2 - SNAPSHOT FORMAT CHANGES:
3 - The object slice syntax has been simplified; backup restores
4 are still possible but rebuilding the local database from
5 snapshots before v0.8 will require translation.
6 - Timestamps encoded into the top-level snapshot file name are
7 now UTC instead of the local time zone. This should avoid
8 ambiguity and ensure sorted order is chronological order in
9 the face of time zone changes.
10 - The format identifier is now "Format: Cumulus Snapshot v0.nn"
11 instead of "Format: LBS Snapshot v0.nn" (but the updated tools
12 will accept either variation).
13 - Additional checksum algorithms: sha224, sha256 (sha1 is still
15 - LOCAL DATABASE CHANGES:
16 - Segment utilization data (per snapshot) is tracked
17 differently, to allow better segment cleaning decisions.
18 - New, greatly-enhanced file include/exclude filtering language.
19 This is based on the filter language is rsync (though simplified)
20 and allows glob-like patterns. It also supports filter rules
21 merged at backup-time from files in the file system; this allows,
22 for example, users to control which files in their home
23 directories are included in backups. See doc/exclude.rst for
25 - A new Python-based command-line interface to Cumulus. Eventually
26 this should be the only command directly invoked by the user; it
27 will automate backup expiration, segment cleaning, backup
28 mirroring, etc. Configuration options are read from a config file
29 so they can be applied consistently.
32 - Make a release that packages up various long-existing patches.
33 - Add FTP and SFTP storage backends (from Ralf Schlatterbeck and
38 - Rework storage layer, to make it easier to support local files,
39 S3, and in the future more storage backends. cumulus-util now
40 permits URLS for specifying storage locations, including s3:// for
42 - The script-level interface between the main cumulus binary and
43 remote storage has changed. It should not yet be considered
45 - Add a tool (cumulus-sync) for copying snapshots between storage
47 - Add a (mostly proof-of-concept) FUSE interface for accessing
49 - Implement a basic garbage-collection command for deleting unused
50 segments. There still needs to be a way to select snapshots to
51 delete (perhaps based on some type of schedule).
52 - Assorted minor bugfixes.
55 - SNAPSHOT FORMAT CHANGE: The snapshot format has been extended to
56 allow the local database to be partially rebuilt from the contents
57 of the metadata log. This may be useful for disaster recovery and
58 other scenarios. The changes are small, but old tools will not
60 - Add the --rebuild-statcache option which will re-read all files
61 instead of depending on the statcache file to be correct. This
62 will additionally print warnings if a file has changed but would
63 not have been detected as changed based on the statcache, and will
64 add subfile incremental signatures for existing files that do not
68 - LOCAL DATABASE CHANGE: The addition of subfile incrementals has
69 requires an extension to the local database. The upgrade script
70 contrib/upgrade0.7-localdb.sql should be run prior to running
71 backups with this version.
72 - Name change: the system is now known as Cumulus (replacing the old
73 name of "LBS"). Some traces of the old name still remain.
74 - Initial support for direct backups to remote storage. A sample
75 script is provided for backing up to Amazon S3. Other scripts
76 should be simple to write. The interface should not yet be
77 considered stable, and may change in a future release.
78 - Support for efficient subfile incremental backups. Changes to a
79 small portion of a file do not require re-uploading substantially
80 all of the file, even in cases where there is an insertion or
81 deletion that shifts most of the file contents.
82 - Allow selected files/directories to be restored, not simply an
83 entire snapshot. Additionally, restore files in an order that
84 should optimize performance (restore files based on how they are
85 grouped into segments, instead of lexicographic order).
86 Currently, the implementation of these changes requires that all
87 metadata be loaded into memory when the restore tool runs, so it
88 is more memory-intensive than the old version. This may be fixed
89 in a future version; in the meantime, if the current restore tool
90 requires too much memory, try the old restore tool or the
92 - Add a verbose ("-v") flag. By default, files will not be listed
93 as they are backed up. The old behavior can be turned back on
97 - SNAPSHOT FORMAT CHANGE: A few minor tweaks have been made to the
98 snapshot format. There is nothing substantial--this is primarily
99 a cleaning up of the format before use becomes more widespread.
100 An efficient and explicit means for representing sparse files has
101 been added to the format. Updated tools will be needed to
102 properly understand the new snapshots. Tools will continue to be
103 able to read the old snapshot format.
104 - LOCAL DATABASE CHANGE: The local database has been changed so that
105 segment utilization data for past snapshots is stored more
106 efficiently. The script in contrib/upgrade0.6-localdb.sql must be
107 run on a database to upgrade it prior to running backups with the
109 - The statcache implementation has been reworked, and the format
110 changed. The first backup will run more slowly since the old
111 statcache data will be ignored. New statcache data will be
112 written to statcache2. The old statcache file can be deleted.
113 - Metadata is shared between snapshots where possible. The
114 --full-metadata option can be used to disable this (completely
115 writing out all metadata with the new snapshot).
116 - Add an "intent" field to snapshots, which specifies informally how
117 long a snapshot is intended to be kept. For example, 1 can be
118 used for daily snapshots, and 7 for weekly snapshots. Segment
119 cleaning is now partly guided by intent values, but tuning of the
120 cleaning algorithms is not yet finished.
123 - Have the lbs-util tool check the version number when reading a
124 snapshot, and signal an error if the format is not recognized.
125 - Include a sample script, contrib/parity-gen, for creating
126 RAID-like parity sets to recover from some segment corruption.
127 The par2 command (http://parchive.sourceforge.net/) is used to
128 actually generate the parity sets; the parity-gen script simply
129 automates maintaining the parity sets.
132 - Much improved Python interface for accessing and manipulating LBS
133 archives and local database information. The interface should not
134 yet be considered completely stable.
135 - Python implementation of lbs-util now includes most of the
136 features of the Perl implementation, plus some other new features.
137 The Perl library and utility are deprecated and have been removed
139 - Preliminary snapshot restore support in lbs-util. This is still
140 not yet extensively tested.
143 - Documentation improvements: a getting started README, and a
144 description of some of the implementation details.
145 - Include an example filter script, lbs-filter-gpg, for encrypting
146 segment data as it is written.
147 - Add support for signed snapshot descriptor files (via the
148 --signature-filter option). Due to the checksums contained in the
149 snapshot descriptor, the signature covers the entire contents of
151 - Preview release of lbs-util.py, a Python-based program for
152 managing LBS snapshots. It currently implements a very simple
153 form of automatic segment cleaning, though this isn't much tested.
154 - Improved segment repacking support; after cleaning segments,
155 objects can be written out in different groups based on age. This
156 may help with long-term segment cleaning efficiency, but probably
160 - LOCAL DATABASE CHANGE: A checksums file is now written out along
161 with the snapshot descriptor that contains checksums of segments
162 used in a snapshot. This will allow quick integrity checks of a
163 snapshot, without needing to decompress or decrypt the segment
164 files. The local database schema was modified to store the
165 checksums for segments.
166 - Build dependence on libtar has been dropped. All necessary
167 support for writing TAR files is now directly included with the
169 - The snapshot format is now documented, at least in part. See
171 - Snapshots now include link count and inode number for files with
172 multiple hard links, so that a restore program could determine
173 which files should be hard linked on restore. The reference
174 restore.pl script does not use this.
175 - Bugfix: Print a help message instead of crashing if no files are
177 - Bugfix: File descriptors of files being backed up were closed
178 twice. In some cases this might have led to an unrelated file
179 being closed (if the file descriptor was quickly re-used).
180 - Preview of a new lbs-util command for maintaining snapshots.
181 Functionality is currently limited.
184 - Update reference restore.pl script to handle octal/hexadecimal
185 (format change introduced in 0.2).
186 - Better support for mixing multiple backup schemes: the statcache
187 file is now per-scheme, so performance should be better in cases
188 where multiple different backups (of different source directories)
189 are made, all sharing the same local database directory.
192 - Bugfix: Do not print an error when only one directory is specified
196 - SNAPSHOT FORMAT CHANGE: Metadata logs in backups can now contain
197 octal and hexadecimal values. Old restore programs may not be
198 able to read these new backups. The reference restore.pl script
199 has not yet been updated.
200 - LOCAL DATABASE CHANGE: Backups may be assigned a name with the
201 --scheme= command-line option, to allow different backup sets to
202 be mixed in the same directory. The database schema for the local
204 - Improved selection (inclusion/exclusion) of files to be included
206 - Device major/minor numbers are dumped for block and character