# vim: tabstop=4 shiftwidth=4 softtabstop=4 # Copyright 2012 Michael Still and Canonical 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. """Image cache manager. The cache manager implements the specification at http://wiki.openstack.org/nova-image-cache-management. """ import hashlib import json import os import re import time from oslo.config import cfg from nova.compute import task_states from nova.compute import vm_states from nova.openstack.common import fileutils from nova.openstack.common import jsonutils from nova.openstack.common import log as logging from nova import utils from nova.virt.libvirt import utils as virtutils LOG = logging.getLogger(__name__) imagecache_opts = [ cfg.StrOpt('base_dir_name', default='_base', help="Where cached images are stored under $instances_path." "This is NOT the full path - just a folder name." "For per-compute-host cached images, set to _base_$my_ip"), cfg.StrOpt('image_info_filename_pattern', default='$instances_path/$base_dir_name/%(image)s.info', help='Allows image information files to be stored in ' 'non-standard locations'), cfg.BoolOpt('remove_unused_base_images', default=True, help='Should unused base images be removed?'), cfg.BoolOpt('remove_unused_kernels', default=False, help='Should unused kernel images be removed? This is only ' 'safe to enable if all compute nodes have been updated ' 'to support this option. This will enabled by default ' 'in future.'), cfg.IntOpt('remove_unused_resized_minimum_age_seconds', default=3600, help='Unused resized base images younger than this will not be ' 'removed'), cfg.IntOpt('remove_unused_original_minimum_age_seconds', default=(24 * 3600), help='Unused unresized base images younger than this will not ' 'be removed'), cfg.BoolOpt('checksum_base_images', default=False, help='Write a checksum for files in _base to disk'), cfg.IntOpt('checksum_interval_seconds', default=3600, help='How frequently to checksum base images'), ] CONF = cfg.CONF CONF.register_opts(imagecache_opts) CONF.import_opt('host', 'nova.netconf') CONF.import_opt('instances_path', 'nova.compute.manager') def get_cache_fname(images, key): """Return a filename based on the SHA1 hash of a given image ID. Image files stored in the _base directory that match this pattern are considered for cleanup by the image cache manager. The cache manager considers the file to be in use if it matches an instance's image_ref, kernel_id or ramdisk_id property. However, in grizzly-3 and before, only the image_ref property was considered. This means that it's unsafe to store kernel and ramdisk images using this pattern until we're sure that all compute nodes are running a cache manager newer than grizzly-3. For now, we require admins to confirm that by setting the remove_unused_kernels boolean but, at some point in the future, we'll be safely able to assume this. """ image_id = str(images[key]) if not CONF.remove_unused_kernels and key in ['kernel_id', 'ramdisk_id']: return image_id else: return hashlib.sha1(image_id).hexdigest() def get_info_filename(base_path): """Construct a filename for storing additional information about a base image. Returns a filename. """ base_file = os.path.basename(base_path) return (CONF.image_info_filename_pattern % {'image': base_file}) def is_valid_info_file(path): """Test if a given path matches the pattern for info files.""" digest_size = hashlib.sha1().digestsize * 2 regexp = (CONF.image_info_filename_pattern % {'image': ('([0-9a-f]{%(digest_size)d}|' '[0-9a-f]{%(digest_size)d}_sm|' '[0-9a-f]{%(digest_size)d}_[0-9]+)' % {'digest_size': digest_size})}) m = re.match(regexp, path) if m: return True return False def _read_possible_json(serialized, info_file): try: d = jsonutils.loads(serialized) except ValueError as e: LOG.error(_('Error reading image info file %(filename)s: ' '%(error)s'), {'filename': info_file, 'error': e}) d = {} return d def read_stored_info(target, field=None, timestamped=False): """Read information about an image. Returns an empty dictionary if there is no info, just the field value if a field is requested, or the entire dictionary otherwise. """ info_file = get_info_filename(target) if not os.path.exists(info_file): # NOTE(mikal): Special case to handle essex checksums being converted. # There is an assumption here that target is a base image filename. old_filename = target + '.sha1' if field == 'sha1' and os.path.exists(old_filename): hash_file = open(old_filename) hash_value = hash_file.read() hash_file.close() write_stored_info(target, field=field, value=hash_value) os.remove(old_filename) d = {field: hash_value} else: d = {} else: lock_name = 'info-%s' % os.path.split(target)[-1] lock_path = os.path.join(CONF.instances_path, 'locks') @utils.synchronized(lock_name, external=True, lock_path=lock_path) def read_file(info_file): LOG.debug(_('Reading image info file: %s'), info_file) with open(info_file, 'r') as f: return f.read().rstrip() serialized = read_file(info_file) d = _read_possible_json(serialized, info_file) if field: if timestamped: return (d.get(field, None), d.get('%s-timestamp' % field, None)) else: return d.get(field, None) return d def write_stored_info(target, field=None, value=None): """Write information about an image.""" if not field: return info_file = get_info_filename(target) LOG.info(_('Writing stored info to %s'), info_file) fileutils.ensure_tree(os.path.dirname(info_file)) lock_name = 'info-%s' % os.path.split(target)[-1] lock_path = os.path.join(CONF.instances_path, 'locks') @utils.synchronized(lock_name, external=True, lock_path=lock_path) def write_file(info_file, field, value): d = {} if os.path.exists(info_file): with open(info_file, 'r') as f: d = _read_possible_json(f.read(), info_file) d[field] = value d['%s-timestamp' % field] = time.time() with open(info_file, 'w') as f: f.write(json.dumps(d)) write_file(info_file, field, value) def read_stored_checksum(target, timestamped=True): """Read the checksum. Returns the checksum (as hex) or None. """ return read_stored_info(target, field='sha1', timestamped=timestamped) def write_stored_checksum(target): """Write a checksum to disk for a file in _base.""" with open(target, 'r') as img_file: checksum = utils.hash_file(img_file) write_stored_info(target, field='sha1', value=checksum) class ImageCacheManager(object): def __init__(self): self.lock_path = os.path.join(CONF.instances_path, 'locks') self._reset_state() def _reset_state(self): """Reset state variables used for each pass.""" self.used_images = {} self.image_popularity = {} self.instance_names = set() self.active_base_files = [] self.corrupt_base_files = [] self.originals = [] self.removable_base_files = [] self.unexplained_images = [] def _store_image(self, base_dir, ent, original=False): """Store a base image for later examination.""" entpath = os.path.join(base_dir, ent) if os.path.isfile(entpath): self.unexplained_images.append(entpath) if original: self.originals.append(entpath) def _list_base_images(self, base_dir): """Return a list of the images present in _base. Determine what images we have on disk. There will be other files in this directory so we only grab the ones which are the right length to be disk images. Note that this does not return a value. It instead populates a class variable with a list of images that we need to try and explain. """ digest_size = hashlib.sha1().digestsize * 2 for ent in os.listdir(base_dir): if len(ent) == digest_size: self._store_image(base_dir, ent, original=True) elif (len(ent) > digest_size + 2 and ent[digest_size] == '_' and not is_valid_info_file(os.path.join(base_dir, ent))): self._store_image(base_dir, ent, original=False) def _list_running_instances(self, context, all_instances): """List running instances (on all compute nodes).""" self.used_images = {} self.image_popularity = {} self.instance_names = set() for instance in all_instances: # NOTE(mikal): "instance name" here means "the name of a directory # which might contain an instance" and therefore needs to include # historical permutations as well as the current one. self.instance_names.add(instance['name']) self.instance_names.add(instance['uuid']) resize_states = [task_states.RESIZE_PREP, task_states.RESIZE_MIGRATING, task_states.RESIZE_MIGRATED, task_states.RESIZE_FINISH] if instance['task_state'] in resize_states or \ instance['vm_state'] == vm_states.RESIZED: self.instance_names.add(instance['name'] + '_resize') self.instance_names.add(instance['uuid'] + '_resize') for image_key in ['image_ref', 'kernel_id', 'ramdisk_id']: try: image_ref_str = str(instance[image_key]) except KeyError: continue local, remote, insts = self.used_images.get(image_ref_str, (0, 0, [])) if instance['host'] == CONF.host: local += 1 else: remote += 1 insts.append(instance['name']) self.used_images[image_ref_str] = (local, remote, insts) self.image_popularity.setdefault(image_ref_str, 0) self.image_popularity[image_ref_str] += 1 def _list_backing_images(self): """List the backing images currently in use.""" inuse_images = [] for ent in os.listdir(CONF.instances_path): if ent in self.instance_names: LOG.debug(_('%s is a valid instance name'), ent) disk_path = os.path.join(CONF.instances_path, ent, 'disk') if os.path.exists(disk_path): LOG.debug(_('%s has a disk file'), ent) backing_file = virtutils.get_disk_backing_file(disk_path) LOG.debug(_('Instance %(instance)s is backed by ' '%(backing)s'), {'instance': ent, 'backing': backing_file}) if backing_file: backing_path = os.path.join(CONF.instances_path, CONF.base_dir_name, backing_file) if backing_path not in inuse_images: inuse_images.append(backing_path) if backing_path in self.unexplained_images: LOG.warning(_('Instance %(instance)s is using a ' 'backing file %(backing)s which ' 'does not appear in the image ' 'service'), {'instance': ent, 'backing': backing_file}) self.unexplained_images.remove(backing_path) return inuse_images def _find_base_file(self, base_dir, fingerprint): """Find the base file matching this fingerprint. Yields the name of the base file, a boolean which is True if the image is "small", and a boolean which indicates if this is a resized image. Note that is is possible for more than one yield to result from this check. If no base file is found, then nothing is yielded. """ # The original file from glance base_file = os.path.join(base_dir, fingerprint) if os.path.exists(base_file): yield base_file, False, False # An older naming style which can be removed sometime after Folsom base_file = os.path.join(base_dir, fingerprint + '_sm') if os.path.exists(base_file): yield base_file, True, False # Resized images resize_re = re.compile('.*/%s_[0-9]+$' % fingerprint) for img in self.unexplained_images: m = resize_re.match(img) if m: yield img, False, True def _verify_checksum(self, img_id, base_file, create_if_missing=True): """Compare the checksum stored on disk with the current file. Note that if the checksum fails to verify this is logged, but no actual action occurs. This is something sysadmins should monitor for and handle manually when it occurs. """ if not CONF.checksum_base_images: return None lock_name = 'hash-%s' % os.path.split(base_file)[-1] # Protect against other nova-computes performing checksums at the same # time if we are using shared storage @utils.synchronized(lock_name, external=True, lock_path=self.lock_path) def inner_verify_checksum(): (stored_checksum, stored_timestamp) = read_stored_checksum( base_file, timestamped=True) if stored_checksum: # NOTE(mikal): Checksums are timestamped. If we have recently # checksummed (possibly on another compute node if we are using # shared storage), then we don't need to checksum again. if (stored_timestamp and time.time() - stored_timestamp < CONF.checksum_interval_seconds): return True # NOTE(mikal): If there is no timestamp, then the checksum was # performed by a previous version of the code. if not stored_timestamp: write_stored_info(base_file, field='sha1', value=stored_checksum) with open(base_file, 'r') as f: current_checksum = utils.hash_file(f) if current_checksum != stored_checksum: LOG.error(_('image %(id)s at (%(base_file)s): image ' 'verification failed'), {'id': img_id, 'base_file': base_file}) return False else: return True else: LOG.info(_('image %(id)s at (%(base_file)s): image ' 'verification skipped, no hash stored'), {'id': img_id, 'base_file': base_file}) # NOTE(mikal): If the checksum file is missing, then we should # create one. We don't create checksums when we download images # from glance because that would delay VM startup. if CONF.checksum_base_images and create_if_missing: LOG.info(_('%(id)s (%(base_file)s): generating checksum'), {'id': img_id, 'base_file': base_file}) write_stored_checksum(base_file) return None return inner_verify_checksum() def _remove_base_file(self, base_file): """Remove a single base file if it is old enough. Returns nothing. """ if not os.path.exists(base_file): LOG.debug(_('Cannot remove %(base_file)s, it does not exist'), base_file) return mtime = os.path.getmtime(base_file) age = time.time() - mtime maxage = CONF.remove_unused_resized_minimum_age_seconds if base_file in self.originals: maxage = CONF.remove_unused_original_minimum_age_seconds if age < maxage: LOG.info(_('Base file too young to remove: %s'), base_file) else: LOG.info(_('Removing base file: %s'), base_file) try: os.remove(base_file) signature = get_info_filename(base_file) if os.path.exists(signature): os.remove(signature) except OSError as e: LOG.error(_('Failed to remove %(base_file)s, ' 'error was %(error)s'), {'base_file': base_file, 'error': e}) def _handle_base_image(self, img_id, base_file): """Handle the checks for a single base image.""" image_bad = False image_in_use = False LOG.info(_('image %(id)s at (%(base_file)s): checking'), {'id': img_id, 'base_file': base_file}) if base_file in self.unexplained_images: self.unexplained_images.remove(base_file) if (base_file and os.path.exists(base_file) and os.path.isfile(base_file)): # _verify_checksum returns True if the checksum is ok, and None if # there is no checksum file checksum_result = self._verify_checksum(img_id, base_file) if checksum_result is not None: image_bad = not checksum_result # Give other threads a chance to run time.sleep(0) instances = [] if img_id in self.used_images: local, remote, instances = self.used_images[img_id] if local > 0 or remote > 0: image_in_use = True LOG.info(_('image %(id)s at (%(base_file)s): ' 'in use: on this node %(local)d local, ' '%(remote)d on other nodes sharing this instance ' 'storage'), {'id': img_id, 'base_file': base_file, 'local': local, 'remote': remote}) self.active_base_files.append(base_file) if not base_file: LOG.warning(_('image %(id)s at (%(base_file)s): warning ' '-- an absent base file is in use! ' 'instances: %(instance_list)s'), {'id': img_id, 'base_file': base_file, 'instance_list': ' '.join(instances)}) if image_bad: self.corrupt_base_files.append(base_file) if base_file: if not image_in_use: LOG.debug(_('image %(id)s at (%(base_file)s): image is not in ' 'use'), {'id': img_id, 'base_file': base_file}) self.removable_base_files.append(base_file) else: LOG.debug(_('image %(id)s at (%(base_file)s): image is in ' 'use'), {'id': img_id, 'base_file': base_file}) if os.path.exists(base_file): virtutils.chown(base_file, os.getuid()) os.utime(base_file, None) def verify_base_images(self, context, all_instances): """Verify that base images are in a reasonable state.""" # NOTE(mikal): The new scheme for base images is as follows -- an # image is streamed from the image service to _base (filename is the # sha1 hash of the image id). If CoW is enabled, that file is then # resized to be the correct size for the instance (filename is the # same as the original, but with an underscore and the resized size # in bytes). This second file is then CoW'd to the instance disk. If # CoW is disabled, the resize occurs as part of the copy from the # cache to the instance directory. Files ending in _sm are no longer # created, but may remain from previous versions. self._reset_state() base_dir = os.path.join(CONF.instances_path, CONF.base_dir_name) if not os.path.exists(base_dir): LOG.debug(_('Skipping verification, no base directory at %s'), base_dir) return LOG.debug(_('Verify base images')) self._list_base_images(base_dir) self._list_running_instances(context, all_instances) # Determine what images are on disk because they're in use for img in self.used_images: fingerprint = hashlib.sha1(img).hexdigest() LOG.debug(_('Image id %(id)s yields fingerprint %(fingerprint)s'), {'id': img, 'fingerprint': fingerprint}) for result in self._find_base_file(base_dir, fingerprint): base_file, image_small, image_resized = result self._handle_base_image(img, base_file) if not image_small and not image_resized: self.originals.append(base_file) # Elements remaining in unexplained_images might be in use inuse_backing_images = self._list_backing_images() for backing_path in inuse_backing_images: if backing_path not in self.active_base_files: self.active_base_files.append(backing_path) # Anything left is an unknown base image for img in self.unexplained_images: LOG.warning(_('Unknown base file: %s'), img) self.removable_base_files.append(img) # Dump these lists if self.active_base_files: LOG.info(_('Active base files: %s'), ' '.join(self.active_base_files)) if self.corrupt_base_files: LOG.info(_('Corrupt base files: %s'), ' '.join(self.corrupt_base_files)) if self.removable_base_files: LOG.info(_('Removable base files: %s'), ' '.join(self.removable_base_files)) if CONF.remove_unused_base_images: for base_file in self.removable_base_files: self._remove_base_file(base_file) # That's it LOG.debug(_('Verification complete'))