glance/glance/api/middleware/cache.py

344 lines
14 KiB
Python

# 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.
"""
Transparent image file caching middleware, designed to live on
Glance API nodes. When images are requested from the API node,
this middleware caches the returned image file to local filesystem.
When subsequent requests for the same image file are received,
the local cached copy of the image file is returned.
"""
import re
import six
from oslo_log import log as logging
from six.moves import http_client as http
import webob
from glance.api.common import size_checked_iter
from glance.api import policy
from glance.api.v1 import images
from glance.common import exception
from glance.common import utils
from glance.common import wsgi
import glance.db
from glance.i18n import _LE, _LI
from glance import image_cache
from glance import notifier
import glance.registry.client.v1.api as registry
LOG = logging.getLogger(__name__)
PATTERNS = {
('v1', 'GET'): re.compile(r'^/v1/images/([^\/]+)$'),
('v1', 'DELETE'): re.compile(r'^/v1/images/([^\/]+)$'),
('v2', 'GET'): re.compile(r'^/v2/images/([^\/]+)/file$'),
('v2', 'DELETE'): re.compile(r'^/v2/images/([^\/]+)$')
}
class CacheFilter(wsgi.Middleware):
def __init__(self, app):
self.cache = image_cache.ImageCache()
self.serializer = images.ImageSerializer()
self.policy = policy.Enforcer()
LOG.info(_LI("Initialized image cache middleware"))
super(CacheFilter, self).__init__(app)
def _verify_metadata(self, image_meta):
"""
Sanity check the 'deleted' and 'size' metadata values.
"""
# NOTE: admins can see image metadata in the v1 API, but shouldn't
# be able to download the actual image data.
if image_meta['status'] == 'deleted' and image_meta['deleted']:
raise exception.NotFound()
if not image_meta['size']:
# override image size metadata with the actual cached
# file size, see LP Bug #900959
if not isinstance(image_meta, policy.ImageTarget):
image_meta['size'] = self.cache.get_image_size(
image_meta['id'])
else:
image_meta.target.size = self.cache.get_image_size(
image_meta['id'])
@staticmethod
def _match_request(request):
"""Determine the version of the url and extract the image id
:returns: tuple of version and image id if the url is a cacheable,
otherwise None
"""
for ((version, method), pattern) in PATTERNS.items():
if request.method != method:
continue
match = pattern.match(request.path_info)
if match is None:
continue
image_id = match.group(1)
# Ensure the image id we got looks like an image id to filter
# out a URI like /images/detail. See LP Bug #879136
if image_id != 'detail':
return (version, method, image_id)
def _enforce(self, req, action, target=None):
"""Authorize an action against our policies"""
if target is None:
target = {}
try:
self.policy.enforce(req.context, action, target)
except exception.Forbidden as e:
LOG.debug("User not permitted to perform '%s' action", action)
raise webob.exc.HTTPForbidden(explanation=e.msg, request=req)
def _get_v1_image_metadata(self, request, image_id):
"""
Retrieves image metadata using registry for v1 api and creates
dictionary-like mash-up of image core and custom properties.
"""
try:
image_metadata = registry.get_image_metadata(request.context,
image_id)
return utils.create_mashup_dict(image_metadata)
except exception.NotFound as e:
LOG.debug("No metadata found for image '%s'", image_id)
raise webob.exc.HTTPNotFound(explanation=e.msg, request=request)
def _get_v2_image_metadata(self, request, image_id):
"""
Retrieves image and for v2 api and creates adapter like object
to access image core or custom properties on request.
"""
db_api = glance.db.get_api()
image_repo = glance.db.ImageRepo(request.context, db_api)
try:
image = image_repo.get(image_id)
# Storing image object in request as it is required in
# _process_v2_request call.
request.environ['api.cache.image'] = image
return policy.ImageTarget(image)
except exception.NotFound as e:
raise webob.exc.HTTPNotFound(explanation=e.msg, request=request)
def process_request(self, request):
"""
For requests for an image file, we check the local image
cache. If present, we return the image file, appending
the image metadata in headers. If not present, we pass
the request on to the next application in the pipeline.
"""
match = self._match_request(request)
try:
(version, method, image_id) = match
except TypeError:
# Trying to unpack None raises this exception
return None
self._stash_request_info(request, image_id, method, version)
# Partial image download requests shall not be served from cache
# Bug: 1664709
# TODO(dharinic): If an image is already cached, add support to serve
# only the requested bytes (partial image download) from the cache.
if (request.headers.get('Content-Range') or
request.headers.get('Range')):
return None
if request.method != 'GET' or not self.cache.is_cached(image_id):
return None
method = getattr(self, '_get_%s_image_metadata' % version)
image_metadata = method(request, image_id)
# Deactivated images shall not be served from cache
if image_metadata['status'] == 'deactivated':
return None
try:
self._enforce(request, 'download_image', target=image_metadata)
except exception.Forbidden:
return None
LOG.debug("Cache hit for image '%s'", image_id)
image_iterator = self.get_from_cache(image_id)
method = getattr(self, '_process_%s_request' % version)
try:
return method(request, image_id, image_iterator, image_metadata)
except exception.ImageNotFound:
msg = _LE("Image cache contained image file for image '%s', "
"however the registry did not contain metadata for "
"that image!") % image_id
LOG.error(msg)
self.cache.delete_cached_image(image_id)
@staticmethod
def _stash_request_info(request, image_id, method, version):
"""
Preserve the image id, version and request method for later retrieval
"""
request.environ['api.cache.image_id'] = image_id
request.environ['api.cache.method'] = method
request.environ['api.cache.version'] = version
@staticmethod
def _fetch_request_info(request):
"""
Preserve the cached image id, version for consumption by the
process_response method of this middleware
"""
try:
image_id = request.environ['api.cache.image_id']
method = request.environ['api.cache.method']
version = request.environ['api.cache.version']
except KeyError:
return None
else:
return (image_id, method, version)
def _process_v1_request(self, request, image_id, image_iterator,
image_meta):
# Don't display location
if 'location' in image_meta:
del image_meta['location']
image_meta.pop('location_data', None)
self._verify_metadata(image_meta)
response = webob.Response(request=request)
raw_response = {
'image_iterator': image_iterator,
'image_meta': image_meta,
}
return self.serializer.show(response, raw_response)
def _process_v2_request(self, request, image_id, image_iterator,
image_meta):
# We do some contortions to get the image_metadata so
# that we can provide it to 'size_checked_iter' which
# will generate a notification.
# TODO(mclaren): Make notification happen more
# naturally once caching is part of the domain model.
image = request.environ['api.cache.image']
self._verify_metadata(image_meta)
response = webob.Response(request=request)
response.app_iter = size_checked_iter(response, image_meta,
image_meta['size'],
image_iterator,
notifier.Notifier())
# NOTE (flwang): Set the content-type, content-md5 and content-length
# explicitly to be consistent with the non-cache scenario.
# Besides, it's not worth the candle to invoke the "download" method
# of ResponseSerializer under image_data. Because method "download"
# will reset the app_iter. Then we have to call method
# "size_checked_iter" to avoid missing any notification. But after
# call "size_checked_iter", we will lose the content-md5 and
# content-length got by the method "download" because of this issue:
# https://github.com/Pylons/webob/issues/86
response.headers['Content-Type'] = 'application/octet-stream'
if image.checksum:
response.headers['Content-MD5'] = (image.checksum.encode('utf-8')
if six.PY2 else image.checksum)
response.headers['Content-Length'] = str(image.size)
return response
def process_response(self, resp):
"""
We intercept the response coming back from the main
images Resource, removing image file from the cache
if necessary
"""
status_code = self.get_status_code(resp)
if not 200 <= status_code < 300:
return resp
# Note(dharinic): Bug: 1664709: Do not cache partial images.
if status_code == http.PARTIAL_CONTENT:
return resp
try:
(image_id, method, version) = self._fetch_request_info(
resp.request)
except TypeError:
return resp
if method == 'GET' and status_code == http.NO_CONTENT:
# Bugfix:1251055 - Don't cache non-existent image files.
# NOTE: Both GET for an image without locations and DELETE return
# 204 but DELETE should be processed.
return resp
method_str = '_process_%s_response' % method
try:
process_response_method = getattr(self, method_str)
except AttributeError:
LOG.error(_LE('could not find %s') % method_str)
# Nothing to do here, move along
return resp
else:
return process_response_method(resp, image_id, version=version)
def _process_DELETE_response(self, resp, image_id, version=None):
if self.cache.is_cached(image_id):
LOG.debug("Removing image %s from cache", image_id)
self.cache.delete_cached_image(image_id)
return resp
def _process_GET_response(self, resp, image_id, version=None):
image_checksum = resp.headers.get('Content-MD5')
if not image_checksum:
# API V1 stores the checksum in a different header:
image_checksum = resp.headers.get('x-image-meta-checksum')
if not image_checksum:
LOG.error(_LE("Checksum header is missing."))
# fetch image_meta on the basis of version
image_metadata = None
if version:
method = getattr(self, '_get_%s_image_metadata' % version)
image_metadata = method(resp.request, image_id)
# NOTE(zhiyan): image_cache return a generator object and set to
# response.app_iter, it will be called by eventlet.wsgi later.
# So we need enforce policy firstly but do it by application
# since eventlet.wsgi could not catch webob.exc.HTTPForbidden and
# return 403 error to client then.
self._enforce(resp.request, 'download_image', target=image_metadata)
resp.app_iter = self.cache.get_caching_iter(image_id, image_checksum,
resp.app_iter)
return resp
def get_status_code(self, response):
"""
Returns the integer status code from the response, which
can be either a Webob.Response (used in testing) or httplib.Response
"""
if hasattr(response, 'status_int'):
return response.status_int
return response.status
def get_from_cache(self, image_id):
"""Called if cache hit"""
with self.cache.open_for_read(image_id) as cache_file:
chunks = utils.chunkiter(cache_file)
for chunk in chunks:
yield chunk