the
        option is currently 'auto', then it is set to either true or false
        based upon if this is a new Cinder installation. The RemoteFS variable
        '_execute_as_root' will be updated for this driver.

        :param is_new_cinder_install: bool indication of new Cinder install
        """
        doc_html = "https://docs.openstack.org/cinder/latest" \
                   "/admin/blockstorage-nfs-backend.html"

        self._ensure_shares_mounted()
        if not self._mounted_shares:
            raise exception.NfsNoSharesMounted()

        nfs_mount = self._get_mount_point_for_share(self._mounted_shares[0])

        self.configuration.nas_secure_file_permissions = \
            self._determine_nas_security_option_setting(
                self.configuration.nas_secure_file_permissions,
                nfs_mount, is_new_cinder_install)

        LOG.debug('NAS variable secure_file_permissions setting is: %s',
                  self.configuration.nas_secure_file_permissions)

        if self.configuration.nas_secure_file_permissions == 'false':
            LOG.warning("The NAS file permissions mode will be 666 "
                        "(allowing other/world read & write access). "
                        "This is considered an insecure NAS environment. "
                        "Please see %s for information on a secure "
                        "NFS configuration.",
                        doc_html)

        self.configuration.nas_secure_file_operations = \
            self._determine_nas_security_option_setting(
                self.configuration.nas_secure_file_operations,
                nfs_mount, is_new_cinder_install)

        # If secure NAS, update the '_execute_as_root' flag to not
        # run as the root user; run as process' user ID.

        # TODO(eharney): need to separate secure NAS vs. execute as root.
        # There are requirements to run some commands as root even
        # when running in secure NAS mode. (i.e. read volume file
        # attached to an instance and owned by qemu:qemu)
        if self.configuration.nas_secure_file_operations == 'true':
            self._execute_as_root = False

        LOG.debug('NAS secure file operations setting is: %s',
                  self.configuration.nas_secure_file_operations)

        if self.configuration.nas_secure_file_operations == 'false':
            LOG.warning("The NAS file operations will be run as "
                        "root: allowing root level access at the storage "
                        "backend. This is considered an insecure NAS "
                        "environment. Please see %s "
                        "for information on a secure NAS configuration.",
                        doc_html)

    def update_migrated_volume(self, ctxt, volume, new_volume,
                               original_volume_status):
        """Return the keys and values updated from NFS for migrated volume.

        This method should rename the back-end volume name(id) on the
        destination host back to its original name(id) on the source host.

        Due to this renaming, inheriting drivers may need to also re-associate
        other entities (such as backend QoS) with the new name.  Alternatively
        they could overwrite this method and raise NotImplemented.

        In any case, driver's code should always use the volume OVO's 'name_id'
        field instead of 'id' to get the volume's real UUID, as the renaming
        could fail.

        :param ctxt: The context used to run the method update_migrated_volume
        :param volume: The original volume that was migrated to this backend
        :param new_volume: The migration volume object that was created on
                           this backend as part of the migration process
        :param original_volume_status: The status of the original volume
        :returns: model_update to update DB with any needed changes
        """
        name_id = None
        if (original_volume_status == 'available' and
                volume.provider_location != new_volume.provider_location):
            current_name = CONF.volume_name_template % new_volume.id
            original_volume_name = CONF.volume_name_template % volume.id
            current_path = self.local_path(new_volume)
            # Replace the volume name with the original volume name
            original_path = current_path.replace(current_name,
                                                 original_volume_name)
            try:
                os.rename(current_path, original_path)
            except OSError:
                LOG.exception('Unable to rename the logical volume '
                              'for volume: %s', volume.id)
                # If the rename fails, _name_id should be set to the new
                # volume id and provider_location should be set to the
                # one from the new volume as well.
                name_id = new_volume._name_id or new_volume.id
        else:
            # The back-end will not be renamed.
            name_id = new_volume._name_id or new_volume.id
        return {'_name_id': name_id,
                'provider_location': new_volume.provider_location}

    def _update_volume_stats(self):
        """Retrieve stats info from volume group."""

        super(NfsDriver, self)._update_volume_stats()
        self._stats['sparse_copy_volume'] = True
        data = self._stats

        global_capacity = data['total_capacity_gb']
        global_free = data['free_capacity_gb']

        thin_enabled = self.configuration.nfs_sparsed_volumes
        if thin_enabled:
            provisioned_capacity = self._get_provisioned_capacity()
        else:
            provisioned_capacity = round(global_capacity - global_free, 2)

        data['provisioned_capacity_gb'] = provisioned_capacity
        data['max_over_subscription_ratio'] = self.max_over_subscription_ratio
        data['reserved_percentage'] = self.reserved_percentage
        data['thin_provisioning_support'] = thin_enabled
        data['thick_provisioning_support'] = not thin_enabled

        self._stats = data

    @coordination.synchronized('{self.driver_prefix}-{volume.id}')
    def create_volume(self, volume):
        """Apply locking to the create volume operation."""

        return super(NfsDriver, self).create_volume(volume)

    @coordination.synchronized('{self.driver_prefix}-{volume.id}')
    def delete_volume(self, volume):
        """Deletes a logical volume."""

        LOG.debug('Deleting volume %(vol)s, provider_location: %(loc)s',
                  {'vol': volume.id, 'loc': volume.provider_location})

        if not volume.provider_location:
            LOG.warning('Volume %s does not have provider_location '
                        'specified, skipping', volume.name)
            return

        info_path = self._local_path_volume_info(volume)
        info = self._read_info_file(info_path, empty_if_missing=True)

        if info:
            base_volume_path = os.path.join(self._local_volume_dir(volume),
                                            info['active'])
            self._delete(info_path)
        else:
            base_volume_path = self._local_path_volume(volume)

        self._delete(base_volume_path)

    def _qemu_img_info(self, path, volume_name):
        return super(NfsDriver, self)._qemu_img_info_base(
            path,
            volume_name,
            self.configuration.nfs_mount_point_base,
            force_share=True,
            run_as_root=True)

    def _check_snapshot_support(self, setup_checking=False):
        """Ensure snapshot support is enabled in config."""

        if (not self.configuration.nfs_snapshot_support and
                not setup_checking):
            msg = _("NFS driver snapshot support is disabled in cinder.conf.")
            raise exception.VolumeDriverException(message=msg)

        if (self.configuration.nas_secure_file_operations == 'true' and
                self.configuration.nfs_snapshot_support):
            msg = _("Snapshots are not supported with "
                    "nas_secure_file_operations enabled ('true' or 'auto'). "
                    "Please set it to 'false' if you intend to  have "
                    "it enabled.")
            LOG.error(msg)
            raise exception.VolumeDriverException(message=msg)

    @coordination.synchronized('{self.driver_prefix}-{snapshot.volume.id}')
    def create_snapshot(self, snapshot):
        """Apply locking to the create snapshot operation."""

        self._check_snapshot_support()
        return self._create_snapshot(snapshot)

    @coordination.synchronized('{self.driver_prefix}-{snapshot.volume.id}')
    def delete_snapshot(self, snapshot):
        """Apply locking to the delete snapshot operation."""

        return self._delete_snapshot(snapshot)

    def _copy_volume_from_snapshot(self, snapshot, volume, volume_size,
                                   src_encryption_key_id=None,
                                   new_encryption_key_id=None):
        """Copy data from snapshot to destination volume.

        This is done with a qemu-img convert to raw/qcow2 from the snapshot
        qcow2.
        """

        LOG.debug("Copying snapshot: %(snap)s -> volume: %(vol)s, "
                  "volume_size: %(size)s GB",
                  {'snap': snapshot.id,
                   'vol': volume.id,
                   'size': volume_size})

        info_path = self._local_path_volume_info(snapshot.volume)
        snap_info = self._read_info_file(info_path)
        vol_path = self._local_volume_dir(snapshot.volume)
        forward_file = snap_info[snapshot.id]
        forward_path = os.path.join(vol_path, forward_file)

        # Find the file which backs this file, which represents the point
        # when this snapshot was created.
        img_info = self._qemu_img_info(forward_path, snapshot.volume.name)
        path_to_snap_img = os.path.join(vol_path, img_info.backing_file)

        path_to_new_vol = self._local_path_volume(volume)

        LOG.debug("will copy from snapshot at %s", path_to_snap_img)

        if self.configuration.nfs_qcow2_volumes:
            out_format = 'qcow2'
        else:
            out_format = 'raw'

        if new_encryption_key_id is not None:
            if src_encryption_key_id is None:
                message = _("Can't create an encrypted volume %(format)s "
                            "from an unencrypted source."
                            ) % {'format': out_format}
                LOG.error(message)
                # TODO(enriquetaso): handle unencrypted snap->encrypted vol
                raise exception.NfsException(message)
            keymgr = key_manager.API(CONF)
            new_key = keymgr.get(volume.obj_context, new_encryption_key_id)
            new_passphrase = \
                binascii.hexlify(new_key.get_encoded()).decode('utf-8')

            # volume.obj_context is the owner of this request
            src_key = keymgr.get(volume.obj_context, src_encryption_key_id)
            src_passphrase = \
                binascii.hexlify(src_key.get_encoded()).decode('utf-8')

            tmp_dir = volume_utils.image_conversion_dir()
            with tempfile.NamedTemporaryFile(prefix='luks_',
                                             dir=tmp_dir) as src_pass_file:
                with open(src_pass_file.name, 'w') as f:
                    f.write(src_passphrase)

                with tempfile.NamedTemporaryFile(prefix='luks_',
                                                 dir=tmp_dir) as new_pass_file:
                    with open(new_pass_file.name, 'w') as f:
                        f.write(new_passphrase)

                    image_utils.convert_image(
                        path_to_snap_img,
                        path_to_new_vol,
                        'luks',
                        passphrase_file=new_pass_file.name,
                        src_passphrase_file=src_pass_file.name,
                        run_as_root=self._execute_as_root)
        else:
            image_utils.convert_image(path_to_snap_img,
                                      path_to_new_vol,
                                      out_format,
                                      run_as_root=self._execute_as_root)

        self._set_rw_permissions_for_all(path_to_new_vol)