¡@

Home 

OpenStack Study: api.py

OpenStack Index

**** CubicPower OpenStack Study ****

# Copyright 2010 United States Government as represented by the

# Administrator of the National Aeronautics and Space Administration.

#

# Copyright 2011, Piston Cloud Computing, 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.

"""

Utility methods to resize, repartition, and modify disk images.

Includes injection of SSH PGP keys into authorized_keys file.

"""

import os

import random

import tempfile

if os.name != 'nt':

import crypt

from oslo.config import cfg

from nova import exception

from nova.openstack.common.gettextutils import _

from nova.openstack.common import jsonutils

from nova.openstack.common import log as logging

from nova.openstack.common import processutils

from nova import paths

from nova import utils

from nova.virt.disk.mount import api as mount

from nova.virt.disk.vfs import api as vfs

from nova.virt import images

LOG = logging.getLogger(__name__)

disk_opts = [

cfg.StrOpt('injected_network_template',

default=paths.basedir_def('nova/virt/interfaces.template'),

help='Template file for injected network'),

# NOTE(yamahata): ListOpt won't work because the command may include a

# comma. For example:

#

# mkfs.ext3 -O dir_index,extent -E stride=8,stripe-width=16

# --label %(fs_label)s %(target)s

#

# list arguments are comma separated and there is no way to

# escape such commas.

#

cfg.MultiStrOpt('virt_mkfs',

default=[],

help='Name of the mkfs commands for ephemeral device. '

'The format is ='),

cfg.BoolOpt('resize_fs_using_block_device',

default=False,

help='Attempt to resize the filesystem by accessing the '

'image over a block device. This is done by the host '

'and may not be necessary if the image contains a recent '

'version of cloud-init. Possible mechanisms require '

'the nbd driver (for qcow and raw), or loop (for raw).'),

]

CONF = cfg.CONF

CONF.register_opts(disk_opts)

CONF.import_opt('default_ephemeral_format', 'nova.virt.driver')

_MKFS_COMMAND = {}

_DEFAULT_MKFS_COMMAND = None

_DEFAULT_FS_BY_OSTYPE = {'linux': 'ext3',

'windows': 'ntfs'}

for s in CONF.virt_mkfs:

# NOTE(yamahata): mkfs command may includes '=' for its options.

# So item.partition('=') doesn't work here

os_type, mkfs_command = s.split('=', 1)

if os_type:

_MKFS_COMMAND[os_type] = mkfs_command

if os_type == 'default':

_DEFAULT_MKFS_COMMAND = mkfs_command

**** CubicPower OpenStack Study ****

def get_fs_type_for_os_type(os_type):

    return os_type if _MKFS_COMMAND.get(os_type) else 'default'

**** CubicPower OpenStack Study ****

def mkfs(os_type, fs_label, target, run_as_root=True):

    """Format a file or block device using

       a user provided command for each os type.

       If user has not provided any configuration,

       format type will be used according to a

       default_ephemeral_format configuration

       or a system defaults.

    """

    mkfs_command = (_MKFS_COMMAND.get(os_type, _DEFAULT_MKFS_COMMAND) or

                    '') % {'fs_label': fs_label, 'target': target}

    if mkfs_command:

        utils.execute(*mkfs_command.split(), run_as_root=run_as_root)

    else:

        default_fs = CONF.default_ephemeral_format

        if not default_fs:

            default_fs = _DEFAULT_FS_BY_OSTYPE.get(os_type, 'ext3')

        utils.mkfs(default_fs, target, fs_label, run_as_root=run_as_root)

**** CubicPower OpenStack Study ****

def resize2fs(image, check_exit_code=False, run_as_root=False):

    try:

        utils.execute('e2fsck',

                      '-fp',

                      image,

                      check_exit_code=[0, 1, 2],

                      run_as_root=run_as_root)

    except processutils.ProcessExecutionError as exc:

        LOG.debug(_("Checking the file system with e2fsck has failed, "

                    "the resize will be aborted. (%s)"), exc)

    else:

        utils.execute('resize2fs',

                      image,

                      check_exit_code=check_exit_code,

                      run_as_root=run_as_root)

**** CubicPower OpenStack Study ****

def get_disk_size(path):

    """Get the (virtual) size of a disk image

    :param path: Path to the disk image

    :returns: Size (in bytes) of the given disk image as it would be seen

              by a virtual machine.

    """

    return images.qemu_img_info(path).virtual_size

**** CubicPower OpenStack Study ****

def extend(image, size, use_cow=False):

    """Increase image to size."""

    if not can_resize_image(image, size):

        return

    utils.execute('qemu-img', 'resize', image, size)

    # if we can't access the filesystem, we can't do anything more

    if not is_image_partitionless(image, use_cow):

        return

    def safe_resize2fs(dev, run_as_root=False, finally_call=lambda: None):

        try:

            resize2fs(dev, run_as_root=run_as_root, check_exit_code=[0])

        except processutils.ProcessExecutionError as exc:

            LOG.debug(_("Resizing the file system with resize2fs "

                        "has failed with error: %s"), exc)

        finally:

            finally_call()

    # NOTE(vish): attempts to resize filesystem

    if use_cow:

        if CONF.resize_fs_using_block_device:

            # in case of non-raw disks we can't just resize the image, but

            # rather the mounted device instead

            mounter = mount.Mount.instance_for_format(

                image, None, None, 'qcow2')

            if mounter.get_dev():

                safe_resize2fs(mounter.device,

                               run_as_root=True,

                               finally_call=mounter.unget_dev)

    else:

        safe_resize2fs(image)

**** CubicPower OpenStack Study ****

 def safe_resize2fs(dev, run_as_root=False, finally_call=lambda: None):

        try:

            resize2fs(dev, run_as_root=run_as_root, check_exit_code=[0])

        except processutils.ProcessExecutionError as exc:

            LOG.debug(_("Resizing the file system with resize2fs "

                        "has failed with error: %s"), exc)

        finally:

            finally_call()

    # NOTE(vish): attempts to resize filesystem

    if use_cow:

        if CONF.resize_fs_using_block_device:

            # in case of non-raw disks we can't just resize the image, but

            # rather the mounted device instead

            mounter = mount.Mount.instance_for_format(

                image, None, None, 'qcow2')

            if mounter.get_dev():

                safe_resize2fs(mounter.device,

                               run_as_root=True,

                               finally_call=mounter.unget_dev)

    else:

        safe_resize2fs(image)

**** CubicPower OpenStack Study ****

def can_resize_image(image, size):

    """Check whether we can resize the container image file."""

    LOG.debug(_('Checking if we can resize image %(image)s. '

                'size=%(size)s'), {'image': image, 'size': size})

    # Check that we're increasing the size

    virt_size = get_disk_size(image)

    if virt_size >= size:

        LOG.debug(_('Cannot resize image %s to a smaller size.'),

                  image)

        return False

    return True

**** CubicPower OpenStack Study ****

def is_image_partitionless(image, use_cow=False):

    """Check whether we can resize contained file system."""

    LOG.debug(_('Checking if we can resize filesystem inside %(image)s. '

                'CoW=%(use_cow)s'), {'image': image, 'use_cow': use_cow})

    # Check the image is unpartitioned

    if use_cow:

        try:

            fs = vfs.VFS.instance_for_image(image, 'qcow2', None)

            fs.setup()

            fs.teardown()

        except exception.NovaException as e:

            LOG.debug(_('Unable to mount image %(image)s with '

                        'error %(error)s. Cannot resize.'),

                      {'image': image,

                       'error': e})

            return False

    else:

        # For raw, we can directly inspect the file system

        try:

            utils.execute('e2label', image)

        except processutils.ProcessExecutionError as e:

            LOG.debug(_('Unable to determine label for image %(image)s with '

                        'error %(error)s. Cannot resize.'),

                      {'image': image,

                       'error': e})

            return False

    return True

**** CubicPower OpenStack Study ****

class _DiskImage(object):

"""Provide operations on a disk image file."""

tmp_prefix = 'openstack-disk-mount-tmp'

**** CubicPower OpenStack Study ****

    def __init__(self, image, partition=None, use_cow=False, mount_dir=None):

        # These passed to each mounter

        self.image = image

        self.partition = partition

        self.mount_dir = mount_dir

        self.use_cow = use_cow

        # Internal

        self._mkdir = False

        self._mounter = None

        self._errors = []

        if mount_dir:

            device = self._device_for_path(mount_dir)

            if device:

                self._reset(device)

    @staticmethod

**** CubicPower OpenStack Study ****

    def _device_for_path(path):

        device = None

        path = os.path.realpath(path)

        with open("/proc/mounts", 'r') as ifp:

            for line in ifp:

                fields = line.split()

                if fields[1] == path:

                    device = fields[0]

                    break

        return device

**** CubicPower OpenStack Study ****

    def _reset(self, device):

        """Reset internal state for a previously mounted directory."""

        self._mounter = mount.Mount.instance_for_device(self.image,

                                                        self.mount_dir,

                                                        self.partition,

                                                        device)

        mount_name = os.path.basename(self.mount_dir or '')

        self._mkdir = mount_name.startswith(self.tmp_prefix)

    @property

**** CubicPower OpenStack Study ****

    def errors(self):

        """Return the collated errors from all operations."""

        return '\n--\n'.join([''] + self._errors)

**** CubicPower OpenStack Study ****

    def mount(self):

        """Mount a disk image, using the object attributes.

        The first supported means provided by the mount classes is used.

        True, or False is returned and the 'errors' attribute

        contains any diagnostics.

        """

        if self._mounter:

            raise exception.NovaException(_('image already mounted'))

        if not self.mount_dir:

            self.mount_dir = tempfile.mkdtemp(prefix=self.tmp_prefix)

            self._mkdir = True

        imgfmt = "raw"

        if self.use_cow:

            imgfmt = "qcow2"

        mounter = mount.Mount.instance_for_format(self.image,

                                                  self.mount_dir,

                                                  self.partition,

                                                  imgfmt)

        if mounter.do_mount():

            self._mounter = mounter

            return self._mounter.device

        else:

            LOG.debug(mounter.error)

            self._errors.append(mounter.error)

            return None

**** CubicPower OpenStack Study ****

    def umount(self):

        """Umount a mount point from the filesystem."""

        if self._mounter:

            self._mounter.do_umount()

            self._mounter = None

**** CubicPower OpenStack Study ****

    def teardown(self):

        """Remove a disk image from the file system."""

        try:

            if self._mounter:

                self._mounter.do_teardown()

                self._mounter = None

        finally:

            if self._mkdir:

                os.rmdir(self.mount_dir)

# Public module functions

def inject_data(image, key=None, net=None, metadata=None, admin_password=None,

                files=None, partition=None, use_cow=False, mandatory=()):

    """Inject the specified items into a disk image.

    If an item name is not specified in the MANDATORY iterable, then a warning

    is logged on failure to inject that item, rather than raising an exception.

    it will mount the image as a fully partitioned disk and attempt to inject

    into the specified partition number.

    If PARTITION is not specified the image is mounted as a single partition.

    Returns True if all requested operations completed without issue.

    Raises an exception if a mandatory item can't be injected.

    """

    LOG.debug(_("Inject data image=%(image)s key=%(key)s net=%(net)s "

                "metadata=%(metadata)s admin_password= "

                "files=%(files)s partition=%(partition)s use_cow=%(use_cow)s"),

              {'image': image, 'key': key, 'net': net, 'metadata': metadata,

               'files': files, 'partition': partition, 'use_cow': use_cow})

    fmt = "raw"

    if use_cow:

        fmt = "qcow2"

    try:

        # Note(mrda): Test if the image exists first to short circuit errors

        os.stat(image)

        fs = vfs.VFS.instance_for_image(image, fmt, partition)

        fs.setup()

    except Exception as e:

        # If a mandatory item is passed to this function,

        # then reraise the exception to indicate the error.

        for inject in mandatory:

            inject_val = locals()[inject]

            if inject_val:

                raise

        LOG.warn(_('Ignoring error injecting data into image '

                   '(%(e)s)'), {'e': e})

        return False

    try:

        return inject_data_into_fs(fs, key, net, metadata,

                                   admin_password, files, mandatory)

    finally:

        fs.teardown()

def setup_container(image, container_dir, use_cow=False):

    """Setup the LXC container.

    It will mount the loopback image to the container directory in order

    to create the root filesystem for the container.

    Returns path of image device which is mounted to the container directory.

    """

    img = _DiskImage(image=image, use_cow=use_cow, mount_dir=container_dir)

    dev = img.mount()

    if dev is None:

        LOG.error(_("Failed to mount container filesystem '%(image)s' "

                    "on '%(target)s': %(errors)s"),

                  {"image": img, "target": container_dir,

                   "errors": img.errors})

        raise exception.NovaException(img.errors)

    return dev

def teardown_container(container_dir, container_root_device=None):

    """Teardown the container rootfs mounting once it is spawned.

    It will umount the container that is mounted,

    and delete any linked devices.

    """

    try:

        img = _DiskImage(image=None, mount_dir=container_dir)

        img.teardown()

        # Make sure container_root_device is released when teardown container.

        if container_root_device:

            if 'loop' in container_root_device:

                LOG.debug(_("Release loop device %s"), container_root_device)

                utils.execute('losetup', '--detach', container_root_device,

                              run_as_root=True, attempts=3)

            else:

                LOG.debug(_('Release nbd device %s'), container_root_device)

                utils.execute('qemu-nbd', '-d', container_root_device,

                              run_as_root=True)

    except Exception as exn:

        LOG.exception(_('Failed to teardown container filesystem: %s'), exn)

def clean_lxc_namespace(container_dir):

    """Clean up the container namespace rootfs mounting one spawned.

    It will umount the mounted names that are mounted

    but leave the linked devices alone.

    """

    try:

        img = _DiskImage(image=None, mount_dir=container_dir)

        img.umount()

    except Exception as exn:

        LOG.exception(_('Failed to umount container filesystem: %s'), exn)

def inject_data_into_fs(fs, key, net, metadata, admin_password, files,

                        mandatory=()):

    """Injects data into a filesystem already mounted by the caller.

    Virt connections can call this directly if they mount their fs

    in a different way to inject_data.

    If an item name is not specified in the MANDATORY iterable, then a warning

    is logged on failure to inject that item, rather than raising an exception.

    Returns True if all requested operations completed without issue.

    Raises an exception if a mandatory item can't be injected.

    """

    status = True

    for inject in ('key', 'net', 'metadata', 'admin_password', 'files'):

        inject_val = locals()[inject]

        inject_func = globals()['_inject_%s_into_fs' % inject]

        if inject_val:

            try:

                inject_func(inject_val, fs)

            except Exception as e:

                if inject in mandatory:

                    raise

                LOG.warn(_('Ignoring error injecting %(inject)s into image '

                           '(%(e)s)'), {'e': e, 'inject': inject})

                status = False

    return status

def _inject_files_into_fs(files, fs):

    for (path, contents) in files:

        # NOTE(wangpan): Ensure the parent dir of injecting file exists

        parent_dir = os.path.dirname(path)

        if (len(parent_dir) > 0 and parent_dir != "/"

                and not fs.has_file(parent_dir)):

            fs.make_path(parent_dir)

            fs.set_ownership(parent_dir, "root", "root")

            fs.set_permissions(parent_dir, 0o744)

        _inject_file_into_fs(fs, path, contents)

def _inject_file_into_fs(fs, path, contents, append=False):

    LOG.debug(_("Inject file fs=%(fs)s path=%(path)s append=%(append)s"),

              {'fs': fs, 'path': path, 'append': append})

    if append:

        fs.append_file(path, contents)

    else:

        fs.replace_file(path, contents)

def _inject_metadata_into_fs(metadata, fs):

    LOG.debug(_("Inject metadata fs=%(fs)s metadata=%(metadata)s"),

              {'fs': fs, 'metadata': metadata})

    _inject_file_into_fs(fs, 'meta.js', jsonutils.dumps(metadata))

def _setup_selinux_for_keys(fs, sshdir):

    """Get selinux guests to ensure correct context on injected keys."""

    if not fs.has_file(os.path.join("etc", "selinux")):

        return

    rclocal = os.path.join('etc', 'rc.local')

    rc_d = os.path.join('etc', 'rc.d')

    if not fs.has_file(rclocal) and fs.has_file(rc_d):

        rclocal = os.path.join(rc_d, 'rc.local')

    # Note some systems end rc.local with "exit 0"

    # and so to append there you'd need something like:

    #  utils.execute('sed', '-i', '${/^exit 0$/d}' rclocal, run_as_root=True)

    restorecon = [

        '\n',

        '# Added by Nova to ensure injected ssh keys have the right context\n',

        'restorecon -RF %s 2>/dev/null || :\n' % sshdir,

    ]

    if not fs.has_file(rclocal):

        restorecon.insert(0, '#!/bin/sh')

    _inject_file_into_fs(fs, rclocal, ''.join(restorecon), append=True)

    fs.set_permissions(rclocal, 0o700)

def _inject_key_into_fs(key, fs):

    """Add the given public ssh key to root's authorized_keys.

    key is an ssh key string.

    fs is the path to the base of the filesystem into which to inject the key.

    """

    LOG.debug(_("Inject key fs=%(fs)s key=%(key)s"), {'fs': fs, 'key': key})

    sshdir = os.path.join('root', '.ssh')

    fs.make_path(sshdir)

    fs.set_ownership(sshdir, "root", "root")

    fs.set_permissions(sshdir, 0o700)

    keyfile = os.path.join(sshdir, 'authorized_keys')

    key_data = ''.join([

        '\n',

        '# The following ssh key was injected by Nova',

        '\n',

        key.strip(),

        '\n',

    ])

    _inject_file_into_fs(fs, keyfile, key_data, append=True)

    fs.set_permissions(keyfile, 0o600)

    _setup_selinux_for_keys(fs, sshdir)

def _inject_net_into_fs(net, fs):

    """Inject /etc/network/interfaces into the filesystem rooted at fs.

    net is the contents of /etc/network/interfaces.

    """

    LOG.debug(_("Inject key fs=%(fs)s net=%(net)s"), {'fs': fs, 'net': net})

    netdir = os.path.join('etc', 'network')

    fs.make_path(netdir)

    fs.set_ownership(netdir, "root", "root")

    fs.set_permissions(netdir, 0o744)

    netfile = os.path.join('etc', 'network', 'interfaces')

    _inject_file_into_fs(fs, netfile, net)

def _inject_admin_password_into_fs(admin_passwd, fs):

    """Set the root password to admin_passwd

    admin_password is a root password

    fs is the path to the base of the filesystem into which to inject

    the key.

    This method modifies the instance filesystem directly,

    and does not require a guest agent running in the instance.

    """

    # The approach used here is to copy the password and shadow

    # files from the instance filesystem to local files, make any

    # necessary changes, and then copy them back.

    LOG.debug(_("Inject admin password fs=%(fs)s "

                "admin_passwd="), {'fs': fs})

    admin_user = 'root'

    passwd_path = os.path.join('etc', 'passwd')

    shadow_path = os.path.join('etc', 'shadow')

    passwd_data = fs.read_file(passwd_path)

    shadow_data = fs.read_file(shadow_path)

    new_shadow_data = _set_passwd(admin_user, admin_passwd,

                                  passwd_data, shadow_data)

    fs.replace_file(shadow_path, new_shadow_data)

def _generate_salt():

    salt_set = ('abcdefghijklmnopqrstuvwxyz'

                'ABCDEFGHIJKLMNOPQRSTUVWXYZ'

                '0123456789./')

    salt = 16 * ' '

    return ''.join([random.choice(salt_set) for c in salt])

def _set_passwd(username, admin_passwd, passwd_data, shadow_data):

    """set the password for username to admin_passwd

    The passwd_file is not modified.  The shadow_file is updated.

    if the username is not found in both files, an exception is raised.

    :param username: the username

    :param encrypted_passwd: the  encrypted password

    :param passwd_file: path to the passwd file

    :param shadow_file: path to the shadow password file

    :returns: nothing

    :raises: exception.NovaException(), IOError()

    """

    if os.name == 'nt':

        raise exception.NovaException(_('Not implemented on Windows'))

    # encryption algo - id pairs for crypt()

    algos = {'SHA-512': '$6$', 'SHA-256': '$5$', 'MD5': '$1$', 'DES': ''}

    salt = _generate_salt()

    # crypt() depends on the underlying libc, and may not support all

    # forms of hash. We try md5 first. If we get only 13 characters back,

    # then the underlying crypt() didn't understand the '$n$salt' magic,

    # so we fall back to DES.

    # md5 is the default because it's widely supported. Although the

    # local crypt() might support stronger SHA, the target instance

    # might not.

    encrypted_passwd = crypt.crypt(admin_passwd, algos['MD5'] + salt)

    if len(encrypted_passwd) == 13:

        encrypted_passwd = crypt.crypt(admin_passwd, algos['DES'] + salt)

    p_file = passwd_data.split("\n")

    s_file = shadow_data.split("\n")

     # username MUST exist in passwd file or it's an error

    found = False

    for entry in p_file:

        split_entry = entry.split(':')

        if split_entry[0] == username:

            found = True

            break

    if not found:

        msg = _('User %(username)s not found in password file.')

        raise exception.NovaException(msg % username)

    # update password in the shadow file.It's an error if the

    # the user doesn't exist.

    new_shadow = list()

    found = False

    for entry in s_file:

        split_entry = entry.split(':')

        if split_entry[0] == username:

            split_entry[1] = encrypted_passwd

            found = True

        new_entry = ':'.join(split_entry)

        new_shadow.append(new_entry)

    if not found:

        msg = _('User %(username)s not found in shadow file.')

        raise exception.NovaException(msg % username)

    return "\n".join(new_shadow)

**** CubicPower OpenStack Study ****

def inject_data(image, key=None, net=None, metadata=None, admin_password=None,

                files=None, partition=None, use_cow=False, mandatory=()):

    """Inject the specified items into a disk image.

    If an item name is not specified in the MANDATORY iterable, then a warning

    is logged on failure to inject that item, rather than raising an exception.

    it will mount the image as a fully partitioned disk and attempt to inject

    into the specified partition number.

    If PARTITION is not specified the image is mounted as a single partition.

    Returns True if all requested operations completed without issue.

    Raises an exception if a mandatory item can't be injected.

    """

    LOG.debug(_("Inject data image=%(image)s key=%(key)s net=%(net)s "

                "metadata=%(metadata)s admin_password= "

                "files=%(files)s partition=%(partition)s use_cow=%(use_cow)s"),

              {'image': image, 'key': key, 'net': net, 'metadata': metadata,

               'files': files, 'partition': partition, 'use_cow': use_cow})

    fmt = "raw"

    if use_cow:

        fmt = "qcow2"

    try:

        # Note(mrda): Test if the image exists first to short circuit errors

        os.stat(image)

        fs = vfs.VFS.instance_for_image(image, fmt, partition)

        fs.setup()

    except Exception as e:

        # If a mandatory item is passed to this function,

        # then reraise the exception to indicate the error.

        for inject in mandatory:

            inject_val = locals()[inject]

            if inject_val:

                raise

        LOG.warn(_('Ignoring error injecting data into image '

                   '(%(e)s)'), {'e': e})

        return False

    try:

        return inject_data_into_fs(fs, key, net, metadata,

                                   admin_password, files, mandatory)

    finally:

        fs.teardown()

**** CubicPower OpenStack Study ****

def setup_container(image, container_dir, use_cow=False):

    """Setup the LXC container.

    It will mount the loopback image to the container directory in order

    to create the root filesystem for the container.

    Returns path of image device which is mounted to the container directory.

    """

    img = _DiskImage(image=image, use_cow=use_cow, mount_dir=container_dir)

    dev = img.mount()

    if dev is None:

        LOG.error(_("Failed to mount container filesystem '%(image)s' "

                    "on '%(target)s': %(errors)s"),

                  {"image": img, "target": container_dir,

                   "errors": img.errors})

        raise exception.NovaException(img.errors)

    return dev

**** CubicPower OpenStack Study ****

def teardown_container(container_dir, container_root_device=None):

    """Teardown the container rootfs mounting once it is spawned.

    It will umount the container that is mounted,

    and delete any linked devices.

    """

    try:

        img = _DiskImage(image=None, mount_dir=container_dir)

        img.teardown()

        # Make sure container_root_device is released when teardown container.

        if container_root_device:

            if 'loop' in container_root_device:

                LOG.debug(_("Release loop device %s"), container_root_device)

                utils.execute('losetup', '--detach', container_root_device,

                              run_as_root=True, attempts=3)

            else:

                LOG.debug(_('Release nbd device %s'), container_root_device)

                utils.execute('qemu-nbd', '-d', container_root_device,

                              run_as_root=True)

    except Exception as exn:

        LOG.exception(_('Failed to teardown container filesystem: %s'), exn)

**** CubicPower OpenStack Study ****

def clean_lxc_namespace(container_dir):

    """Clean up the container namespace rootfs mounting one spawned.

    It will umount the mounted names that are mounted

    but leave the linked devices alone.

    """

    try:

        img = _DiskImage(image=None, mount_dir=container_dir)

        img.umount()

    except Exception as exn:

        LOG.exception(_('Failed to umount container filesystem: %s'), exn)

**** CubicPower OpenStack Study ****

def inject_data_into_fs(fs, key, net, metadata, admin_password, files,

                        mandatory=()):

    """Injects data into a filesystem already mounted by the caller.

    Virt connections can call this directly if they mount their fs

    in a different way to inject_data.

    If an item name is not specified in the MANDATORY iterable, then a warning

    is logged on failure to inject that item, rather than raising an exception.

    Returns True if all requested operations completed without issue.

    Raises an exception if a mandatory item can't be injected.

    """

    status = True

    for inject in ('key', 'net', 'metadata', 'admin_password', 'files'):

        inject_val = locals()[inject]

        inject_func = globals()['_inject_%s_into_fs' % inject]

        if inject_val:

            try:

                inject_func(inject_val, fs)

            except Exception as e:

                if inject in mandatory:

                    raise

                LOG.warn(_('Ignoring error injecting %(inject)s into image '

                           '(%(e)s)'), {'e': e, 'inject': inject})

                status = False

    return status

**** CubicPower OpenStack Study ****

def _inject_files_into_fs(files, fs):

    for (path, contents) in files:

        # NOTE(wangpan): Ensure the parent dir of injecting file exists

        parent_dir = os.path.dirname(path)

        if (len(parent_dir) > 0 and parent_dir != "/"

                and not fs.has_file(parent_dir)):

            fs.make_path(parent_dir)

            fs.set_ownership(parent_dir, "root", "root")

            fs.set_permissions(parent_dir, 0o744)

        _inject_file_into_fs(fs, path, contents)

**** CubicPower OpenStack Study ****

def _inject_file_into_fs(fs, path, contents, append=False):

    LOG.debug(_("Inject file fs=%(fs)s path=%(path)s append=%(append)s"),

              {'fs': fs, 'path': path, 'append': append})

    if append:

        fs.append_file(path, contents)

    else:

        fs.replace_file(path, contents)

**** CubicPower OpenStack Study ****

def _inject_metadata_into_fs(metadata, fs):

    LOG.debug(_("Inject metadata fs=%(fs)s metadata=%(metadata)s"),

              {'fs': fs, 'metadata': metadata})

    _inject_file_into_fs(fs, 'meta.js', jsonutils.dumps(metadata))

**** CubicPower OpenStack Study ****

def _setup_selinux_for_keys(fs, sshdir):

    """Get selinux guests to ensure correct context on injected keys."""

    if not fs.has_file(os.path.join("etc", "selinux")):

        return

    rclocal = os.path.join('etc', 'rc.local')

    rc_d = os.path.join('etc', 'rc.d')

    if not fs.has_file(rclocal) and fs.has_file(rc_d):

        rclocal = os.path.join(rc_d, 'rc.local')

    # Note some systems end rc.local with "exit 0"

    # and so to append there you'd need something like:

    #  utils.execute('sed', '-i', '${/^exit 0$/d}' rclocal, run_as_root=True)

    restorecon = [

        '\n',

        '# Added by Nova to ensure injected ssh keys have the right context\n',

        'restorecon -RF %s 2>/dev/null || :\n' % sshdir,

    ]

    if not fs.has_file(rclocal):

        restorecon.insert(0, '#!/bin/sh')

    _inject_file_into_fs(fs, rclocal, ''.join(restorecon), append=True)

    fs.set_permissions(rclocal, 0o700)

**** CubicPower OpenStack Study ****

def _inject_key_into_fs(key, fs):

    """Add the given public ssh key to root's authorized_keys.

    key is an ssh key string.

    fs is the path to the base of the filesystem into which to inject the key.

    """

    LOG.debug(_("Inject key fs=%(fs)s key=%(key)s"), {'fs': fs, 'key': key})

    sshdir = os.path.join('root', '.ssh')

    fs.make_path(sshdir)

    fs.set_ownership(sshdir, "root", "root")

    fs.set_permissions(sshdir, 0o700)

    keyfile = os.path.join(sshdir, 'authorized_keys')

    key_data = ''.join([

        '\n',

        '# The following ssh key was injected by Nova',

        '\n',

        key.strip(),

        '\n',

    ])

    _inject_file_into_fs(fs, keyfile, key_data, append=True)

    fs.set_permissions(keyfile, 0o600)

    _setup_selinux_for_keys(fs, sshdir)

**** CubicPower OpenStack Study ****

def _inject_net_into_fs(net, fs):

    """Inject /etc/network/interfaces into the filesystem rooted at fs.

    net is the contents of /etc/network/interfaces.

    """

    LOG.debug(_("Inject key fs=%(fs)s net=%(net)s"), {'fs': fs, 'net': net})

    netdir = os.path.join('etc', 'network')

    fs.make_path(netdir)

    fs.set_ownership(netdir, "root", "root")

    fs.set_permissions(netdir, 0o744)

    netfile = os.path.join('etc', 'network', 'interfaces')

    _inject_file_into_fs(fs, netfile, net)

**** CubicPower OpenStack Study ****

def _inject_admin_password_into_fs(admin_passwd, fs):

    """Set the root password to admin_passwd

    admin_password is a root password

    fs is the path to the base of the filesystem into which to inject

    the key.

    This method modifies the instance filesystem directly,

    and does not require a guest agent running in the instance.

    """

    # The approach used here is to copy the password and shadow

    # files from the instance filesystem to local files, make any

    # necessary changes, and then copy them back.

    LOG.debug(_("Inject admin password fs=%(fs)s "

                "admin_passwd="), {'fs': fs})

    admin_user = 'root'

    passwd_path = os.path.join('etc', 'passwd')

    shadow_path = os.path.join('etc', 'shadow')

    passwd_data = fs.read_file(passwd_path)

    shadow_data = fs.read_file(shadow_path)

    new_shadow_data = _set_passwd(admin_user, admin_passwd,

                                  passwd_data, shadow_data)

    fs.replace_file(shadow_path, new_shadow_data)

**** CubicPower OpenStack Study ****

def _generate_salt():

    salt_set = ('abcdefghijklmnopqrstuvwxyz'

                'ABCDEFGHIJKLMNOPQRSTUVWXYZ'

                '0123456789./')

    salt = 16 * ' '

    return ''.join([random.choice(salt_set) for c in salt])

**** CubicPower OpenStack Study ****

def _set_passwd(username, admin_passwd, passwd_data, shadow_data):

    """set the password for username to admin_passwd

    The passwd_file is not modified.  The shadow_file is updated.

    if the username is not found in both files, an exception is raised.

    :param username: the username

    :param encrypted_passwd: the  encrypted password

    :param passwd_file: path to the passwd file

    :param shadow_file: path to the shadow password file

    :returns: nothing

    :raises: exception.NovaException(), IOError()

    """

    if os.name == 'nt':

        raise exception.NovaException(_('Not implemented on Windows'))

    # encryption algo - id pairs for crypt()

    algos = {'SHA-512': '$6$', 'SHA-256': '$5$', 'MD5': '$1$', 'DES': ''}

    salt = _generate_salt()

    # crypt() depends on the underlying libc, and may not support all

    # forms of hash. We try md5 first. If we get only 13 characters back,

    # then the underlying crypt() didn't understand the '$n$salt' magic,

    # so we fall back to DES.

    # md5 is the default because it's widely supported. Although the

    # local crypt() might support stronger SHA, the target instance

    # might not.

    encrypted_passwd = crypt.crypt(admin_passwd, algos['MD5'] + salt)

    if len(encrypted_passwd) == 13:

        encrypted_passwd = crypt.crypt(admin_passwd, algos['DES'] + salt)

    p_file = passwd_data.split("\n")

    s_file = shadow_data.split("\n")

     # username MUST exist in passwd file or it's an error

    found = False

    for entry in p_file:

        split_entry = entry.split(':')

        if split_entry[0] == username:

            found = True

            break

    if not found:

        msg = _('User %(username)s not found in password file.')

        raise exception.NovaException(msg % username)

    # update password in the shadow file.It's an error if the

    # the user doesn't exist.

    new_shadow = list()

    found = False

    for entry in s_file:

        split_entry = entry.split(':')

        if split_entry[0] == username:

            split_entry[1] = encrypted_passwd

            found = True

        new_entry = ':'.join(split_entry)

        new_shadow.append(new_entry)

    if not found:

        msg = _('User %(username)s not found in shadow file.')

        raise exception.NovaException(msg % username)

    return "\n".join(new_shadow)