¡@

Home 

OpenStack Study: xattr.py

OpenStack Index

**** CubicPower OpenStack Study ****

# Copyright 2011 OpenStack Foundation

# 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.

"""

Cache driver that uses xattr file tags and requires a filesystem

that has atimes set.

Assumptions

===========

1. Cache data directory exists on a filesytem that updates atime on

reads ('noatime' should NOT be set)

2. Cache data directory exists on a filesystem that supports xattrs.

This is optional, but highly recommended since it allows us to

present ops with useful information pertaining to the cache, like

human readable filenames and statistics.

3. `glance-prune` is scheduled to run as a periodic job via cron. This

is needed to run the LRU prune strategy to keep the cache size

within the limits set by the config file.

Cache Directory Notes

=====================

The image cache data directory contains the main cache path, where the

active cache entries and subdirectories for handling partial downloads

and errored-out cache images.

The layout looks like:

$image_cache_dir/

entry1

entry2

...

incomplete/

invalid/

queue/

"""

from __future__ import absolute_import

from contextlib import contextmanager

import errno

import os

import stat

import time

from oslo.config import cfg

import six

import xattr

from glance.common import exception

from glance.image_cache.drivers import base

import glance.openstack.common.log as logging

LOG = logging.getLogger(__name__)

CONF = cfg.CONF

**** CubicPower OpenStack Study ****

class Driver(base.Driver):

"""

Cache driver that uses xattr file tags and requires a filesystem

that has atimes set.

"""

**** CubicPower OpenStack Study ****

    def configure(self):

        """

        Configure the driver to use the stored configuration options

        Any store that needs special configuration should implement

        this method. If the store was not able to successfully configure

        itself, it should raise `exception.BadDriverConfiguration`

        """

        # Here we set up the various file-based image cache paths

        # that we need in order to find the files in different states

        # of cache management.

        self.set_paths()

        # We do a quick attempt to write a user xattr to a temporary file

        # to check that the filesystem is even enabled to support xattrs

        image_cache_dir = self.base_dir

        fake_image_filepath = os.path.join(image_cache_dir, 'checkme')

        with open(fake_image_filepath, 'wb') as fake_file:

            fake_file.write("XXX")

            fake_file.flush()

        try:

            set_xattr(fake_image_filepath, 'hits', '1')

        except IOError as e:

            if e.errno == errno.EOPNOTSUPP:

                msg = (_("The device housing the image cache directory "

                         "%(image_cache_dir)s does not support xattr. It is "

                         "likely you need to edit your fstab and add the "

                         "user_xattr option to the appropriate line for the "

                         "device housing the cache directory.") %

                       {'image_cache_dir': image_cache_dir})

                LOG.error(msg)

                raise exception.BadDriverConfiguration(driver_name="xattr",

                                                       reason=msg)

        else:

            # Cleanup after ourselves...

            if os.path.exists(fake_image_filepath):

                os.unlink(fake_image_filepath)

**** CubicPower OpenStack Study ****

    def get_cache_size(self):

        """

        Returns the total size in bytes of the image cache.

        """

        sizes = []

        for path in get_all_regular_files(self.base_dir):

            file_info = os.stat(path)

            sizes.append(file_info[stat.ST_SIZE])

        return sum(sizes)

**** CubicPower OpenStack Study ****

    def get_hit_count(self, image_id):

        """

        Return the number of hits that an image has.

        :param image_id: Opaque image identifier

        """

        if not self.is_cached(image_id):

            return 0

        path = self.get_image_filepath(image_id)

        return int(get_xattr(path, 'hits', default=0))

**** CubicPower OpenStack Study ****

    def get_cached_images(self):

        """

        Returns a list of records about cached images.

        """

        LOG.debug(_("Gathering cached image entries."))

        entries = []

        for path in get_all_regular_files(self.base_dir):

            image_id = os.path.basename(path)

            entry = {}

            entry['image_id'] = image_id

            file_info = os.stat(path)

            entry['last_modified'] = file_info[stat.ST_MTIME]

            entry['last_accessed'] = file_info[stat.ST_ATIME]

            entry['size'] = file_info[stat.ST_SIZE]

            entry['hits'] = self.get_hit_count(image_id)

            entries.append(entry)

        entries.sort()  # Order by ID

        return entries

**** CubicPower OpenStack Study ****

    def is_cached(self, image_id):

        """

        Returns True if the image with the supplied ID has its image

        file cached.

        :param image_id: Image ID

        """

        return os.path.exists(self.get_image_filepath(image_id))

**** CubicPower OpenStack Study ****

    def is_cacheable(self, image_id):

        """

        Returns True if the image with the supplied ID can have its

        image file cached, False otherwise.

        :param image_id: Image ID

        """

        # Make sure we're not already cached or caching the image

        return not (self.is_cached(image_id) or

                    self.is_being_cached(image_id))

**** CubicPower OpenStack Study ****

    def is_being_cached(self, image_id):

        """

        Returns True if the image with supplied id is currently

        in the process of having its image file cached.

        :param image_id: Image ID

        """

        path = self.get_image_filepath(image_id, 'incomplete')

        return os.path.exists(path)

**** CubicPower OpenStack Study ****

    def is_queued(self, image_id):

        """

        Returns True if the image identifier is in our cache queue.

        """

        path = self.get_image_filepath(image_id, 'queue')

        return os.path.exists(path)

**** CubicPower OpenStack Study ****

    def delete_all_cached_images(self):

        """

        Removes all cached image files and any attributes about the images

        """

        deleted = 0

        for path in get_all_regular_files(self.base_dir):

            delete_cached_file(path)

            deleted += 1

        return deleted

**** CubicPower OpenStack Study ****

    def delete_cached_image(self, image_id):

        """

        Removes a specific cached image file and any attributes about the image

        :param image_id: Image ID

        """

        path = self.get_image_filepath(image_id)

        delete_cached_file(path)

**** CubicPower OpenStack Study ****

    def delete_all_queued_images(self):

        """

        Removes all queued image files and any attributes about the images

        """

        files = [f for f in get_all_regular_files(self.queue_dir)]

        for file in files:

            os.unlink(file)

        return len(files)

**** CubicPower OpenStack Study ****

    def delete_queued_image(self, image_id):

        """

        Removes a specific queued image file and any attributes about the image

        :param image_id: Image ID

        """

        path = self.get_image_filepath(image_id, 'queue')

        if os.path.exists(path):

            os.unlink(path)

**** CubicPower OpenStack Study ****

    def get_least_recently_accessed(self):

        """

        Return a tuple containing the image_id and size of the least recently

        accessed cached file, or None if no cached files.

        """

        stats = []

        for path in get_all_regular_files(self.base_dir):

            file_info = os.stat(path)

            stats.append((file_info[stat.ST_ATIME],  # access time

                          file_info[stat.ST_SIZE],   # size in bytes

                          path))                     # absolute path

        if not stats:

            return None

        stats.sort()

        return os.path.basename(stats[0][2]), stats[0][1]

    @contextmanager

**** CubicPower OpenStack Study ****

    def open_for_write(self, image_id):

        """

        Open a file for writing the image file for an image

        with supplied identifier.

        :param image_id: Image ID

        """

        incomplete_path = self.get_image_filepath(image_id, 'incomplete')

        def set_attr(key, value):

            set_xattr(incomplete_path, key, value)

        def commit():

            set_attr('hits', 0)

            final_path = self.get_image_filepath(image_id)

            LOG.debug(_("Fetch finished, moving "

                        "'%(incomplete_path)s' to '%(final_path)s'"),

                      dict(incomplete_path=incomplete_path,

                           final_path=final_path))

            os.rename(incomplete_path, final_path)

            # Make sure that we "pop" the image from the queue...

            if self.is_queued(image_id):

                LOG.debug(_("Removing image '%s' from queue after "

                            "caching it."), image_id)

                os.unlink(self.get_image_filepath(image_id, 'queue'))

        def rollback(e):

            set_attr('error', six.text_type(e))

            invalid_path = self.get_image_filepath(image_id, 'invalid')

            LOG.debug(_("Fetch of cache file failed (%(e)s), rolling back by "

                        "moving '%(incomplete_path)s' to "

                        "'%(invalid_path)s'"),

                      {'e': six.text_type(e),

                       'incomplete_path': incomplete_path,

                       'invalid_path': invalid_path})

            os.rename(incomplete_path, invalid_path)

        try:

            with open(incomplete_path, 'wb') as cache_file:

                yield cache_file

        except Exception as e:

            rollback(e)

            raise

        else:

            commit()

        finally:

            # if the generator filling the cache file neither raises an

            # exception, nor completes fetching all data, neither rollback

            # nor commit will have been called, so the incomplete file

            # will persist - in that case remove it as it is unusable

            # example: ^c from client fetch

            if os.path.exists(incomplete_path):

                rollback('incomplete fetch')

    @contextmanager

**** CubicPower OpenStack Study ****

        def set_attr(key, value):

            set_xattr(incomplete_path, key, value)

**** CubicPower OpenStack Study ****

        def commit():

            set_attr('hits', 0)

            final_path = self.get_image_filepath(image_id)

            LOG.debug(_("Fetch finished, moving "

                        "'%(incomplete_path)s' to '%(final_path)s'"),

                      dict(incomplete_path=incomplete_path,

                           final_path=final_path))

            os.rename(incomplete_path, final_path)

            # Make sure that we "pop" the image from the queue...

            if self.is_queued(image_id):

                LOG.debug(_("Removing image '%s' from queue after "

                            "caching it."), image_id)

                os.unlink(self.get_image_filepath(image_id, 'queue'))

**** CubicPower OpenStack Study ****

        def rollback(e):

            set_attr('error', six.text_type(e))

            invalid_path = self.get_image_filepath(image_id, 'invalid')

            LOG.debug(_("Fetch of cache file failed (%(e)s), rolling back by "

                        "moving '%(incomplete_path)s' to "

                        "'%(invalid_path)s'"),

                      {'e': six.text_type(e),

                       'incomplete_path': incomplete_path,

                       'invalid_path': invalid_path})

            os.rename(incomplete_path, invalid_path)

        try:

            with open(incomplete_path, 'wb') as cache_file:

                yield cache_file

        except Exception as e:

            rollback(e)

            raise

        else:

            commit()

        finally:

            # if the generator filling the cache file neither raises an

            # exception, nor completes fetching all data, neither rollback

            # nor commit will have been called, so the incomplete file

            # will persist - in that case remove it as it is unusable

            # example: ^c from client fetch

            if os.path.exists(incomplete_path):

                rollback('incomplete fetch')

    @contextmanager

**** CubicPower OpenStack Study ****

    def open_for_read(self, image_id):

        """

        Open and yield file for reading the image file for an image

        with supplied identifier.

        :param image_id: Image ID

        """

        path = self.get_image_filepath(image_id)

        with open(path, 'rb') as cache_file:

            yield cache_file

        path = self.get_image_filepath(image_id)

        inc_xattr(path, 'hits', 1)

**** CubicPower OpenStack Study ****

    def queue_image(self, image_id):

        """

        This adds a image to be cache to the queue.

        If the image already exists in the queue or has already been

        cached, we return False, True otherwise

        :param image_id: Image ID

        """

        if self.is_cached(image_id):

            msg = _("Not queueing image '%s'. Already cached.") % image_id

            LOG.warn(msg)

            return False

        if self.is_being_cached(image_id):

            msg = _("Not queueing image '%s'. Already being "

                    "written to cache") % image_id

            LOG.warn(msg)

            return False

        if self.is_queued(image_id):

            msg = _("Not queueing image '%s'. Already queued.") % image_id

            LOG.warn(msg)

            return False

        path = self.get_image_filepath(image_id, 'queue')

        LOG.debug(_("Queueing image '%s'."), image_id)

        # Touch the file to add it to the queue

        with open(path, "w"):

            pass

        return True

**** CubicPower OpenStack Study ****

    def get_queued_images(self):

        """

        Returns a list of image IDs that are in the queue. The

        list should be sorted by the time the image ID was inserted

        into the queue.

        """

        files = [f for f in get_all_regular_files(self.queue_dir)]

        items = []

        for path in files:

            mtime = os.path.getmtime(path)

            items.append((mtime, os.path.basename(path)))

        items.sort()

        return [image_id for (mtime, image_id) in items]

**** CubicPower OpenStack Study ****

    def _reap_old_files(self, dirpath, entry_type, grace=None):

        now = time.time()

        reaped = 0

        for path in get_all_regular_files(dirpath):

            mtime = os.path.getmtime(path)

            age = now - mtime

            if not grace:

                LOG.debug(_("No grace period, reaping '%(path)s'"

                            " immediately"), {'path': path})

                delete_cached_file(path)

                reaped += 1

            elif age > grace:

                LOG.debug(_("Cache entry '%(path)s' exceeds grace period, "

                            "(%(age)i s > %(grace)i s)"),

                          {'path': path, 'age': age, 'grace': grace})

                delete_cached_file(path)

                reaped += 1

        LOG.info(_("Reaped %(reaped)s %(entry_type)s cache entries"),

                 {'reaped': reaped, 'entry_type': entry_type})

        return reaped

**** CubicPower OpenStack Study ****

    def reap_invalid(self, grace=None):

        """Remove any invalid cache entries

        :param grace: Number of seconds to keep an invalid entry around for

                      debugging purposes. If None, then delete immediately.

        """

        return self._reap_old_files(self.invalid_dir, 'invalid', grace=grace)

**** CubicPower OpenStack Study ****

    def reap_stalled(self, grace=None):

        """Remove any stalled cache entries

        :param grace: Number of seconds to keep an invalid entry around for

                      debugging purposes. If None, then delete immediately.

        """

        return self._reap_old_files(self.incomplete_dir, 'stalled',

                                    grace=grace)

**** CubicPower OpenStack Study ****

    def clean(self, stall_time=None):

        """

        Delete any image files in the invalid directory and any

        files in the incomplete directory that are older than a

        configurable amount of time.

        """

        self.reap_invalid()

        if stall_time is None:

            stall_time = CONF.image_cache_stall_time

        self.reap_stalled(stall_time)

def get_all_regular_files(basepath):

    for fname in os.listdir(basepath):

        path = os.path.join(basepath, fname)

        if os.path.isfile(path):

            yield path

def delete_cached_file(path):

    if os.path.exists(path):

        LOG.debug(_("Deleting image cache file '%s'"), path)

        os.unlink(path)

    else:

        LOG.warn(_("Cached image file '%s' doesn't exist, unable to"

                   " delete"), path)

def _make_namespaced_xattr_key(key, namespace='user'):

    """

    Create a fully-qualified xattr-key by including the intended namespace.

    Namespacing differs among OSes[1]:

        FreeBSD: user, system

        Linux: user, system, trusted, security

        MacOS X: not needed

    Mac OS X won't break if we include a namespace qualifier, so, for

    simplicity, we always include it.

    --

    [1] http://en.wikipedia.org/wiki/Extended_file_attributes

    """

    namespaced_key = ".".join([namespace, key])

    return namespaced_key

def get_xattr(path, key, **kwargs):

    """Return the value for a particular xattr

    If the key doesn't not exist, or xattrs aren't supported by the file

    system then a KeyError will be raised, that is, unless you specify a

    default using kwargs.

    """

    namespaced_key = _make_namespaced_xattr_key(key)

    try:

        return xattr.getxattr(path, namespaced_key)

    except IOError:

        if 'default' in kwargs:

            return kwargs['default']

        else:

            raise

def set_xattr(path, key, value):

    """Set the value of a specified xattr.

    If xattrs aren't supported by the file-system, we skip setting the value.

    """

    namespaced_key = _make_namespaced_xattr_key(key)

    xattr.setxattr(path, namespaced_key, str(value))

def inc_xattr(path, key, n=1):

    """

    Increment the value of an xattr (assuming it is an integer).

    BEWARE, this code *does* have a RACE CONDITION, since the

    read/update/write sequence is not atomic.

    Since the use-case for this function is collecting stats--not critical--

    the benefits of simple, lock-free code out-weighs the possibility of an

    occasional hit not being counted.

    """

    count = int(get_xattr(path, key))

    count += n

    set_xattr(path, key, str(count))

**** CubicPower OpenStack Study ****

def get_all_regular_files(basepath):

    for fname in os.listdir(basepath):

        path = os.path.join(basepath, fname)

        if os.path.isfile(path):

            yield path

**** CubicPower OpenStack Study ****

def delete_cached_file(path):

    if os.path.exists(path):

        LOG.debug(_("Deleting image cache file '%s'"), path)

        os.unlink(path)

    else:

        LOG.warn(_("Cached image file '%s' doesn't exist, unable to"

                   " delete"), path)

**** CubicPower OpenStack Study ****

def _make_namespaced_xattr_key(key, namespace='user'):

    """

    Create a fully-qualified xattr-key by including the intended namespace.

    Namespacing differs among OSes[1]:

        FreeBSD: user, system

        Linux: user, system, trusted, security

        MacOS X: not needed

    Mac OS X won't break if we include a namespace qualifier, so, for

    simplicity, we always include it.

    --

    [1] http://en.wikipedia.org/wiki/Extended_file_attributes

    """

    namespaced_key = ".".join([namespace, key])

    return namespaced_key

**** CubicPower OpenStack Study ****

def get_xattr(path, key, **kwargs):

    """Return the value for a particular xattr

    If the key doesn't not exist, or xattrs aren't supported by the file

    system then a KeyError will be raised, that is, unless you specify a

    default using kwargs.

    """

    namespaced_key = _make_namespaced_xattr_key(key)

    try:

        return xattr.getxattr(path, namespaced_key)

    except IOError:

        if 'default' in kwargs:

            return kwargs['default']

        else:

            raise

**** CubicPower OpenStack Study ****

def set_xattr(path, key, value):

    """Set the value of a specified xattr.

    If xattrs aren't supported by the file-system, we skip setting the value.

    """

    namespaced_key = _make_namespaced_xattr_key(key)

    xattr.setxattr(path, namespaced_key, str(value))

**** CubicPower OpenStack Study ****

def inc_xattr(path, key, n=1):

    """

    Increment the value of an xattr (assuming it is an integer).

    BEWARE, this code *does* have a RACE CONDITION, since the

    read/update/write sequence is not atomic.

    Since the use-case for this function is collecting stats--not critical--

    the benefits of simple, lock-free code out-weighs the possibility of an

    occasional hit not being counted.

    """

    count = int(get_xattr(path, key))

    count += n

    set_xattr(path, key, str(count))