path: root/ironic/drivers/modules/drac/raid.py

#
# 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.

"""
DRAC RAID specific methods
"""

from collections import defaultdict
import math

from futurist import periodics
from ironic_lib import metrics_utils
from oslo_log import log as logging
from oslo_utils import importutils
from oslo_utils import units

from ironic.common import exception
from ironic.common.i18n import _
from ironic.common import raid as raid_common
from ironic.conductor import task_manager
from ironic.conductor import utils as manager_utils
from ironic.conf import CONF
from ironic.drivers import base
from ironic.drivers.modules import deploy_utils
from ironic.drivers.modules.drac import common as drac_common
from ironic.drivers.modules.drac import job as drac_job

drac_exceptions = importutils.try_import('dracclient.exceptions')
drac_constants = importutils.try_import('dracclient.constants')

LOG = logging.getLogger(__name__)

METRICS = metrics_utils.get_metrics_logger(__name__)

RAID_LEVELS = {
    '0': {
        'min_disks': 1,
        'max_disks': 1000,
        'type': 'simple',
        'overhead': 0
    },
    '1': {
        'min_disks': 2,
        'max_disks': 2,
        'type': 'simple',
        'overhead': 1
    },
    '5': {
        'min_disks': 3,
        'max_disks': 1000,
        'type': 'simple',
        'overhead': 1
    },
    '6': {
        'min_disks': 4,
        'max_disks': 1000,
        'type': 'simple',
        'overhead': 2
    },
    '1+0': {
        'type': 'spanned',
        'span_type': '1'
    },
    '5+0': {
        'type': 'spanned',
        'span_type': '5'
    },
    '6+0': {
        'type': 'spanned',
        'span_type': '6'
    }
}


def list_raid_controllers(node):
    """List the RAID controllers of the node.

    :param node: an ironic node object.
    :returns: a list of RAIDController objects from dracclient.
    :raises: DracOperationError on an error from python-dracclient.
    """
    client = drac_common.get_drac_client(node)

    try:
        return client.list_raid_controllers()
    except drac_exceptions.BaseClientException as exc:
        LOG.error('DRAC driver failed to get the list of RAID controllers '
                  'for node %(node_uuid)s. Reason: %(error)s.',
                  {'node_uuid': node.uuid, 'error': exc})
        raise exception.DracOperationError(error=exc)


def list_virtual_disks(node):
    """List the virtual disks of the node.

    :param node: an ironic node object.
    :returns: a list of VirtualDisk objects from dracclient.
    :raises: DracOperationError on an error from python-dracclient.
    """
    client = drac_common.get_drac_client(node)

    try:
        return client.list_virtual_disks()
    except drac_exceptions.BaseClientException as exc:
        LOG.error('DRAC driver failed to get the list of virtual disks '
                  'for node %(node_uuid)s. Reason: %(error)s.',
                  {'node_uuid': node.uuid, 'error': exc})
        raise exception.DracOperationError(error=exc)


def list_physical_disks(node):
    """List the physical disks of the node.

    :param node: an ironic node object.
    :returns: a list of PhysicalDisk objects from dracclient.
    :raises: DracOperationError on an error from python-dracclient.
    """
    client = drac_common.get_drac_client(node)

    try:
        return client.list_physical_disks()
    except drac_exceptions.BaseClientException as exc:
        LOG.error('DRAC driver failed to get the list of physical disks '
                  'for node %(node_uuid)s. Reason: %(error)s.',
                  {'node_uuid': node.uuid, 'error': exc})
        raise exception.DracOperationError(error=exc)


def _is_raid_controller(node, raid_controller_fqdd, raid_controllers=None):
    """Find out if object's fqdd is for a raid controller or not

    :param node: an ironic node object
    :param raid_controller_fqdd: The object's fqdd we are testing to see
                                 if it is a raid controller or not.
    :param raid_controllers: A list of RAIDControllers used to check for
                             the presence of BOSS cards.  If None, the
                             iDRAC will be queried for the list of
                             controllers.
    :returns: boolean, True if the device is a RAID controller,
              False if not.
    """
    client = drac_common.get_drac_client(node)

    try:
        return client.is_raid_controller(raid_controller_fqdd,
                                         raid_controllers)
    except drac_exceptions.BaseClientException as exc:
        LOG.error('Unable to determine if controller %(raid_controller_fqdd)s '
                  'on node %(node_uuid)s is a RAID controller. '
                  'Reason: %(error)s. ',
                  {'raid_controller_fqdd': raid_controller_fqdd,
                   'node_uuid': node.uuid, 'error': exc})

        raise exception.DracOperationError(error=exc)


def _validate_job_queue(node, raid_controller=None):
    """Validate that there are no pending jobs for this controller.

    :param node: an ironic node object.
    :param raid_controller: id of the RAID controller.
    """
    kwargs = {}
    if raid_controller:
        kwargs["name_prefix"] = "Config:RAID:%s" % raid_controller
    drac_job.validate_job_queue(node, **kwargs)


def create_virtual_disk(node, raid_controller, physical_disks, raid_level,
                        size_mb, disk_name=None, span_length=None,
                        span_depth=None):
    """Create a single virtual disk on a RAID controller.

    The created virtual disk will be in pending state. The DRAC card will do
    the actual configuration once the changes are applied by calling the
    ``commit_config`` method.

    :param node: an ironic node object.
    :param raid_controller: id of the RAID controller.
    :param physical_disks: ids of the physical disks.
    :param raid_level: RAID level of the virtual disk.
    :param size_mb: size of the virtual disk.
    :param disk_name: name of the virtual disk. (optional)
    :param span_depth: Number of spans in virtual disk. (optional)
    :param span_length: Number of disks per span. (optional)
    :returns: a dictionary containing the commit_needed key with a boolean
              value indicating whether a config job must be created for the
              values to be applied.
    :raises: DracOperationError on an error from python-dracclient.
    """
    # This causes config to fail, because the boot mode is set via a config
    # job.
    _validate_job_queue(node, raid_controller)

    client = drac_common.get_drac_client(node)

    try:
        return client.create_virtual_disk(raid_controller, physical_disks,
                                          raid_level, size_mb, disk_name,
                                          span_length, span_depth)
    except drac_exceptions.BaseClientException as exc:
        LOG.error('DRAC driver failed to create virtual disk for node '
                  '%(node_uuid)s. Reason: %(error)s.',
                  {'node_uuid': node.uuid,
                   'error': exc})
        raise exception.DracOperationError(error=exc)


def delete_virtual_disk(node, virtual_disk):
    """Delete a single virtual disk on a RAID controller.

    The deleted virtual disk will be in pending state. The DRAC card will do
    the actual configuration once the changes are applied by calling the
    ``commit_config`` method.

    :param node: an ironic node object.
    :param virtual_disk: id of the virtual disk.
    :returns: a dictionary containing the commit_needed key with a boolean
              value indicating whether a config job must be created for the
              values to be applied.
    :raises: DracOperationError on an error from python-dracclient.
    """
    # NOTE(mgoddard): Cannot specify raid_controller as we don't know it.
    _validate_job_queue(node)

    client = drac_common.get_drac_client(node)

    try:
        return client.delete_virtual_disk(virtual_disk)
    except drac_exceptions.BaseClientException as exc:
        LOG.error('DRAC driver failed to delete virtual disk '
                  '%(virtual_disk_fqdd)s for node %(node_uuid)s. '
                  'Reason: %(error)s.',
                  {'virtual_disk_fqdd': virtual_disk,
                   'node_uuid': node.uuid,
                   'error': exc})
        raise exception.DracOperationError(error=exc)


def _reset_raid_config(node, raid_controller):
    """Delete all virtual disk and unassign all hotspares physical disk

    :param node: an ironic node object.
    :param raid_controller: id of the RAID controller.
    :returns: a dictionary containing
              - The is_commit_required needed key with a
              boolean value indicating whether a config job must be created
              for the values to be applied.
              - The is_reboot_required key with a RebootRequired enumerated
              value indicating whether the server must be rebooted to
              reset configuration.
    :raises: DracOperationError on an error from python-dracclient.
    """
    try:

        _validate_job_queue(node, raid_controller)

        client = drac_common.get_drac_client(node)
        return client.reset_raid_config(raid_controller)
    except drac_exceptions.BaseClientException as exc:
        LOG.error('DRAC driver failed to delete all virtual disk '
                  'and unassign all hotspares '
                  'on %(raid_controller_fqdd)s '
                  'for node %(node_uuid)s. '
                  'Reason: %(error)s.',
                  {'raid_controller_fqdd': raid_controller,
                   'node_uuid': node.uuid,
                   'error': exc})
        raise exception.DracOperationError(error=exc)


def clear_foreign_config(node, raid_controller):
    """Free up the foreign drives.

    :param node: an ironic node object.
    :param raid_controller: id of the RAID controller.
    :returns: a dictionary containing
              - The is_commit_required needed key with a
              boolean value indicating whether a config job must be created
              for the values to be applied.
              - The is_reboot_required key with a RebootRequired enumerated
              value indicating whether the server must be rebooted to
              clear foreign configuration.
    :raises: DracOperationError on an error from python-dracclient.
    """
    try:

        _validate_job_queue(node, raid_controller)

        client = drac_common.get_drac_client(node)
        return client.clear_foreign_config(raid_controller)
    except drac_exceptions.BaseClientException as exc:
        LOG.error('DRAC driver failed to free foreign driver '
                  'on %(raid_controller_fqdd)s '
                  'for node %(node_uuid)s. '
                  'Reason: %(error)s.',
                  {'raid_controller_fqdd': raid_controller,
                   'node_uuid': node.uuid,
                   'error': exc})
        raise exception.DracOperationError(error=exc)


def change_physical_disk_state(node, mode=None,
                               controllers_to_physical_disk_ids=None):
    """Convert disks RAID status

    This method converts the requested physical disks from
    RAID to JBOD or vice versa.  It does this by only converting the
    disks that are not already in the correct state.

    :param node: an ironic node object.
    :param mode: the mode to change the disks either to RAID or JBOD.
    :param controllers_to_physical_disk_ids: Dictionary of controllers and
           corresponding disk ids to convert to the requested mode.
    :return: a dictionary containing:
             - conversion_results, a dictionary that maps controller ids
             to the conversion results for that controller.
             The conversion results are a dict that contains:
             - The is_commit_required key with the value always set to
             True indicating that a config job must be created to
             complete disk conversion.
             - The is_reboot_required key with a RebootRequired
             enumerated value indicating whether the server must be
             rebooted to complete disk conversion.
    :raises: DRACOperationError on an error from python-dracclient.
    """
    try:
        drac_job.validate_job_queue(node)
        client = drac_common.get_drac_client(node)
        return client.change_physical_disk_state(
            mode, controllers_to_physical_disk_ids)
    except drac_exceptions.BaseClientException as exc:
        LOG.error('DRAC driver failed to change physical drives '
                  'to %(mode)s mode for node %(node_uuid)s. '
                  'Reason: %(error)s.',
                  {'mode': mode, 'node_uuid': node.uuid, 'error': exc})
        raise exception.DracOperationError(error=exc)


def commit_config(node, raid_controller, reboot=False, realtime=False):
    """Apply all pending changes on a RAID controller.

    :param node: an ironic node object.
    :param raid_controller: id of the RAID controller.
    :param reboot: indicates whether a reboot job should be automatically
                   created with the config job. (optional, defaults to False)
    :param realtime: indicates RAID controller supports realtime.
                     (optional, defaults to False)
    :returns: id of the created job
    :raises: DracOperationError on an error from python-dracclient.
    """
    client = drac_common.get_drac_client(node)

    try:
        return client.commit_pending_raid_changes(
            raid_controller=raid_controller,
            reboot=reboot,
            realtime=realtime)
    except drac_exceptions.BaseClientException as exc:
        LOG.error('DRAC driver failed to commit pending RAID config for'
                  ' controller %(raid_controller_fqdd)s on node '
                  '%(node_uuid)s. Reason: %(error)s.',
                  {'raid_controller_fqdd': raid_controller,
                   'node_uuid': node.uuid,
                   'error': exc})
        raise exception.DracOperationError(error=exc)


def _change_physical_disk_mode(node, mode=None,
                               controllers_to_physical_disk_ids=None,
                               substep="completed"):
    """Physical drives conversion from RAID to JBOD or vice-versa.

    :param node: an ironic node object.
    :param mode: the mode to change the disks either to RAID or JBOD.
    :param controllers_to_physical_disk_ids: Dictionary of controllers and
           corresponding disk ids to convert to the requested mode.
    :returns: states.CLEANWAIT if deletion is in progress asynchronously
              or None if it is completed.
    """
    change_disk_state = change_physical_disk_state(
        node, mode, controllers_to_physical_disk_ids)

    controllers = list()
    if 'conversion_results' in change_disk_state:
        conversion_results = change_disk_state['conversion_results']
        for controller_id, result in conversion_results.items():
            controller = {'raid_controller': controller_id,
                          'is_reboot_required': result['is_reboot_required'],
                          'is_commit_required': result['is_commit_required']}
            controllers.append(controller)
    else:
        controller_ids = change_disk_state['commit_required_ids']
        for controller_id in controller_ids:
            controller = {'raid_controller': controller_id,
                          'is_reboot_required':
                          change_disk_state['is_reboot_required'],
                          'is_commit_required': True}
            controllers.append(controller)

    return _commit_to_controllers(
        node,
        controllers, substep=substep)


def abandon_config(node, raid_controller):
    """Deletes all pending changes on a RAID controller.

    :param node: an ironic node object.
    :param raid_controller: id of the RAID controller.
    :raises: DracOperationError on an error from python-dracclient.
    """
    client = drac_common.get_drac_client(node)

    try:
        client.abandon_pending_raid_changes(raid_controller)
    except drac_exceptions.BaseClientException as exc:
        LOG.error('DRAC driver failed to delete pending RAID config '
                  'for controller %(raid_controller_fqdd)s on node '
                  '%(node_uuid)s. Reason: %(error)s.',
                  {'raid_controller_fqdd': raid_controller,
                   'node_uuid': node.uuid,
                   'error': exc})
        raise exception.DracOperationError(error=exc)


def _calculate_spans(raid_level, disks_count):
    """Calculates number of spans for a RAID level given a physical disk count

    :param raid_level: RAID level of the virtual disk.
    :param disk_count: number of physical disks used for the virtual disk.
    :returns: number of spans.
    """
    if raid_level in ['0', '1', '5', '6']:
        return 1
    elif raid_level in ['5+0', '6+0']:
        return 2
    elif raid_level in ['1+0']:
        return disks_count >> 1
    else:
        reason = (_('Cannot calculate spans for RAID level "%s"') %
                  raid_level)
        raise exception.DracOperationError(error=reason)


def _usable_disks_count(raid_level, disks_count):
    """Calculates the number of disks usable for a RAID level

    ...given a physical disk count

    :param raid_level: RAID level of the virtual disk.
    :param disk_count: number of physical disks used for the virtual disk.
    :returns: number of disks.
    """
    if raid_level in ['0', '1', '5', '6']:
        return disks_count
    elif raid_level in ['5+0', '6+0', '1+0']:
        # largest even number less than disk_count
        return (disks_count >> 1) << 1
    else:
        reason = (_('RAID level %(raid_level)s is not supported by the '
                    'driver. Supported RAID levels: %(supported_raid_levels)s')
                  % {'raid_level': raid_level,
                     'supported_raid_levels': list(RAID_LEVELS)})
        raise exception.DracOperationError(error=reason)


def _raid_level_min_disks(raid_level, spans_count=1):
    try:
        raid_level_info = RAID_LEVELS[raid_level]
    except KeyError:
        reason = (_('RAID level %(raid_level)s is not supported by the '
                    'driver. Supported RAID levels: %(supported_raid_levels)s')
                  % {'raid_level': raid_level,
                     'supported_raid_levels': list(RAID_LEVELS)})
        raise exception.DracOperationError(error=reason)

    if raid_level_info['type'] == 'spanned':
        if spans_count <= 1:
            reason = _('Spanned RAID volumes cannot contain a single span')
            raise exception.DracOperationError(error=reason)

        span_type = raid_level_info['span_type']
        raid_level_info = RAID_LEVELS[span_type]

    return raid_level_info['min_disks'] * spans_count


def _raid_level_max_disks(raid_level, spans_count=1):
    try:
        raid_level_info = RAID_LEVELS[raid_level]
    except KeyError:
        reason = (_('RAID level %(raid_level)s is not supported by the '
                    'driver. Supported RAID levels: %(supported_raid_levels)s')
                  % {'raid_level': raid_level,
                     'supported_raid_levels': list(RAID_LEVELS)})
        raise exception.DracOperationError(error=reason)

    if raid_level_info['type'] == 'spanned':
        if spans_count <= 1:
            reason = _('Spanned RAID volumes cannot contain a single span')
            raise exception.DracOperationError(error=reason)

        span_type = raid_level_info['span_type']
        raid_level_info = RAID_LEVELS[span_type]

    return raid_level_info['max_disks'] * spans_count


def _raid_level_overhead(raid_level, spans_count=1):
    try:
        raid_level_info = RAID_LEVELS[raid_level]
    except KeyError:
        reason = (_('RAID level %(raid_level)s is not supported by the '
                    'driver. Supported RAID levels: %(supported_raid_levels)s')
                  % {'raid_level': raid_level,
                     'supported_raid_levels': list(RAID_LEVELS)})
        raise exception.DracOperationError(error=reason)

    if raid_level_info['type'] == 'spanned':
        if spans_count <= 1:
            reason = _('Spanned RAID volumes cannot contain a single span')
            raise exception.DracOperationError(error=reason)

        span_type = raid_level_info['span_type']
        raid_level_info = RAID_LEVELS[span_type]

    return raid_level_info['overhead'] * spans_count


def _max_volume_size_mb(raid_level, physical_disks, free_space_mb,
                        spans_count=1, stripe_size_kb=64 * units.Ki):
    # restrict the size to the smallest available space
    free_spaces = [free_space_mb[disk] for disk in physical_disks]
    size_kb = min(free_spaces) * units.Ki

    # NOTE(ifarkas): using math.floor so we get a volume size that does not
    #                exceed the available space
    stripes_per_disk = int(math.floor(float(size_kb) / stripe_size_kb))

    disks_count = len(physical_disks)
    overhead_disks_count = _raid_level_overhead(raid_level, spans_count)

    return int(stripes_per_disk * stripe_size_kb
               * (disks_count - overhead_disks_count) / units.Ki)


def _volume_usage_per_disk_mb(logical_disk, physical_disks, spans_count=1,
                              stripe_size_kb=64 * units.Ki):
    disks_count = len(physical_disks)
    overhead_disks_count = _raid_level_overhead(logical_disk['raid_level'],
                                                spans_count)
    volume_size_kb = logical_disk['size_mb'] * units.Ki
    # NOTE(ifarkas): using math.ceil so we get the largest disk usage
    #                possible, so we can avoid over-committing
    stripes_per_volume = math.ceil(float(volume_size_kb) / stripe_size_kb)

    stripes_per_disk = math.ceil(
        float(stripes_per_volume) / (disks_count - overhead_disks_count))
    return int(stripes_per_disk * stripe_size_kb / units.Ki)


def _find_configuration(logical_disks, physical_disks, pending_delete):
    """Find RAID configuration.

    This method transforms the RAID configuration defined in Ironic to a format
    that is required by dracclient. This includes matching the physical disks
    to RAID volumes when it's not pre-defined, or in general calculating
    missing properties.

    :param logical_disks: list of logical disk definitions.
    :param physical_disks: list of physical disk definitions.
    :param pending_delete: Whether there is a pending deletion of virtual
        disks that should be accounted for.
    """

    # shared physical disks of RAID volumes size_gb='MAX' should be
    # deprioritized during the matching process to reserve as much space as
    # possible. Reserved means it won't be used during matching.
    volumes_with_reserved_physical_disks = [
        volume for volume in logical_disks
        if ('physical_disks' in volume and volume['size_mb'] == 'MAX'
            and volume.get('share_physical_disks', False))]
    reserved_physical_disks = [
        disk for disk in physical_disks
        for volume in volumes_with_reserved_physical_disks
        if disk.id in volume['physical_disks']]

    # we require each logical disk contain only homogeneous physical disks, so
    # sort them by type
    physical_disks_by_type = {}
    reserved_physical_disks_by_type = {}
    free_space_mb = {}
    for disk in physical_disks:
        # calculate free disk space
        free_space_mb[disk] = _get_disk_free_size_mb(disk, pending_delete)

        disk_type = (disk.controller, disk.media_type, disk.interface_type,
                     disk.size_mb)
        if disk_type not in physical_disks_by_type:
            physical_disks_by_type[disk_type] = []
            reserved_physical_disks_by_type[disk_type] = []

        if disk in reserved_physical_disks:
            reserved_physical_disks_by_type[disk_type].append(disk)
        else:
            physical_disks_by_type[disk_type].append(disk)

    # exclude non-shared physical disks (predefined by the user) from
    # physical_disks_by_type because they are not going to be used during
    # matching
    for volume in logical_disks:
        if ('physical_disks' in volume
                and not volume.get('share_physical_disks', False)):
            for disk in physical_disks:
                if disk.id in volume['physical_disks']:
                    disk_type = (disk.controller, disk.media_type,
                                 disk.interface_type, disk.size_mb)
                    if disk in physical_disks_by_type[disk_type]:
                        physical_disks_by_type[disk_type].remove(disk)

    processed_volumes = []

    # step 1 - process volumes with predefined disks and exact size
    for volume in [volume for volume in logical_disks
                   if ('physical_disks' in volume
                       and volume['size_mb'] != 'MAX')]:
        _calculate_volume_props(volume, physical_disks, free_space_mb)
        processed_volumes.append(volume)

    # step 2 - process volumes without predefined disks
    volumes_without_disks = [disk for disk in logical_disks
                             if 'physical_disks' not in disk]

    if volumes_without_disks:
        result, free_space_mb = (
            _assign_disks_to_volume(volumes_without_disks,
                                    physical_disks_by_type, free_space_mb,
                                    pending_delete))
        if not result:
            # try again using the reserved physical disks in addition
            for disk_type, disks in physical_disks_by_type.items():
                physical_disks_by_type[disk_type] += (
                    reserved_physical_disks_by_type[disk_type])

            result, free_space_mb = (
                _assign_disks_to_volume(volumes_without_disks,
                                        physical_disks_by_type,
                                        free_space_mb,
                                        pending_delete))
            if not result:
                error_msg = _('failed to find matching physical disks for all '
                              'logical disks')
                LOG.error('DRAC driver failed to create RAID '
                          'configuration. Reason: %(error)s.',
                          {'error': error_msg})
                raise exception.DracOperationError(error=error_msg)

    processed_volumes += volumes_without_disks

    # step 3 - process volumes with predefined disks and size_mb == 'MAX'
    for volume in [volume for volume in logical_disks
                   if ('physical_disks' in volume
                       and volume['size_mb'] == 'MAX')]:
        _calculate_volume_props(volume, physical_disks, free_space_mb)
        processed_volumes.append(volume)

    return processed_volumes


def _calculate_volume_props(logical_disk, physical_disks, free_space_mb):
    selected_disks = [disk for disk in physical_disks
                      if disk.id in logical_disk['physical_disks']]

    spans_count = _calculate_spans(
        logical_disk['raid_level'], len(selected_disks))

    if len(selected_disks) % spans_count != 0:
        error_msg = _('invalid number of physical disks was provided')
        raise exception.DracOperationError(error=error_msg)

    disks_per_span = int(len(selected_disks) / spans_count)

    # Best practice is to not pass span_length and span_depth when creating a
    # RAID10.  The iDRAC will dynamically calculate these values using maximum
    # values obtained from the RAID controller.
    logical_disk['span_depth'] = None
    logical_disk['span_length'] = None
    if logical_disk['raid_level'] != '1+0':
        logical_disk['span_depth'] = spans_count
        logical_disk['span_length'] = disks_per_span

    max_volume_size_mb = _max_volume_size_mb(
        logical_disk['raid_level'], selected_disks, free_space_mb,
        spans_count=spans_count)

    if logical_disk['size_mb'] == 'MAX':
        if max_volume_size_mb == 0:
            error_msg = _("size set to 'MAX' but could not allocate physical "
                          "disk space")
            raise exception.DracOperationError(error=error_msg)

        logical_disk['size_mb'] = max_volume_size_mb
    elif max_volume_size_mb < logical_disk['size_mb']:
        if max_volume_size_mb == 0:
            error_msg = _('not enough physical disk space for the logical '
                          'disk')
            raise exception.DracOperationError(error=error_msg)

    disk_usage = _volume_usage_per_disk_mb(logical_disk, selected_disks,
                                           spans_count=spans_count)

    for disk in selected_disks:
        if free_space_mb[disk] < disk_usage:
            error_msg = _('not enough free space on physical disks for the '
                          'logical disk')
            raise exception.DracOperationError(error=error_msg)
        else:
            free_space_mb[disk] -= disk_usage

    if 'controller' not in logical_disk:
        logical_disk['controller'] = selected_disks[0].controller


def _assign_disks_to_volume(logical_disks, physical_disks_by_type,
                            free_space_mb, pending_delete):
    logical_disk = logical_disks.pop(0)
    raid_level = logical_disk['raid_level']

    # iterate over all possible configurations
    for (controller, disk_type,
         interface_type, size_mb), disks in physical_disks_by_type.items():

        if ('disk_type' in logical_disk
            and logical_disk['disk_type'] != disk_type):
            continue

        if ('interface_type' in logical_disk
            and logical_disk['interface_type'] != interface_type):
            continue

        # filter out disks without free disk space
        disks = [disk for disk in disks if free_space_mb[disk] > 0]

        # sort disks by free size which is important if we have max disks limit
        # on a volume
        disks = sorted(
            disks,
            key=lambda disk: free_space_mb[disk])

        # filter out disks already in use if sharing is disabled
        if ('share_physical_disks' not in logical_disk
                or not logical_disk['share_physical_disks']):
            initial_free_size_mb = {
                disk: _get_disk_free_size_mb(disk, pending_delete)
                for disk in disks
            }
            disks = [disk for disk in disks
                     if initial_free_size_mb[disk] == free_space_mb[disk]]

        max_spans = _calculate_spans(raid_level, len(disks))
        min_spans = min([2, max_spans])
        min_disks = _raid_level_min_disks(raid_level,
                                          spans_count=min_spans)
        max_disks = _raid_level_max_disks(raid_level,
                                          spans_count=max_spans)
        candidate_max_disks = min([max_disks, len(disks)])

        for disks_count in range(min_disks, candidate_max_disks + 1):
            if ('number_of_physical_disks' in logical_disk
                and logical_disk['number_of_physical_disks'] != disks_count):
                    continue

            # skip invalid disks_count
            if disks_count != _usable_disks_count(logical_disk['raid_level'],
                                                  disks_count):
                continue

            selected_disks = disks[0:disks_count]

            candidate_volume = logical_disk.copy()
            candidate_free_space_mb = free_space_mb.copy()
            candidate_volume['physical_disks'] = [disk.id for disk
                                                  in selected_disks]
            try:
                _calculate_volume_props(candidate_volume, selected_disks,
                                        candidate_free_space_mb)
            except exception.DracOperationError:
                continue

            if len(logical_disks) > 0:
                result, candidate_free_space_mb = (
                    _assign_disks_to_volume(logical_disks,
                                            physical_disks_by_type,
                                            candidate_free_space_mb,
                                            pending_delete))
                if result:
                    logical_disks.append(candidate_volume)
                    return (True, candidate_free_space_mb)
            else:
                logical_disks.append(candidate_volume)
                return (True, candidate_free_space_mb)
    else:
        # put back the logical_disk to queue
        logical_disks.insert(0, logical_disk)
        return (False, free_space_mb)


def _filter_logical_disks(logical_disks, include_root_volume,
                          include_nonroot_volumes):
    filtered_disks = []
    for disk in logical_disks:
        if include_root_volume and disk.get('is_root_volume'):
            filtered_disks.append(disk)

        if include_nonroot_volumes and not disk.get('is_root_volume'):
            filtered_disks.append(disk)

    return filtered_disks


def _create_config_job(node, controller, reboot=False, realtime=False,
                       raid_config_job_ids=[],
                       raid_config_parameters=[]):
    job_id = commit_config(node, raid_controller=controller,
                           reboot=reboot, realtime=realtime)

    raid_config_job_ids.append(job_id)
    if controller not in raid_config_parameters:
        raid_config_parameters.append(controller)

    LOG.info('Change has been committed to RAID controller '
             '%(controller)s on node %(node)s. '
             'DRAC job id: %(job_id)s',
             {'controller': controller, 'node': node.uuid,
              'job_id': job_id})
    return {'raid_config_job_ids': raid_config_job_ids,
            'raid_config_parameters': raid_config_parameters}


def _commit_to_controllers(node, controllers, substep="completed"):
    """Commit changes to RAID controllers on the node.

    :param node: an ironic node object
    :param controllers: a list of dictionary containing
                        - The raid_controller key with raid controller
                        fqdd value indicating on which raid configuration
                        job needs to be perform.
                        - The is_commit_required needed key with a
                        boolean value indicating whether a config job must
                        be created.
                        - The is_reboot_required key with a RebootRequired
                        enumerated value indicating whether the server must
                        be rebooted only if raid controller does not support
                        realtime.
    :param substep: contain sub cleaning or deploy step which executes any raid
                    configuration job if set after cleaning or deploy step.
                    (default to completed)
    :returns: states.CLEANWAIT (cleaning) or states.DEPLOYWAIT (deployment) if
              configuration is in progress asynchronously or None if it is
              completed.
    """
    # remove controller which does not require configuration job
    controllers = [controller for controller in controllers
                   if controller['is_commit_required']]

    if not controllers:
        LOG.debug('No changes on any of the controllers on node %s',
                  node.uuid)
        driver_internal_info = node.driver_internal_info
        driver_internal_info['raid_config_substep'] = substep
        driver_internal_info['raid_config_parameters'] = []
        node.driver_internal_info = driver_internal_info
        node.save()
        return

    driver_internal_info = node.driver_internal_info
    driver_internal_info['raid_config_substep'] = substep
    driver_internal_info['raid_config_parameters'] = []

    if 'raid_config_job_ids' not in driver_internal_info:
        driver_internal_info['raid_config_job_ids'] = []

    optional = drac_constants.RebootRequired.optional
    all_realtime = all(cntlr['is_reboot_required'] == optional
                       for cntlr in controllers)
    raid_config_job_ids = []
    raid_config_parameters = []
    if all_realtime:
        for controller in controllers:
            realtime_controller = controller['raid_controller']
            job_details = _create_config_job(
                node, controller=realtime_controller,
                reboot=False, realtime=True,
                raid_config_job_ids=raid_config_job_ids,
                raid_config_parameters=raid_config_parameters)

    else:
        for controller in controllers:
            mix_controller = controller['raid_controller']
            reboot = (controller == controllers[-1])
            job_details = _create_config_job(
                node, controller=mix_controller,
                reboot=reboot, realtime=False,
                raid_config_job_ids=raid_config_job_ids,
                raid_config_parameters=raid_config_parameters)

    driver_internal_info['raid_config_job_ids'].extend(job_details[
        'raid_config_job_ids'])

    driver_internal_info['raid_config_parameters'].extend(job_details[
        'raid_config_parameters'])

    node.driver_internal_info = driver_internal_info

    # Signal whether the node has been rebooted, that we do not need to execute
    # the step again, and that this completion of this step is triggered
    # through async polling.
    # NOTE(mgoddard): set_async_step_flags calls node.save().
    deploy_utils.set_async_step_flags(
        node,
        reboot=not all_realtime,
        skip_current_step=True,
        polling=True)

    return deploy_utils.get_async_step_return_state(node)


def _create_virtual_disks(task, node):
    logical_disks_to_create = node.driver_internal_info[
        'logical_disks_to_create']

    controllers = list()
    for logical_disk in logical_disks_to_create:
        controller = dict()
        controller_cap = create_virtual_disk(
            node,
            raid_controller=logical_disk['controller'],
            physical_disks=logical_disk['physical_disks'],
            raid_level=logical_disk['raid_level'],
            size_mb=logical_disk['size_mb'],
            disk_name=logical_disk.get('name'),
            span_length=logical_disk.get('span_length'),
            span_depth=logical_disk.get('span_depth'))
        controller['raid_controller'] = logical_disk['controller']
        controller['is_reboot_required'] = controller_cap[
            'is_reboot_required']
        controller['is_commit_required'] = controller_cap[
            'is_commit_required']
        if controller not in controllers:
            controllers.append(controller)

    return _commit_to_controllers(node, controllers)


def _get_disk_free_size_mb(disk, pending_delete):
    """Return the size of free space on the disk in MB.

    :param disk: a PhysicalDisk object.
    :param pending_delete: Whether there is a pending deletion of all virtual
        disks.
    """
    return disk.size_mb if pending_delete else disk.free_size_mb


class DracWSManRAID(base.RAIDInterface):

    def get_properties(self):
        """Return the properties of the interface."""
        return drac_common.COMMON_PROPERTIES

    @base.deploy_step(priority=0,
                      argsinfo=base.RAID_APPLY_CONFIGURATION_ARGSINFO)
    def apply_configuration(self, task, raid_config, create_root_volume=True,
                            create_nonroot_volumes=False,
                            delete_existing=True):
        return super(DracRAID, self).apply_configuration(
            task, raid_config, create_root_volume=create_root_volume,
            create_nonroot_volumes=create_nonroot_volumes,
            delete_existing=delete_existing)

    @METRICS.timer('DracRAID.create_configuration')
    @base.clean_step(priority=0, abortable=False, argsinfo={
        'create_root_volume': {
            'description': (
                'This specifies whether to create the root volume. '
                'Defaults to `True`.'
            ),
            'required': False
        },
        'create_nonroot_volumes': {
            'description': (
                'This specifies whether to create the non-root volumes. '
                'Defaults to `True`.'
            ),
            'required': False
        },
        "delete_existing": {
            "description": (
                "Setting this to 'True' indicates to delete existing RAID "
                "configuration prior to creating the new configuration. "
                "Default value is 'False'."
            ),
            "required": False,
        }
    })
    def create_configuration(self, task,
                             create_root_volume=True,
                             create_nonroot_volumes=True,
                             delete_existing=False):
        """Create the RAID configuration.

        This method creates the RAID configuration on the given node.

        :param task: a TaskManager instance containing the node to act on.
        :param create_root_volume: If True, a root volume is created
            during RAID configuration. Otherwise, no root volume is
            created. Default is True.
        :param create_nonroot_volumes: If True, non-root volumes are
            created. If False, no non-root volumes are created. Default
            is True.
        :param delete_existing: Setting this to True indicates to delete RAID
            configuration prior to creating the new configuration. Default is
            False.
        :returns: states.CLEANWAIT (cleaning) or states.DEPLOYWAIT (deployment)
            if creation is in progress asynchronously or None if it is
            completed.
        :raises: MissingParameterValue, if node.target_raid_config is missing
            or empty.
        :raises: DracOperationError on an error from python-dracclient.
        """
        node = task.node

        logical_disks = node.target_raid_config['logical_disks']

        for disk in logical_disks:
            if disk['size_gb'] == 'MAX' and 'physical_disks' not in disk:
                raise exception.InvalidParameterValue(
                    _("create_configuration called with invalid "
                      "target_raid_configuration for node %(node_id)s. "
                      "'physical_disks' is missing from logical_disk while "
                      "'size_gb'='MAX' was requested: "
                      "%(logical_disk)s") % {'node_id': node.uuid,
                                             'logical_disk': disk})

            if disk['size_gb'] == 'MAX':
                disk['size_mb'] = 'MAX'
            else:
                disk['size_mb'] = disk['size_gb'] * units.Ki

            del disk['size_gb']

        if delete_existing:
            self._delete_configuration_no_commit(task)

        physical_disks = list_physical_disks(node)
        logical_disks = _find_configuration(logical_disks, physical_disks,
                                            pending_delete=delete_existing)

        logical_disks_to_create = _filter_logical_disks(
            logical_disks, create_root_volume, create_nonroot_volumes)

        controllers_to_physical_disk_ids = defaultdict(list)
        for logical_disk in logical_disks_to_create:
            # Not applicable to JBOD logical disks.
            if logical_disk['raid_level'] == 'JBOD':
                continue

            for physical_disk_name in logical_disk['physical_disks']:
                controllers_to_physical_disk_ids[
                    logical_disk['controller']].append(
                    physical_disk_name)

        # adding logical_disks to driver_internal_info to create virtual disks
        driver_internal_info = node.driver_internal_info
        driver_internal_info[
            "logical_disks_to_create"] = logical_disks_to_create
        node.driver_internal_info = driver_internal_info
        node.save()

        commit_results = None
        if logical_disks_to_create:
            LOG.debug(
                "Converting physical disks configured to back RAID "
                "logical disks to RAID mode for node %(node_uuid)s ",
                {"node_uuid": node.uuid})
            raid_mode = drac_constants.RaidStatus.raid
            commit_results = _change_physical_disk_mode(
                node, raid_mode,
                controllers_to_physical_disk_ids,
                substep="create_virtual_disks")

        if commit_results:
            return commit_results
        else:
            LOG.debug("Controller does not support drives conversion "
                      "so creating virtual disks")
            return _create_virtual_disks(task, node)

    @METRICS.timer('DracRAID.delete_configuration')
    @base.clean_step(priority=0)
    @base.deploy_step(priority=0)
    def delete_configuration(self, task):
        """Delete the RAID configuration.

        :param task: a TaskManager instance containing the node to act on.
        :returns: states.CLEANWAIT (cleaning) or states.DEPLOYWAIT (deployment)
            if deletion is in progress asynchronously or None if it is
            completed.
        :raises: DracOperationError on an error from python-dracclient.
        """

        controllers = self._delete_configuration_no_commit(task)
        return _commit_to_controllers(task.node, controllers,
                                      substep="delete_foreign_config")

    @METRICS.timer('DracRAID.get_logical_disks')
    def get_logical_disks(self, task):
        """Get the RAID configuration of the node.

        :param task: a TaskManager instance containing the node to act on.
        :returns: A dictionary of properties.
        :raises: DracOperationError on an error from python-dracclient.
        """
        node = task.node

        logical_disks = []
        for disk in list_virtual_disks(node):
            logical_disk = {
                'id': disk.id,
                'controller': disk.controller,
                'size_gb': int(disk.size_mb / units.Ki),
                'raid_level': disk.raid_level
            }

            if disk.name is not None:
                logical_disk['name'] = disk.name

            logical_disks.append(logical_disk)

        return {'logical_disks': logical_disks}

    @METRICS.timer('DracRAID._query_raid_config_job_status')
    @periodics.periodic(
        spacing=CONF.drac.query_raid_config_job_status_interval)
    def _query_raid_config_job_status(self, manager, context):
        """Periodic task to check the progress of running RAID config jobs."""

        filters = {'reserved': False, 'maintenance': False}
        fields = ['driver_internal_info']

        node_list = manager.iter_nodes(fields=fields, filters=filters)
        for (node_uuid, driver, conductor_group,
             driver_internal_info) in node_list:
            try:
                lock_purpose = 'checking async raid configuration jobs'
                with task_manager.acquire(context, node_uuid,
                                          purpose=lock_purpose,
                                          shared=True) as task:
                    if not isinstance(task.driver.raid, DracWSManRAID):
                        continue

                    job_ids = driver_internal_info.get('raid_config_job_ids')
                    if not job_ids:
                        continue

                    self._check_node_raid_jobs(task)

            except exception.NodeNotFound:
                LOG.info("During query_raid_config_job_status, node "
                         "%(node)s was not found and presumed deleted by "
                         "another process.", {'node': node_uuid})
            except exception.NodeLocked:
                LOG.info("During query_raid_config_job_status, node "
                         "%(node)s was already locked by another process. "
                         "Skip.", {'node': node_uuid})

    @METRICS.timer('DracRAID._check_node_raid_jobs')
    def _check_node_raid_jobs(self, task):
        """Check the progress of running RAID config jobs of a node."""

        node = task.node
        raid_config_job_ids = node.driver_internal_info['raid_config_job_ids']
        finished_job_ids = []

        for config_job_id in raid_config_job_ids:
            config_job = drac_job.get_job(node, job_id=config_job_id)

            if config_job is None or config_job.status == 'Completed':
                finished_job_ids.append(config_job_id)
            elif config_job.status == 'Failed':
                finished_job_ids.append(config_job_id)
                self._set_raid_config_job_failure(node)

        if not finished_job_ids:
            return

        task.upgrade_lock()
        self._delete_cached_config_job_id(node, finished_job_ids)

        if not node.driver_internal_info.get('raid_config_job_failure',
                                             False):
            if 'raid_config_substep' in node.driver_internal_info:
                substep = node.driver_internal_info['raid_config_substep']

                if substep == 'delete_foreign_config':
                    foreign_drives = self._execute_foreign_drives(task, node)
                    if foreign_drives is None:
                        return self._convert_drives(task, node)
                elif substep == 'physical_disk_conversion':
                    self._convert_drives(task, node)
                elif substep == "create_virtual_disks":
                    return _create_virtual_disks(task, node)
                elif substep == 'completed':
                    self._complete_raid_substep(task, node)
            else:
                self._complete_raid_substep(task, node)
        else:
            self._clear_raid_substep(node)
            self._clear_raid_config_job_failure(node)
            self._set_failed(task, config_job)

    def _execute_foreign_drives(self, task, node):
        controllers = list()
        jobs_required = False
        for controller_id in node.driver_internal_info[
                'raid_config_parameters']:
            controller_cap = clear_foreign_config(
                node, controller_id)
            controller = {
                'raid_controller': controller_id,
                'is_reboot_required': controller_cap['is_reboot_required'],
                'is_commit_required': controller_cap['is_commit_required']}
            controllers.append(controller)
            jobs_required = jobs_required or controller_cap[
                'is_commit_required']

        if not jobs_required:
            LOG.info(
                "No foreign drives detected, so "
                "resume %s", "cleaning" if node.clean_step else "deployment")
            return None
        else:
            return _commit_to_controllers(
                node,
                controllers,
                substep='physical_disk_conversion')

    def _complete_raid_substep(self, task, node):
        self._clear_raid_substep(node)
        self._resume(task)

    def _convert_drives(self, task, node):
        jbod = drac_constants.RaidStatus.jbod
        drives_results = _change_physical_disk_mode(
            node, mode=jbod)
        if drives_results is None:
            LOG.debug("Controller does not support drives "
                      "conversion on %(node_uuid)s",
                      {'node_uuid': node.uuid})
            self._complete_raid_substep(task, node)

    def _clear_raid_substep(self, node):
        driver_internal_info = node.driver_internal_info
        driver_internal_info.pop('raid_config_substep', None)
        driver_internal_info.pop('raid_config_parameters', None)
        node.driver_internal_info = driver_internal_info
        node.save()

    def _set_raid_config_job_failure(self, node):
        driver_internal_info = node.driver_internal_info
        driver_internal_info['raid_config_job_failure'] = True
        node.driver_internal_info = driver_internal_info
        node.save()

    def _clear_raid_config_job_failure(self, node):
        driver_internal_info = node.driver_internal_info
        del driver_internal_info['raid_config_job_failure']
        node.driver_internal_info = driver_internal_info
        node.save()

    def _delete_cached_config_job_id(self, node, finished_config_job_ids=None):
        if finished_config_job_ids is None:
            finished_config_job_ids = []
        driver_internal_info = node.driver_internal_info
        unfinished_job_ids = [job_id for job_id
                              in driver_internal_info['raid_config_job_ids']
                              if job_id not in finished_config_job_ids]
        driver_internal_info['raid_config_job_ids'] = unfinished_job_ids
        node.driver_internal_info = driver_internal_info
        node.save()

    def _set_failed(self, task, config_job):
        LOG.error("RAID configuration job failed for node %(node)s. "
                  "Failed config job: %(config_job_id)s. "
                  "Message: '%(message)s'.",
                  {'node': task.node.uuid, 'config_job_id': config_job.id,
                   'message': config_job.message})
        task.node.last_error = config_job.message
        task.process_event('fail')

    def _resume(self, task):
        raid_common.update_raid_info(
            task.node, self.get_logical_disks(task))
        if task.node.clean_step:
            manager_utils.notify_conductor_resume_clean(task)
        else:
            manager_utils.notify_conductor_resume_deploy(task)

    def _delete_configuration_no_commit(self, task):
        """Delete existing RAID configuration without committing the change.

        :param task: A TaskManager instance.
        :returns: A set of names of RAID controllers which need RAID changes to
            be committed.
        """
        node = task.node
        controllers = list()
        drac_raid_controllers = list_raid_controllers(node)
        for cntrl in drac_raid_controllers:
            if _is_raid_controller(node, cntrl.id, drac_raid_controllers):
                controller = dict()
                controller_cap = _reset_raid_config(node, cntrl.id)
                controller["raid_controller"] = cntrl.id
                controller["is_reboot_required"] = controller_cap[
                    "is_reboot_required"]
                controller["is_commit_required"] = controller_cap[
                    "is_commit_required"]
                controllers.append(controller)
        return controllers


class DracRAID(DracWSManRAID):
    """Class alias of class DracWSManRAID.

    This class provides ongoing support of the deprecated 'idrac' RAID
    interface implementation entrypoint.

    All bug fixes and new features should be implemented in its base
    class, DracWSManRAID. That makes them available to both the
    deprecated 'idrac' and new 'idrac-wsman' entrypoints. Such changes
    should not be made to this class.
    """

    def __init__(self):
        super(DracRAID, self).__init__()
        LOG.warning("RAID interface 'idrac' is deprecated and may be removed "
                    "in a future release. Use 'idrac-wsman' instead.")