def update_metadata_and_save(self, bundle, new_bundle):
"""
Updates the metadata about the contents of the bundle, including
data_size as well as the total amount of disk used by the user.
If |new_bundle| is True, saves the bundle as a new bundle. Otherwise,
updates it.
"""
bundle_path = self._bundle_store.get_bundle_location(bundle.uuid)
dirs_and_files = None
if os.path.isdir(bundle_path):
dirs_and_files = path_util.recursive_ls(bundle_path)
else:
dirs_and_files = [], [bundle_path]
data_hash = '0x%s' % (path_util.hash_directory(bundle_path, dirs_and_files))
data_size = path_util.get_size(bundle_path, dirs_and_files)
if new_bundle:
bundle.data_hash = data_hash
bundle.metadata.set_metadata_key('data_size', data_size)
self._bundle_model.save_bundle(bundle)
else:
bundle_update = {
'data_hash': data_hash,
'metadata': {
'data_size': data_size,
},
}
self._bundle_model.update_bundle(bundle, bundle_update)
self._bundle_model.update_user_disk_used(bundle.owner_id)
def update_metadata_and_save(self, bundle, enforce_disk_quota=False):
"""
Updates the metadata about the contents of the bundle, including
data_size as well as the total amount of disk used by the user.
If |new_bundle| is True, saves the bundle as a new bundle. Otherwise,
updates it.
"""
bundle_path = self._bundle_store.get_bundle_location(bundle.uuid)
dirs_and_files = None
if os.path.isdir(bundle_path):
dirs_and_files = path_util.recursive_ls(bundle_path)
else:
dirs_and_files = [], [bundle_path]
data_hash = '0x%s' % (path_util.hash_directory(bundle_path,
dirs_and_files))
data_size = path_util.get_size(bundle_path, dirs_and_files)
if enforce_disk_quota:
disk_left = self._bundle_model.get_user_disk_quota_left(
bundle.owner_id)
if data_size > disk_left:
raise UsageError(
"Can't save bundle, bundle size %s greater than user's disk quota left: %s"
% (data_size, disk_left))
bundle_update = {
'data_hash': data_hash,
'metadata': {
'data_size': data_size,
},
}
self._bundle_model.update_bundle(bundle, bundle_update)
self._bundle_model.update_user_disk_used(bundle.owner_id)
def update_metadata_and_save(self, bundle, new_bundle):
"""
Updates the metadata about the contents of the bundle, including
data_size as well as the total amount of disk used by the user.
If |new_bundle| is True, saves the bundle as a new bundle. Otherwise,
updates it.
"""
bundle_path = self._bundle_store.get_bundle_location(bundle.uuid)
dirs_and_files = None
if os.path.isdir(bundle_path):
dirs_and_files = path_util.recursive_ls(bundle_path)
else:
dirs_and_files = [], [bundle_path]
data_hash = '0x%s' % (path_util.hash_directory(bundle_path, dirs_and_files))
data_size = path_util.get_size(bundle_path, dirs_and_files)
if new_bundle:
bundle.data_hash = data_hash
bundle.metadata.set_metadata_key('data_size', data_size)
self._bundle_model.save_bundle(bundle)
else:
bundle_update = {
'data_hash': data_hash,
'metadata': {
'data_size': data_size,
},
}
self._bundle_model.update_bundle(bundle, bundle_update)
self._bundle_model.update_user_disk_used(bundle.owner_id)
def upload(self, path, allow_symlinks=False):
'''
Copy the contents of the directory at path into the data subdirectory,
in a subfolder named by a hash of the contents of the new data directory.
Return a (data_hash, metadata) pair, where the metadata is a dict mapping
keys to precomputed statistics about the new data directory.
'''
absolute_path = path_util.normalize(path)
path_util.check_isvalid(absolute_path, 'upload')
# Recursively copy the directory into a new BundleStore temp directory.
temp_directory = uuid.uuid4().hex
temp_path = os.path.join(self.temp, temp_directory)
path_util.copy(absolute_path, temp_path)
# Multiplex between uploading a directory and uploading a file here.
# All other path_util calls will use these lists of directories and files.
if os.path.isdir(temp_path):
dirs_and_files = path_util.recursive_ls(temp_path)
else:
dirs_and_files = ([], [temp_path])
if not allow_symlinks:
path_util.check_for_symlinks(temp_path, dirs_and_files)
path_util.set_permissions(temp_path, 0o755, dirs_and_files)
# Hash the contents of the temporary directory, and then if there is no
# data with this hash value, move this directory into the data directory.
data_hash = '0x%s' % (path_util.hash_directory(temp_path, dirs_and_files),)
data_size = path_util.get_size(temp_path, dirs_and_files)
final_path = os.path.join(self.data, data_hash)
final_path_exists = False
try:
os.utime(final_path, None)
final_path_exists = True
except OSError, e:
if e.errno == errno.ENOENT:
os.rename(temp_path, final_path)
else:
raise
def update_running_bundle(self, status):
'''
Update the database with information about the bundle given by |status|.
If the bundle is completed, then we need to install the bundle and clean up.
'''
status['last_updated'] = int(time.time())
# Update the bundle's data with status (which is the new information).
bundle = status['bundle']
# Update to the database
db_update = {}
# Update state
if 'state' in status and status['state']:
db_update['state'] = status['state']
# Add metadata from the machine
db_update['metadata'] = metadata = {}
bundle_subclass = type(bundle)
for spec in bundle_subclass.METADATA_SPECS:
value = status.get(spec.key)
if value is not None:
metadata[spec.key] = value
#print 'update_running_bundle', status
# See if the bundle is completed.
success = status.get('success')
if success is not None:
# Re-install dependencies.
# - For RunBundle, remove the dependencies.
# - For MakeBundle, copy. This way, we maintain the invariant that
# we always only need to look back one-level at the dependencies,
# not recurse.
try:
temp_dir = status.get('temp_dir')
if not temp_dir:
temp_dir = bundle.metadata.temp_dir
if isinstance(bundle, RunBundle):
print >>sys.stderr, 'Worker.finalize_bundle: removing dependencies from %s (RunBundle)' % temp_dir
bundle.remove_dependencies(self.bundle_store, self.get_parent_dict(bundle), temp_dir)
else:
print >>sys.stderr, 'Worker.finalize_bundle: installing (copying) dependencies to %s (MakeBundle)' % temp_dir
bundle.install_dependencies(self.bundle_store, self.get_parent_dict(bundle), temp_dir, copy=True)
db_update['data_hash'] = '0x%s' % path_util.hash_path(temp_dir)
metadata.update(data_size=path_util.get_size(temp_dir))
except Exception as e:
print '=== INTERNAL ERROR: %s' % e
traceback.print_exc()
success = False
metadata['failure_message'] = 'Internal error: ' + e.message
# Clean up any state for RunBundles.
if isinstance(bundle, RunBundle):
try:
self.machine.finalize_bundle(bundle)
except Exception as e:
success = False
if 'failure_message' not in metadata:
metadata['failure_message'] = e.message
else:
metadata['failure_message'] += '\n' + e.message
state = State.READY if success else State.FAILED
db_update['state'] = state
print '-- END BUNDLE: %s [%s]' % (bundle, state)
print ''
self._update_events_log('finalize_bundle', bundle, (bundle.uuid, state, metadata))
# Update user statistics
self.model.increment_user_time_used(bundle.owner_id, getattr(bundle.metadata, 'time', 0))
self.model.update_user_disk_used(bundle.owner_id)
# Update database!
self.model.update_bundle(bundle, db_update)
def upload(self, sources, follow_symlinks, exclude_patterns, git, unpack, remove_sources):
'''
|sources|: specifies the locations of the contents to upload. Each element is either a URL or a local path.
|follow_symlinks|: for local path(s), whether to follow (resolve) symlinks
|exclude_patterns|: for local path(s), don't upload these patterns (e.g., *.o)
|git|: for URL, whether |source| is a git repo to clone.
|unpack|: for each source in |sources|, whether to unpack it if it's an archive.
|remove_sources|: remove |sources|.
If |sources| contains one source, then the bundle contents will be that source.
Otherwise, the bundle contents will be a directory with each of the sources.
Exceptions:
- If |git|, then each source is replaced with the result of running 'git clone |source|'
- If |unpack| is True or a source is an archive (zip, tar.gz, etc.), then unpack the source.
Install the contents of the directory at |source| into
DATA_SUBDIRECTORY in a subdirectory named by a hash of the contents.
Return a (data_hash, metadata) pair, where the metadata is a dict mapping
keys to precomputed statistics about the new data directory.
'''
to_delete = []
# Create temporary directory as a staging area and put everything there.
temp_path = tempfile.mkdtemp('-bundle_store_upload')
temp_subpaths = []
for source in sources:
# Where to save |source| to (might change this value if we unpack).
temp_subpath = os.path.join(temp_path, os.path.basename(source))
if remove_sources:
to_delete.append(source)
source_unpack = unpack and zip_util.path_is_archive(source)
if path_util.path_is_url(source):
# Download the URL.
print_util.open_line('BundleStore.upload: downloading %s to %s' % (source, temp_path))
if git:
file_util.git_clone(source, temp_subpath)
else:
file_util.download_url(source, temp_subpath, print_status=True)
if source_unpack:
zip_util.unpack(temp_subpath, zip_util.strip_archive_ext(temp_subpath))
path_util.remove(temp_subpath)
temp_subpath = zip_util.strip_archive_ext(temp_subpath)
print_util.clear_line()
else:
# Copy the local path.
source_path = path_util.normalize(source)
path_util.check_isvalid(source_path, 'upload')
# Recursively copy the directory into a new BundleStore temp directory.
print_util.open_line('BundleStore.upload: %s => %s' % (source_path, temp_subpath))
if source_unpack:
zip_util.unpack(source_path, zip_util.strip_archive_ext(temp_subpath))
temp_subpath = zip_util.strip_archive_ext(temp_subpath)
else:
if remove_sources:
path_util.rename(source_path, temp_subpath)
else:
path_util.copy(source_path, temp_subpath, follow_symlinks=follow_symlinks, exclude_patterns=exclude_patterns)
print_util.clear_line()
temp_subpaths.append(temp_subpath)
# If exactly one source, then upload that directly.
if len(temp_subpaths) == 1:
to_delete.append(temp_path)
temp_path = temp_subpaths[0]
# Multiplex between uploading a directory and uploading a file here.
# All other path_util calls will use these lists of directories and files.
if os.path.isdir(temp_path):
dirs_and_files = path_util.recursive_ls(temp_path)
else:
dirs_and_files = ([], [temp_path])
# Hash the contents of the temporary directory, and then if there is no
# data with this hash value, move this directory into the data directory.
print_util.open_line('BundleStore.upload: hashing %s' % temp_path)
data_hash = '0x%s' % (path_util.hash_directory(temp_path, dirs_and_files),)
print_util.clear_line()
print_util.open_line('BundleStore.upload: computing size of %s' % temp_path)
data_size = path_util.get_size(temp_path, dirs_and_files)
print_util.clear_line()
final_path = os.path.join(self.data, data_hash)
if os.path.exists(final_path):
# Already exists, just delete it
path_util.remove(temp_path)
else:
print >>sys.stderr, 'BundleStore.upload: moving %s to %s' % (temp_path, final_path)
path_util.rename(temp_path, final_path)
# Delete paths.
for path in to_delete:
if os.path.exists(path):
path_util.remove(path)
# After this operation there should always be a directory at the final path.
assert(os.path.lexists(final_path)), 'Uploaded to %s failed!' % (final_path,)
return (data_hash, {'data_size': data_size})
def update_running_bundle(self, status):
"""
Update the database with information about the bundle given by |status|.
If the bundle is completed, then we need to install the bundle and clean up.
"""
status["last_updated"] = int(time.time())
# Update the bundle's data with status (which is the new information).
bundle = status["bundle"]
# Update to the database
db_update = {}
# Update state
if "state" in status and status["state"]:
db_update["state"] = status["state"]
# Add metadata from the machine
db_update["metadata"] = metadata = {}
bundle_subclass = type(bundle)
for spec in bundle_subclass.METADATA_SPECS:
value = status.get(spec.key)
if value is not None:
metadata[spec.key] = value
# print 'update_running_bundle', status
# See if the bundle is completed.
success = status.get("success")
if success is not None:
# Re-install dependencies.
# - For RunBundle, remove the dependencies.
# - For MakeBundle, copy. This way, we maintain the invariant that
# we always only need to look back one-level at the dependencies,
# not recurse.
try:
temp_dir = status.get("temp_dir")
if not temp_dir:
temp_dir = bundle.metadata.temp_dir
if isinstance(bundle, RunBundle):
print >> sys.stderr, "Worker.finalize_bundle: removing dependencies from %s (RunBundle)" % temp_dir
bundle.remove_dependencies(self.bundle_store, self.get_parent_dict(bundle), temp_dir)
else:
print >> sys.stderr, "Worker.finalize_bundle: installing (copying) dependencies to %s (MakeBundle)" % temp_dir
bundle.install_dependencies(self.bundle_store, self.get_parent_dict(bundle), temp_dir, copy=True)
db_update["data_hash"] = "0x%s" % path_util.hash_path(temp_dir)
metadata.update(data_size=path_util.get_size(temp_dir))
except Exception as e:
print "=== INTERNAL ERROR: %s" % e
traceback.print_exc()
success = False
metadata["failure_message"] = "Internal error: " + e.message
# Clean up any state for RunBundles.
if isinstance(bundle, RunBundle):
try:
self.machine.finalize_bundle(bundle)
except Exception as e:
success = False
if "failure_message" not in metadata:
metadata["failure_message"] = e.message
else:
metadata["failure_message"] += "\n" + e.message
state = State.READY if success else State.FAILED
db_update["state"] = state
print "-- END BUNDLE: %s [%s]" % (bundle, state)
print ""
self._update_events_log("finalize_bundle", bundle, (bundle.uuid, state, metadata))
# Update user statistics
self.model.increment_user_time_used(bundle.owner_id, getattr(bundle.metadata, "time", 0))
self.model.update_user_disk_used(bundle.owner_id)
# Update database!
self.model.update_bundle(bundle, db_update)
def upload(self, sources, follow_symlinks, exclude_patterns, git, unpack, remove_sources, uuid):
"""
|sources|: specifies the locations of the contents to upload. Each element is either a URL or a local path.
|follow_symlinks|: for local path(s), whether to follow (resolve) symlinks
|exclude_patterns|: for local path(s), don't upload these patterns (e.g., *.o)
|git|: for URL, whether |source| is a git repo to clone.
|unpack|: for each source in |sources|, whether to unpack it if it's an archive.
|remove_sources|: remove |sources|.
If |sources| contains one source, then the bundle contents will be that source.
Otherwise, the bundle contents will be a directory with each of the sources.
Exceptions:
- If |git|, then each source is replaced with the result of running 'git clone |source|'
- If |unpack| is True or a source is an archive (zip, tar.gz, etc.), then unpack the source.
Install the contents of the directory at |source| into
DATA_SUBDIRECTORY in a subdirectory named by a hash of the contents.
Return a (data_hash, metadata) pair, where the metadata is a dict mapping
keys to precomputed statistics about the new data directory.
"""
to_delete = []
# If just a single file, set the final path to be equal to that file
single_path = len(sources) == 1
# Determine which disk this will go on
disk_choice = self.ring.get_node(uuid)
final_path = os.path.join(self.partitions, disk_choice, self.DATA_SUBDIRECTORY, uuid)
if os.path.exists(final_path):
raise UsageError('Path %s already present in bundle store' % final_path)
# Only make if not there
elif not single_path:
path_util.make_directory(final_path)
# Paths to resources
subpaths = []
for source in sources:
# Where to save |source| to (might change this value if we unpack).
if not single_path:
subpath = os.path.join(final_path, os.path.basename(source))
else:
subpath = final_path
if remove_sources:
to_delete.append(source)
source_unpack = unpack and zip_util.path_is_archive(source)
if source_unpack and single_path:
# Load the file into the bundle store under the given path
subpath += zip_util.get_archive_ext(source)
if path_util.path_is_url(source):
# Download the URL.
print_util.open_line('BundleStore.upload: downloading %s to %s' % (source, subpath))
if git:
file_util.git_clone(source, subpath)
else:
file_util.download_url(source, subpath, print_status=True)
if source_unpack:
zip_util.unpack(subpath, zip_util.strip_archive_ext(subpath))
path_util.remove(subpath)
subpath = zip_util.strip_archive_ext(subpath)
print_util.clear_line()
else:
# Copy the local path.
source_path = path_util.normalize(source)
path_util.check_isvalid(source_path, 'upload')
# Recursively copy the directory into the BundleStore
print_util.open_line('BundleStore.upload: %s => %s' % (source_path, subpath))
if source_unpack:
zip_util.unpack(source_path, zip_util.strip_archive_ext(subpath))
subpath = zip_util.strip_archive_ext(subpath)
else:
if remove_sources:
path_util.rename(source_path, subpath)
else:
path_util.copy(source_path, subpath, follow_symlinks=follow_symlinks, exclude_patterns=exclude_patterns)
print_util.clear_line()
subpaths.append(subpath)
dirs_and_files = None
if os.path.isdir(final_path):
dirs_and_files = path_util.recursive_ls(final_path)
else:
dirs_and_files = [], [final_path]
# Hash the contents of the bundle directory. Update the data_hash attribute
# for the bundle
print_util.open_line('BundleStore.upload: hashing %s' % final_path)
data_hash = '0x%s' % (path_util.hash_directory(final_path, dirs_and_files))
print_util.clear_line()
print_util.open_line('BundleStore.upload: computing size of %s' % final_path)
data_size = path_util.get_size(final_path, dirs_and_files)
print_util.clear_line()
# Delete paths.
for path in to_delete:
if os.path.exists(path):
path_util.remove(path)
# After this operation there should always be a directory at the final path.
assert (os.path.lexists(final_path)), 'Uploaded to %s failed!' % (final_path,)
return (data_hash, {'data_size': data_size})
def upload(self, path, follow_symlinks, exclude_patterns):
'''
Copy the contents of the directory at |path| into the data subdirectory,
in a subfolder named by a hash of the contents of the new data directory.
If |path| is in a temporary directory, then we just move it.
Return a (data_hash, metadata) pair, where the metadata is a dict mapping
keys to precomputed statistics about the new data directory.
'''
# Create temporary directory as a staging area.
# If |path| is already temporary, then we use that directly
# (with the understanding that |path| will be moved)
if not isinstance(path, list) and os.path.realpath(path).startswith(
os.path.realpath(self.temp)):
temp_path = path
else:
temp_path = os.path.join(self.temp, uuid.uuid4().hex)
if not isinstance(path, list) and path_util.path_is_url(path):
# Have to be careful. Want to make sure if we're fetching a URL
# that points to a file, we are allowing this.
if path.startswith('file://'):
path_suffix = path[7:]
if os.path.islink(path_suffix):
raise UsageError('Not allowed to upload symlink %s' %
path_suffix)
if not any(
path_suffix.startswith(f)
for f in self.direct_upload_paths):
raise UsageError(
'Not allowed to upload %s (only %s allowed)' %
(path_suffix, self.direct_upload_paths))
# Download |path| if it is a URL.
print >> sys.stderr, 'BundleStore.upload: downloading %s to %s' % (
path, temp_path)
file_util.download_url(path, temp_path, print_status=True)
elif path != temp_path:
# Copy |path| into the temp_path.
if isinstance(path, list):
absolute_path = [path_util.normalize(p) for p in path]
for p in absolute_path:
path_util.check_isvalid(p, 'upload')
else:
absolute_path = path_util.normalize(path)
path_util.check_isvalid(absolute_path, 'upload')
# Recursively copy the directory into a new BundleStore temp directory.
print_util.open_line('BundleStore.upload: copying %s to %s' %
(absolute_path, temp_path))
path_util.copy(absolute_path,
temp_path,
follow_symlinks=follow_symlinks,
exclude_patterns=exclude_patterns)
print_util.clear_line()
# Multiplex between uploading a directory and uploading a file here.
# All other path_util calls will use these lists of directories and files.
if os.path.isdir(temp_path):
dirs_and_files = path_util.recursive_ls(temp_path)
else:
dirs_and_files = ([], [temp_path])
# Hash the contents of the temporary directory, and then if there is no
# data with this hash value, move this directory into the data directory.
print_util.open_line('BundleStore.upload: hashing %s' % temp_path)
data_hash = '0x%s' % (path_util.hash_directory(temp_path,
dirs_and_files), )
print_util.clear_line()
print_util.open_line('BundleStore.upload: computing size of %s' %
temp_path)
data_size = path_util.get_size(temp_path, dirs_and_files)
print_util.clear_line()
final_path = os.path.join(self.data, data_hash)
final_path_exists = False
try:
# If data_hash already exists, then we don't need to move it over.
os.utime(final_path, None)
final_path_exists = True
except OSError, e:
if e.errno == errno.ENOENT:
print >> sys.stderr, 'BundleStore.upload: moving %s to %s' % (
temp_path, final_path)
path_util.rename(temp_path, final_path)
else:
raise
def upload(self, path, follow_symlinks):
"""
Copy the contents of the directory at |path| into the data subdirectory,
in a subfolder named by a hash of the contents of the new data directory.
If |path| is in a temporary directory, then we just move it.
Return a (data_hash, metadata) pair, where the metadata is a dict mapping
keys to precomputed statistics about the new data directory.
"""
# Create temporary directory as a staging area.
# If |path| is already temporary, then we use that directly
# (with the understanding that |path| will be moved)
if not isinstance(path, list) and os.path.realpath(path).startswith(os.path.realpath(self.temp)):
temp_path = path
else:
temp_path = os.path.join(self.temp, uuid.uuid4().hex)
if not isinstance(path, list) and path_util.path_is_url(path):
# Have to be careful. Want to make sure if we're fetching a URL
# that points to a file, we are allowing this.
if path.startswith("file://"):
path_suffix = path[7:]
if os.path.islink(path_suffix):
raise UsageError("Not allowed to upload symlink %s" % path_suffix)
if not any(path_suffix.startswith(f) for f in self.direct_upload_paths):
raise UsageError(
"Not allowed to upload %s (only %s allowed)" % (path_suffix, self.direct_upload_paths)
)
# Download |path| if it is a URL.
print >>sys.stderr, "BundleStore.upload: downloading %s to %s" % (path, temp_path)
file_util.download_url(path, temp_path, print_status=True)
elif path != temp_path:
# Copy |path| into the temp_path.
if isinstance(path, list):
absolute_path = [path_util.normalize(p) for p in path]
for p in absolute_path:
path_util.check_isvalid(p, "upload")
else:
absolute_path = path_util.normalize(path)
path_util.check_isvalid(absolute_path, "upload")
# Recursively copy the directory into a new BundleStore temp directory.
print >>sys.stderr, "BundleStore.upload: copying %s to %s" % (absolute_path, temp_path)
path_util.copy(absolute_path, temp_path, follow_symlinks=follow_symlinks)
# Multiplex between uploading a directory and uploading a file here.
# All other path_util calls will use these lists of directories and files.
if os.path.isdir(temp_path):
dirs_and_files = path_util.recursive_ls(temp_path)
else:
dirs_and_files = ([], [temp_path])
# Hash the contents of the temporary directory, and then if there is no
# data with this hash value, move this directory into the data directory.
print >>sys.stderr, "BundleStore.upload: hashing %s" % (temp_path)
data_hash = "0x%s" % (path_util.hash_directory(temp_path, dirs_and_files),)
data_size = path_util.get_size(temp_path, dirs_and_files)
final_path = os.path.join(self.data, data_hash)
final_path_exists = False
try:
# If data_hash already exists, then we don't need to move it over.
os.utime(final_path, None)
final_path_exists = True
except OSError, e:
if e.errno == errno.ENOENT:
print >>sys.stderr, "BundleStore.upload: moving %s to %s" % (temp_path, final_path)
path_util.rename(temp_path, final_path)
else:
raise
