1 # Cumulus: Efficient Filesystem Backup to the Cloud
2 # Copyright (C) 2008-2009, 2012 The Cumulus Developers
3 # See the AUTHORS file for a list of contributors.
5 # This program is free software; you can redistribute it and/or modify
6 # it under the terms of the GNU General Public License as published by
7 # the Free Software Foundation; either version 2 of the License, or
8 # (at your option) any later version.
10 # This program is distributed in the hope that it will be useful,
11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 # GNU General Public License for more details.
15 # You should have received a copy of the GNU General Public License along
16 # with this program; if not, write to the Free Software Foundation, Inc.,
17 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
19 """High-level interface for working with Cumulus archives.
21 This module provides an easy interface for reading from and manipulating
22 various parts of a Cumulus archive:
23 - listing the snapshots and segments present
24 - reading segment contents
25 - parsing snapshot descriptors and snapshot metadata logs
26 - reading and maintaining the local object database
29 from __future__ import division
40 import cumulus.store.file
42 # The largest supported snapshot format that can be understood.
43 FORMAT_VERSION = (0, 11) # Cumulus Snapshot v0.11
45 # Maximum number of nested indirect references allowed in a snapshot.
46 MAX_RECURSION_DEPTH = 3
48 # All segments which have been accessed this session.
49 accessed_segments = set()
51 # Table of methods used to filter segments before storage, and corresponding
52 # filename extensions. These are listed in priority order (methods earlier in
53 # the list are tried first).
55 (".gpg", "cumulus-filter-gpg --decrypt"),
57 (".bz2", "bzip2 -dc"),
62 """Decode a URI-encoded (%xx escapes) string."""
63 def hex_decode(m): return chr(int(m.group(1), 16))
64 return re.sub(r"%([0-9a-f]{2})", hex_decode, s)
66 """Encode a string to URI-encoded (%xx escapes) form."""
68 if c > '+' and c < '\x7f' and c != '@':
71 return "%%%02x" % (ord(c),)
72 return ''.join(hex_encode(c) for c in s)
75 """A class which merely acts as a data container.
77 Instances of this class (or its subclasses) are merely used to store data
78 in various attributes. No methods are provided.
82 return "<%s %s>" % (self.__class__, self.__dict__)
84 CHECKSUM_ALGORITHMS = {
86 'sha224': hashlib.sha224,
87 'sha256': hashlib.sha256,
90 class ChecksumCreator:
91 """Compute a Cumulus checksum for provided data.
93 The algorithm used is selectable, but currently defaults to sha1.
96 def __init__(self, algorithm='sha1'):
97 self.algorithm = algorithm
98 self.hash = CHECKSUM_ALGORITHMS[algorithm]()
100 def update(self, data):
101 self.hash.update(data)
105 return "%s=%s" % (self.algorithm, self.hash.hexdigest())
107 class ChecksumVerifier:
108 """Verify whether a checksum from a snapshot matches the supplied data."""
110 def __init__(self, checksumstr):
111 """Create an object to check the supplied checksum."""
113 (algo, checksum) = checksumstr.split("=", 1)
114 self.checksum = checksum
115 self.hash = CHECKSUM_ALGORITHMS[algo]()
117 def update(self, data):
118 self.hash.update(data)
121 """Return a boolean indicating whether the checksum matches."""
123 result = self.hash.hexdigest()
124 return result == self.checksum
126 class SearchPathEntry(object):
127 """Item representing a possible search location for Cumulus files.
129 Some Cumulus files might be stored in multiple possible file locations: due
130 to format (different compression mechanisms with different extensions),
131 locality (different segments might be placed in different directories to
132 control archiving policies), for backwards compatibility (default location
133 changed over time). A SearchPathEntry describes a possible location for a
136 def __init__(self, directory_prefix, suffix, context=None):
137 self._directory_prefix = directory_prefix
138 self._suffix = suffix
139 self._context = context
142 return "%s(%r, %r, %r)" % (self.__class__.__name__,
143 self._directory_prefix, self._suffix,
146 def build_path(self, basename):
147 """Construct the search path to use for a file with name basename.
149 Returns a tuple (pathname, context), where pathname is the path to try
150 and context is any additional data associated with this search entry
153 return (os.path.join(self._directory_prefix, basename + self._suffix),
156 class SearchPath(object):
157 """A collection of locations to search for files and lookup utilities.
159 For looking for a file in a Cumulus storage backend, a SearchPath object
160 contains a list of possible locations to try. A SearchPath can be used to
161 perform the search as well; when a file is found the search path ordering
162 is updated (moving the successful SearchPathEntry to the front of the list
163 for future searches).
165 def __init__(self, name_regex, searchpath):
166 self._regex = re.compile(name_regex)
167 self._path = list(searchpath)
169 def add_search_entry(self, entry):
170 self._path.append(entry)
172 def directories(self):
173 """Return the set of directories to search for a file type."""
174 return set(entry._directory_prefix for entry in self._path)
176 def get(self, backend, basename):
177 for (i, entry) in enumerate(self._path):
179 (pathname, context) = entry.build_path(basename)
180 fp = backend.get(pathname)
181 # On success, move this entry to the front of the search path
182 # to speed future searches.
185 self._path.insert(0, entry)
186 return (fp, pathname, context)
187 except cumulus.store.NotFoundError:
189 raise cumulus.store.NotFoundError(basename)
191 def stat(self, backend, basename):
192 for (i, entry) in enumerate(self._path):
194 (pathname, context) = entry.build_path(basename)
195 stat_data = backend.stat(pathname)
196 # On success, move this entry to the front of the search path
197 # to speed future searches.
200 self._path.insert(0, entry)
201 result = {"path": pathname}
202 result.update(stat_data)
204 except cumulus.store.NotFoundError:
206 raise cumulus.store.NotFoundError(basename)
208 def match(self, filename):
209 return self._regex.match(filename)
211 def list(self, backend):
213 for d in self.directories():
215 for f in backend.list(d):
218 if m: yield (os.path.join(d, f), m)
219 except cumulus.store.NotFoundError:
222 raise cumulus.store.NotFoundError(backend)
224 def _build_segments_searchpath(prefix):
225 for (extension, filter) in SEGMENT_FILTERS:
226 yield SearchPathEntry(prefix, extension, filter)
229 "checksums": SearchPath(
230 r"^snapshot-(.*)\.(\w+)sums$",
231 [SearchPathEntry("meta", ".sha1sums"),
232 SearchPathEntry("checksums", ".sha1sums"),
233 SearchPathEntry("", ".sha1sums")]),
235 r"^snapshot-(.*)\.meta(\.\S+)?$",
236 _build_segments_searchpath("meta")),
237 "segments": SearchPath(
238 (r"^([0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12})"
241 _build_segments_searchpath("segments0"),
242 _build_segments_searchpath("segments1"),
243 _build_segments_searchpath(""),
244 _build_segments_searchpath("segments"))),
245 "snapshots": SearchPath(
246 r"^snapshot-(.*)\.(cumulus|lbs)$",
247 [SearchPathEntry("snapshots", ".cumulus"),
248 SearchPathEntry("snapshots", ".lbs"),
249 SearchPathEntry("", ".cumulus"),
250 SearchPathEntry("", ".lbs")]),
253 class BackendWrapper(object):
254 """Wrapper around a Cumulus storage backend that understands file types.
256 The BackendWrapper class understands different Cumulus file types, such as
257 snapshots and segments, and implements higher-level operations such as
258 "retrieve a snapshot with a specific name" (hiding operations such as
259 searching for the correct file name).
262 def __init__(self, backend):
263 """Initializes a wrapper around the specified storage backend.
265 store may either be a Store object or URL.
267 if type(backend) in (str, unicode):
268 if backend.find(":") >= 0:
269 self._backend = cumulus.store.open(backend)
271 self._backend = cumulus.store.file.FileStore(backend)
273 self._backend = backend
276 def raw_backend(self):
279 def stat_generic(self, basename, filetype):
280 return SEARCH_PATHS[filetype].stat(self._backend, basename)
282 def open_generic(self, basename, filetype):
283 return SEARCH_PATHS[filetype].get(self._backend, basename)
285 def open_snapshot(self, name):
286 return self.open_generic("snapshot-" + name, "snapshots")
288 def open_segment(self, name):
289 return self.open_generic(name + ".tar", "segments")
291 def list_generic(self, filetype):
292 return ((x[1].group(1), x[0])
293 for x in SEARCH_PATHS[filetype].list(self._backend))
295 def prefetch_generic(self):
296 """Calls scan on directories to prefetch file metadata."""
298 for typeinfo in SEARCH_PATHS.values():
299 directories.update(typeinfo.directories())
300 for d in directories:
302 self._backend.scan(d)
305 def __init__(self, backend):
306 if isinstance(backend, BackendWrapper):
307 self.backend = backend
309 self.backend = BackendWrapper(backend)
314 def get_cachedir(self):
315 if self.cachedir is None:
316 self.cachedir = tempfile.mkdtemp("-cumulus")
320 if self.cachedir is not None:
321 # TODO: Avoid use of system, make this safer
322 os.system("rm -rf " + self.cachedir)
326 def parse_ref(refstr):
327 m = re.match(r"^zero\[(\d+)\]$", refstr)
329 return ("zero", None, None, (0, int(m.group(1)), False))
331 m = re.match(r"^([-0-9a-f]+)\/([0-9a-f]+)(\(\S+\))?(\[(((\d+)\+)?(\d+)|=(\d+))\])?$", refstr)
336 checksum = m.group(3)
339 if checksum is not None:
340 checksum = checksum.lstrip("(").rstrip(")")
342 if slice is not None:
343 if m.group(9) is not None:
344 # Size-assertion slice
345 slice = (0, int(m.group(9)), True)
346 elif m.group(6) is None:
348 slice = (0, int(m.group(8)), False)
350 slice = (int(m.group(7)), int(m.group(8)), False)
352 return (segment, object, checksum, slice)
354 def list_snapshots(self):
355 return set(x[0] for x in self.backend.list_generic("snapshots"))
357 def list_segments(self):
358 return set(x[0] for x in self.backend.list_generic("segments"))
360 def load_snapshot(self, snapshot):
361 snapshot_file = self.backend.open_snapshot(snapshot)[0]
362 return snapshot_file.read().splitlines(True)
365 def filter_data(filehandle, filter_cmd):
366 if filter_cmd is None:
368 (input, output) = os.popen2(filter_cmd)
369 def copy_thread(src, dst):
372 block = src.read(BLOCK_SIZE)
373 if len(block) == 0: break
377 thread.start_new_thread(copy_thread, (filehandle, input))
380 def get_segment(self, segment):
381 accessed_segments.add(segment)
383 (segment_fp, path, filter_cmd) = self.backend.open_segment(segment)
384 return self.filter_data(segment_fp, filter_cmd)
386 def load_segment(self, segment):
387 seg = tarfile.open(segment, 'r|', self.get_segment(segment))
389 data_obj = seg.extractfile(item)
390 path = item.name.split('/')
391 if len(path) == 2 and path[0] == segment:
392 yield (path[1], data_obj.read())
394 def extract_segment(self, segment):
395 segdir = os.path.join(self.get_cachedir(), segment)
397 for (object, data) in self.load_segment(segment):
398 f = open(os.path.join(segdir, object), 'wb')
402 def load_object(self, segment, object):
403 accessed_segments.add(segment)
404 path = os.path.join(self.get_cachedir(), segment, object)
405 if not os.access(path, os.R_OK):
406 self.extract_segment(segment)
407 if segment in self._lru_list: self._lru_list.remove(segment)
408 self._lru_list.append(segment)
409 while len(self._lru_list) > self.CACHE_SIZE:
410 os.system("rm -rf " + os.path.join(self.cachedir,
412 self._lru_list = self._lru_list[1:]
413 return open(path, 'rb').read()
415 def get(self, refstr):
416 """Fetch the given object and return it.
418 The input should be an object reference, in string form.
421 (segment, object, checksum, slice) = self.parse_ref(refstr)
423 if segment == "zero":
424 return "\0" * slice[1]
426 data = self.load_object(segment, object)
428 if checksum is not None:
429 verifier = ChecksumVerifier(checksum)
430 verifier.update(data)
431 if not verifier.valid():
434 if slice is not None:
435 (start, length, exact) = slice
436 if exact and len(data) != length: raise ValueError
437 data = data[start:start+length]
438 if len(data) != length: raise IndexError
443 self.backend.prefetch_generic()
445 def parse(lines, terminate=None):
446 """Generic parser for RFC822-style "Key: Value" data streams.
448 This parser can be used to read metadata logs and snapshot root descriptor
451 lines must be an iterable object which yields a sequence of lines of input.
453 If terminate is specified, it is used as a predicate to determine when to
454 stop reading input lines.
461 # Strip off a trailing newline, if present
462 if len(l) > 0 and l[-1] == "\n":
465 if terminate is not None and terminate(l):
466 if len(dict) > 0: yield dict
471 m = re.match(r"^([-\w]+):\s*(.*)$", l)
473 dict[m.group(1)] = m.group(2)
474 last_key = m.group(1)
475 elif len(l) > 0 and l[0].isspace() and last_key is not None:
480 if len(dict) > 0: yield dict
482 def parse_full(lines):
484 return parse(lines).next()
485 except StopIteration:
488 def parse_metadata_version(s):
489 """Convert a string with the snapshot version format to a tuple."""
491 m = re.match(r"^(?:Cumulus|LBS) Snapshot v(\d+(\.\d+)*)$", s)
495 return tuple([int(d) for d in m.group(1).split(".")])
497 def read_metadata(object_store, root):
498 """Iterate through all lines in the metadata log, following references."""
500 # Stack for keeping track of recursion when following references to
501 # portions of the log. The last entry in the stack corresponds to the
502 # object currently being parsed. Each entry is a list of lines which have
503 # been reversed, so that popping successive lines from the end of each list
504 # will return lines of the metadata log in order.
507 def follow_ref(refstr):
508 if len(stack) >= MAX_RECURSION_DEPTH: raise OverflowError
509 lines = object_store.get(refstr).splitlines(True)
515 while len(stack) > 0:
522 # An indirect reference which we must follow?
523 if len(line) > 0 and line[0] == '@':
531 """Metadata for a single file (or directory or...) from a snapshot."""
533 # Functions for parsing various datatypes that can appear in a metadata log
537 """Decode an integer, expressed in decimal, octal, or hexadecimal."""
538 if s.startswith("0x"):
540 elif s.startswith("0"):
547 """Decode a URI-encoded (%xx escapes) string."""
552 """An unecoded string."""
557 """Decode a user/group to a tuple of uid/gid followed by name."""
559 uid = MetadataItem.decode_int(items[0])
562 if items[1].startswith("(") and items[1].endswith(")"):
563 name = MetadataItem.decode_str(items[1][1:-1])
567 def decode_device(s):
568 """Decode a device major/minor number."""
569 (major, minor) = map(MetadataItem.decode_int, s.split("/"))
570 return (major, minor)
574 def __init__(self, fields, object_store):
575 """Initialize from a dictionary of key/value pairs from metadata log."""
578 self.object_store = object_store
580 self.items = self.Items()
581 for (k, v) in fields.items():
582 if k in self.field_types:
583 decoder = self.field_types[k]
584 setattr(self.items, k, decoder(v))
588 """Return an iterator for the data blocks that make up a file."""
590 # This traverses the list of blocks that make up a file, following
591 # indirect references. It is implemented in much the same way as
592 # read_metadata, so see that function for details of the technique.
594 objects = self.fields['data'].split()
598 def follow_ref(refstr):
599 if len(stack) >= MAX_RECURSION_DEPTH: raise OverflowError
600 objects = self.object_store.get(refstr).split()
602 stack.append(objects)
604 while len(stack) > 0:
611 # An indirect reference which we must follow?
612 if len(ref) > 0 and ref[0] == '@':
617 # Description of fields that might appear, and how they should be parsed.
618 MetadataItem.field_types = {
619 'name': MetadataItem.decode_str,
620 'type': MetadataItem.raw_str,
621 'mode': MetadataItem.decode_int,
622 'device': MetadataItem.decode_device,
623 'user': MetadataItem.decode_user,
624 'group': MetadataItem.decode_user,
625 'ctime': MetadataItem.decode_int,
626 'mtime': MetadataItem.decode_int,
627 'links': MetadataItem.decode_int,
628 'inode': MetadataItem.raw_str,
629 'checksum': MetadataItem.decode_str,
630 'size': MetadataItem.decode_int,
631 'contents': MetadataItem.decode_str,
632 'target': MetadataItem.decode_str,
635 def iterate_metadata(object_store, root):
636 for d in parse(read_metadata(object_store, root), lambda l: len(l) == 0):
637 yield MetadataItem(d, object_store)
640 """Access to the local database of snapshot contents and object checksums.
642 The local database is consulted when creating a snapshot to determine what
643 data can be re-used from old snapshots. Segment cleaning is performed by
644 manipulating the data in the local database; the local database also
645 includes enough data to guide the segment cleaning process.
648 def __init__(self, path, dbname="localdb.sqlite"):
649 self.db_connection = sqlite3.connect(path + "/" + dbname)
651 # Low-level database access. Use these methods when there isn't a
652 # higher-level interface available. Exception: do, however, remember to
653 # use the commit() method after making changes to make sure they are
654 # actually saved, even when going through higher-level interfaces.
656 "Commit any pending changes to the local database."
657 self.db_connection.commit()
660 "Roll back any pending changes to the local database."
661 self.db_connection.rollback()
664 "Return a DB-API cursor for directly accessing the local database."
665 return self.db_connection.cursor()
667 def list_schemes(self):
668 """Return the list of snapshots found in the local database.
670 The returned value is a list of tuples (id, scheme, name, time, intent).
674 cur.execute("select distinct scheme from snapshots")
675 schemes = [row[0] for row in cur.fetchall()]
679 def list_snapshots(self, scheme):
680 """Return a list of snapshots for the given scheme."""
682 cur.execute("select name from snapshots")
683 snapshots = [row[0] for row in cur.fetchall()]
687 def delete_snapshot(self, scheme, name):
688 """Remove the specified snapshot from the database.
690 Warning: This does not garbage collect all dependent data in the
691 database, so it must be followed by a call to garbage_collect() to make
692 the database consistent.
695 cur.execute("delete from snapshots where scheme = ? and name = ?",
698 def prune_old_snapshots(self, scheme, intent=1.0):
699 """Delete entries from old snapshots from the database.
701 Only snapshots with the specified scheme name will be deleted. If
702 intent is given, it gives the intended next snapshot type, to determine
703 how aggressively to clean (for example, intent=7 could be used if the
704 next snapshot will be a weekly snapshot).
709 # Find the id of the last snapshot to be created. This is used for
710 # measuring time in a way: we record this value in each segment we
711 # expire on this run, and then on a future run can tell if there have
712 # been intervening backups made.
713 cur.execute("select max(snapshotid) from snapshots")
714 last_snapshotid = cur.fetchone()[0]
716 # Get the list of old snapshots for this scheme. Delete all the old
717 # ones. Rules for what to keep:
718 # - Always keep the most recent snapshot.
719 # - If snapshot X is younger than Y, and X has higher intent, then Y
721 cur.execute("""select snapshotid, name, intent,
722 julianday('now') - timestamp as age
723 from snapshots where scheme = ?
724 order by age""", (scheme,))
728 for (id, name, snap_intent, snap_age) in cur.fetchall():
730 if snap_intent < max_intent:
731 # Delete small-intent snapshots if there is a more recent
732 # large-intent snapshot.
734 elif snap_intent == intent:
735 # Delete previous snapshots with the specified intent level.
738 if can_delete and not first:
739 print "Delete snapshot %d (%s)" % (id, name)
740 cur.execute("delete from snapshots where snapshotid = ?",
743 max_intent = max(max_intent, snap_intent)
745 self.garbage_collect()
747 def garbage_collect(self):
748 """Garbage-collect unreachable segment and object data.
750 Remove all segments and checksums which is not reachable from the
751 current set of snapshots stored in the local database.
755 # Delete entries in the segment_utilization table which are for
756 # non-existent snapshots.
757 cur.execute("""delete from segment_utilization
758 where snapshotid not in
759 (select snapshotid from snapshots)""")
761 # Delete segments not referenced by any current snapshots.
762 cur.execute("""delete from segments where segmentid not in
763 (select segmentid from segment_utilization)""")
765 # Delete dangling objects in the block_index table.
766 cur.execute("""delete from block_index
767 where segmentid not in
768 (select segmentid from segments)""")
770 # Remove sub-block signatures for deleted objects.
771 cur.execute("""delete from subblock_signatures
773 (select blockid from block_index)""")
776 class SegmentInfo(Struct): pass
778 def get_segment_cleaning_list(self, age_boost=0.0):
779 """Return a list of all current segments with information for cleaning.
781 Return all segments which are currently known in the local database
782 (there might be other, older segments in the archive itself), and
783 return usage statistics for each to help decide which segments to
786 The returned list will be sorted by estimated cleaning benefit, with
787 segments that are best to clean at the start of the list.
789 If specified, the age_boost parameter (measured in days) will added to
790 the age of each segment, as a way of adjusting the benefit computation
791 before a long-lived snapshot is taken (for example, age_boost might be
792 set to 7 when cleaning prior to taking a weekly snapshot).
797 cur.execute("""select segmentid, used, size, mtime,
798 julianday('now') - mtime as age from segment_info
799 where expire_time is null""")
801 info = self.SegmentInfo()
803 info.used_bytes = row[1]
804 info.size_bytes = row[2]
806 info.age_days = row[4]
808 # If data is not available for whatever reason, treat it as 0.0.
809 if info.age_days is None:
811 if info.used_bytes is None:
812 info.used_bytes = 0.0
814 # Benefit calculation: u is the estimated fraction of each segment
815 # which is utilized (bytes belonging to objects still in use
816 # divided by total size; this doesn't take compression or storage
817 # overhead into account, but should give a reasonable estimate).
819 # The total benefit is a heuristic that combines several factors:
820 # the amount of space that can be reclaimed (1 - u), an ageing
821 # factor (info.age_days) that favors cleaning old segments to young
822 # ones and also is more likely to clean segments that will be
823 # rewritten for long-lived snapshots (age_boost), and finally a
824 # penalty factor for the cost of re-uploading data (u + 0.1).
825 u = info.used_bytes / info.size_bytes
826 info.cleaning_benefit \
827 = (1 - u) * (info.age_days + age_boost) / (u + 0.1)
829 segments.append(info)
831 segments.sort(cmp, key=lambda s: s.cleaning_benefit, reverse=True)
834 def mark_segment_expired(self, segment):
835 """Mark a segment for cleaning in the local database.
837 The segment parameter should be either a SegmentInfo object or an
838 integer segment id. Objects in the given segment will be marked as
839 expired, which means that any future snapshots that would re-use those
840 objects will instead write out a new copy of the object, and thus no
841 future snapshots will depend upon the given segment.
844 if isinstance(segment, int):
846 elif isinstance(segment, self.SegmentInfo):
849 raise TypeError("Invalid segment: %s, must be of type int or SegmentInfo, not %s" % (segment, type(segment)))
852 cur.execute("select max(snapshotid) from snapshots")
853 last_snapshotid = cur.fetchone()[0]
854 cur.execute("update segments set expire_time = ? where segmentid = ?",
855 (last_snapshotid, id))
856 cur.execute("update block_index set expired = 0 where segmentid = ?",
859 def balance_expired_objects(self):
860 """Analyze expired objects in segments to be cleaned and group by age.
862 Update the block_index table of the local database to group expired
863 objects by age. The exact number of buckets and the cutoffs for each
864 are dynamically determined. Calling this function after marking
865 segments expired will help in the segment cleaning process, by ensuring
866 that when active objects from clean segments are rewritten, they will
867 be placed into new segments roughly grouped by age.
870 # The expired column of the block_index table is used when generating a
871 # new Cumulus snapshot. A null value indicates that an object may be
872 # re-used. Otherwise, an object must be written into a new segment if
873 # needed. Objects with distinct expired values will be written into
874 # distinct segments, to allow for some grouping by age. The value 0 is
875 # somewhat special in that it indicates any rewritten objects can be
876 # placed in the same segment as completely new objects; this can be
877 # used for very young objects which have been expired, or objects not
878 # expected to be encountered.
880 # In the balancing process, all objects which are not used in any
881 # current snapshots will have expired set to 0. Objects which have
882 # been seen will be sorted by age and will have expired values set to
883 # 0, 1, 2, and so on based on age (with younger objects being assigned
884 # lower values). The number of buckets and the age cutoffs is
885 # determined by looking at the distribution of block ages.
889 # Mark all expired objects with expired = 0; these objects will later
890 # have values set to indicate groupings of objects when repacking.
891 cur.execute("""update block_index set expired = 0
892 where expired is not null""")
894 # We will want to aim for at least one full segment for each bucket
895 # that we eventually create, but don't know how many bytes that should
896 # be due to compression. So compute the average number of bytes in
897 # each expired segment as a rough estimate for the minimum size of each
898 # bucket. (This estimate could be thrown off by many not-fully-packed
899 # segments, but for now don't worry too much about that.) If we can't
900 # compute an average, it's probably because there are no expired
901 # segments, so we have no more work to do.
902 cur.execute("""select avg(size) from segments
904 (select distinct segmentid from block_index
905 where expired is not null)""")
906 segment_size_estimate = cur.fetchone()[0]
907 if not segment_size_estimate:
910 # Next, extract distribution of expired objects (number and size) by
911 # age. Save the timestamp for "now" so that the classification of
912 # blocks into age buckets will not change later in the function, after
913 # time has passed. Set any timestamps in the future to now, so we are
914 # guaranteed that for the rest of this function, age is always
916 cur.execute("select julianday('now')")
917 now = cur.fetchone()[0]
919 cur.execute("""update block_index set timestamp = ?
920 where timestamp > ? and expired is not null""",
923 cur.execute("""select round(? - timestamp) as age, count(*), sum(size)
924 from block_index where expired = 0
925 group by age order by age""", (now,))
926 distribution = cur.fetchall()
928 # Start to determine the buckets for expired objects. Heuristics used:
929 # - An upper bound on the number of buckets is given by the number of
930 # segments we estimate it will take to store all data. In fact,
931 # aim for a couple of segments per bucket.
932 # - Place very young objects in bucket 0 (place with new objects)
933 # unless there are enough of them to warrant a separate bucket.
934 # - Try not to create unnecessarily many buckets, since fewer buckets
935 # will allow repacked data to be grouped based on spatial locality
936 # (while more buckets will group by temporal locality). We want a
939 total_bytes = sum([i[2] for i in distribution])
940 target_buckets = 2 * (total_bytes / segment_size_estimate) ** 0.4
941 min_size = 1.5 * segment_size_estimate
942 target_size = max(2 * segment_size_estimate,
943 total_bytes / target_buckets)
945 print "segment_size:", segment_size_estimate
946 print "distribution:", distribution
947 print "total_bytes:", total_bytes
948 print "target_buckets:", target_buckets
949 print "min, target size:", min_size, target_size
951 # Chosen cutoffs. Each bucket consists of objects with age greater
952 # than one cutoff value, but not greater than the next largest cutoff.
955 # Starting with the oldest objects, begin grouping together into
956 # buckets of size at least target_size bytes.
957 distribution.reverse()
959 min_age_bucket = False
960 for (age, items, size) in distribution:
961 if bucket_size >= target_size \
962 or (age < MIN_AGE and not min_age_bucket):
963 if bucket_size < target_size and len(cutoffs) > 0:
970 min_age_bucket = True
972 # The last (youngest) bucket will be group 0, unless it has enough data
973 # to be of size min_size by itself, or there happen to be no objects
974 # less than MIN_AGE at all.
975 if bucket_size >= min_size or not min_age_bucket:
979 print "cutoffs:", cutoffs
981 # Update the database to assign each object to the appropriate bucket.
983 for i in range(len(cutoffs)):
984 cur.execute("""update block_index set expired = ?
985 where round(? - timestamp) > ?
986 and expired is not null""",
987 (i, now, cutoffs[i]))