Ticket #1400: NEWS-rst

File NEWS-rst, 63.5 KB (added by zooko, at 2011-05-11T02:00:38Z)
Line 
1.. -*- coding: utf-8 -*-
2
3==================================
4User-Visible Changes in Tahoe-LAFS
5==================================
6
7Release 1.8.2 (2011-01-30)
8--------------------------
9
10Compatibility and Dependencies
11''''''''''''''''''''''''''''''
12
13- Tahoe is now compatible with Twisted-10.2 (released last month), as
14  well as with earlier versions. The previous Tahoe-1.8.1 release
15  failed to run against Twisted-10.2, raising an AttributeError on
16  StreamServerEndpointService (#1286)
17- Tahoe now depends upon the "mock" testing library, and the foolscap
18  dependency was raised to 0.6.1 . It no longer requires pywin32
19  (which was used only on windows). Future developers should note that
20  reactor.spawnProcess and derivatives may no longer be used inside
21  Tahoe code.
22
23Other Changes
24'''''''''''''
25
26- the default reserved_space value for new storage nodes is 1 GB (#1208)
27- documentation is now in reStructuredText (.rst) format
28- "tahoe cp" should now handle non-ASCII filenames
29- the unmaintained Mac/Windows GUI applications have been removed (#1282)
30- tahoe processes should appear in top and ps as "tahoe", not
31  "python", on some unix platforms. (#174)
32- "tahoe debug trial" can be used to run the test suite (#1296)
33- the SFTP frontend now reports unknown sizes as "0" instead of "?",
34  to improve compatibility with clients like FileZilla (#1337)
35- "tahoe --version" should now report correct values in situations
36  where 1.8.1 might have been wrong (#1287)
37
38
39Release 1.8.1 (2010-10-28)
40--------------------------
41
42Bugfixes and Improvements
43'''''''''''''''''''''''''
44
45- Allow the repairer to improve the health of a file by uploading some
46  shares, even if it cannot achieve the configured happiness
47  threshold. This fixes a regression introduced between v1.7.1 and
48  v1.8.0. (#1212)
49- Fix a memory leak in the ResponseCache which is used during mutable
50  file/directory operations. (#1045)
51- Fix a regression and add a performance improvement in the
52  downloader.  This issue caused repair to fail in some special
53  cases. (#1223)
54- Fix a bug that caused 'tahoe cp' to fail for a grid-to-grid copy
55  involving a non-ASCII filename. (#1224)
56- Fix a rarely-encountered bug involving printing large strings to the
57  console on Windows. (#1232)
58- Perform ~ expansion in the --exclude-from filename argument to
59  'tahoe backup'. (#1241)
60- The CLI's 'tahoe mv' and 'tahoe ln' commands previously would try to
61  use an HTTP proxy if the HTTP_PROXY environment variable was set.
62  These now always connect directly to the WAPI, thus avoiding giving
63  caps to the HTTP proxy (and also avoiding failures in the case that
64  the proxy is failing or requires authentication). (#1253)
65- The CLI now correctly reports failure in the case that 'tahoe mv'
66  fails to unlink the file from its old location. (#1255)
67- 'tahoe start' now gives a more positive indication that the node has
68  started. (#71)
69- The arguments seen by 'ps' or other tools for node processes are now
70  more useful (in particular, they include the path of the 'tahoe'
71  script, rather than an obscure tool named 'twistd'). (#174)
72
73Removed Features
74''''''''''''''''
75
76- The tahoe start/stop/restart and node creation commands no longer
77  accept the -m or --multiple option, for consistency between
78  platforms.  (#1262)
79
80Packaging
81'''''''''
82
83- We now host binary packages so that users on certain operating
84  systems can install without having a compiler.
85  <http://tahoe-lafs.org/source/tahoe-lafs/deps/tahoe-lafs-dep-eggs/README.html>
86- Use a newer version of a dependency if needed, even if an older
87  version is installed. This would previously cause a VersionConflict
88  error. (#1190)
89- Use a precompiled binary of a dependency if one with a sufficiently
90  high version number is available, instead of attempting to compile
91  the dependency from source, even if the source version has a higher
92  version number. (#1233)
93
94Documentation
95'''''''''''''
96
97- All current documentation in .txt format has been converted to .rst
98  format. (#1225)
99- Added docs/backdoors.rst declaring that we won't add backdoors to
100  Tahoe-LAFS, or add anything to facilitate government access to data.
101  (#1216)
102
103
104Release 1.8.0 (2010-09-23)
105--------------------------
106
107New Features
108''''''''''''
109
110- A completely new downloader which improves performance and
111  robustness of immutable-file downloads. It uses the fastest K
112  servers to download the data in K-way parallel. It automatically
113  fails over to alternate servers if servers fail in mid-download. It
114  allows seeking to arbitrary locations in the file (the previous
115  downloader which would only read the entire file sequentially from
116  beginning to end). It minimizes unnecessary round trips and
117  unnecessary bytes transferred to improve performance. It sends
118  requests to fewer servers to reduce the load on servers (the
119  previous one would send a small request to every server for every
120  download) (#287, #288, #448, #798, #800, #990, #1170, #1191)
121- Non-ASCII command-line arguments and non-ASCII outputs now work on
122  Windows. In addition, the command-line tool now works on 64-bit
123  Windows. (#1074)
124
125Bugfixes and Improvements
126'''''''''''''''''''''''''
127
128- Document and clean up the command-line options for specifying the
129  node's base directory. (#188, #706, #715, #772, #1108)
130- The default node directory for Windows is ".tahoe" in the user's
131  home directory, the same as on other platforms. (#890)
132- Fix a case in which full cap URIs could be logged. (#685, #1155)
133- Fix bug in WUI in Python 2.5 when the system clock is set back to
134  1969. Now you can use Tahoe-LAFS with Python 2.5 and set your system
135  clock to 1969 and still use the WUI. (#1055)
136- Many improvements in code organization, tests, logging,
137  documentation, and packaging. (#983, #1074, #1108, #1127, #1129,
138  #1131, #1166, #1175)
139
140Dependency Updates
141''''''''''''''''''
142
143- on x86 and x86-64 platforms, pycryptopp >= 0.5.20
144- pycrypto 2.2 is excluded due to a bug
145
146
147Release 1.7.1 (2010-07-18)
148--------------------------
149
150Bugfixes and Improvements
151'''''''''''''''''''''''''
152
153- Fix bug in which uploader could fail with AssertionFailure or report
154  that it had achieved servers-of-happiness when it hadn't. (#1118)
155- Fix bug in which servers could get into a state where they would
156  refuse to accept shares of a certain file (#1117)
157- Add init scripts for managing the gateway server on Debian/Ubuntu
158  (#961)
159- Fix bug where server version number was always 0 on the welcome page
160  (#1067)
161- Add new command-line command "tahoe unlink" as a synonym for "tahoe
162  rm" (#776)
163- The FTP frontend now encrypts its temporary files, protecting their
164  contents from an attacker who is able to read the disk. (#1083)
165- Fix IP address detection on FreeBSD 7, 8, and 9 (#1098)
166- Fix minor layout issue in the Web User Interface with Internet
167  Explorer (#1097)
168- Fix rarely-encountered incompatibility between Twisted logging
169  utility and the new unicode support added in v1.7.0 (#1099)
170- Forward-compatibility improvements for non-ASCII caps (#1051)
171
172Code improvements
173'''''''''''''''''
174
175- Simplify and tidy-up directories, unicode support, test code (#923,
176  #967, #1072)
177
178
179Release 1.7.0 (2010-06-18)
180--------------------------
181
182New Features
183''''''''''''
184
185- SFTP support
186  Your Tahoe-LAFS gateway now acts like a full-fledged SFTP server. It
187  has been tested with sshfs to provide a virtual filesystem in Linux.
188  Many users have asked for this feature.  We hope that it serves them
189  well! See the docs/frontends/FTP-and-SFTP.txt document to get
190  started.
191- support for non-ASCII character encodings
192  Tahoe-LAFS now correctly handles filenames containing non-ASCII
193  characters on all supported platforms:
194
195 - when reading files in from the local filesystem (such as when you
196   run "tahoe backup" to back up your local files to a Tahoe-LAFS
197   grid);
198 - when writing files out to the local filesystem (such as when you
199   run "tahoe cp -r" to recursively copy files out of a Tahoe-LAFS
200   grid);
201 - when displaying filenames to the terminal (such as when you run
202   "tahoe ls"), subject to limitations of the terminal and locale;
203 - when parsing command-line arguments, except on Windows.
204
205- Servers of Happiness
206  Tahoe-LAFS now measures during immutable file upload to see how well
207  distributed it is across multiple servers. It aborts the upload if
208  the pieces of the file are not sufficiently well-distributed.
209  This behavior is controlled by a configuration parameter called
210  "servers of happiness". With the default settings for its erasure
211  coding, Tahoe-LAFS generates 10 shares for each file, such that any
212  3 of those shares are sufficient to recover the file. The default
213  value of "servers of happiness" is 7, which means that Tahoe-LAFS
214  will guarantee that there are at least 7 servers holding some of the
215  shares, such that any 3 of those servers can completely recover your
216  file.
217  The new upload code also distributes the shares better than the
218  previous version in some cases and takes better advantage of
219  pre-existing shares (when a file has already been previously
220  uploaded). See the architecture.txt document [3] for details.
221
222
223Bugfixes and Improvements
224'''''''''''''''''''''''''
225
226- Premature abort of upload if some shares were already present and
227  some servers fail. (#608)
228- python ./setup.py install -- can't create or remove files in install
229  directory. (#803)
230- Network failure => internal TypeError. (#902)
231- Install of Tahoe on CentOS 5.4. (#933)
232- CLI option --node-url now supports https url. (#1028)
233- HTML/CSS template files were not correctly installed under
234  Windows. (#1033)
235- MetadataSetter does not enforce restriction on setting "tahoe"
236  subkeys.  (#1034)
237- ImportError: No module named
238  setuptools_darcs.setuptools_darcs. (#1054)
239- Renamed Title in xhtml files. (#1062)
240- Increase Python version dependency to 2.4.4, to avoid a critical
241  CPython security bug. (#1066)
242- Typo correction for the munin plugin tahoe_storagespace. (#968)
243- Fix warnings found by pylint. (#973)
244- Changing format of some documentation files. (#1027)
245- the misc/ directory was tied up. (#1068)
246- The 'ctime' and 'mtime' metadata fields are no longer written except
247   by "tahoe backup". (#924)
248- Unicode filenames in Tahoe-LAFS directories are normalized so that
249  names that differ only in how accents are encoded are treated as the
250  same. (#1076)
251- Various small improvements to documentation. (#937, #911, #1024,
252  #1082)
253
254Removals
255''''''''
256
257- The 'tahoe debug consolidate' subcommand (for converting old
258  allmydata Windows client backups to a newer format) has been
259  removed.
260
261Dependency Updates
262''''''''''''''''''
263
264- the Python version dependency is raised to 2.4.4 in some cases
265  (2.4.3 for Redhat-based Linux distributions, 2.4.2 for UCS-2 builds)
266  (#1066)
267- pycrypto >= 2.0.1
268- pyasn1 >= 0.0.8a
269- mock (only required by unit tests)
270
271
272Release 1.6.1 (2010-02-27)
273--------------------------
274
275Bugfixes
276''''''''
277
278- Correct handling of Small Immutable Directories
279
280  Immutable directories can now be deep-checked and listed in the web
281  UI in all cases. (In v1.6.0, some operations, such as deep-check, on
282  a directory graph that included very small immutable directories,
283  would result in an exception causing the whole operation to abort.)
284  (#948)
285
286Usability Improvements
287''''''''''''''''''''''
288
289- Improved user interface messages and error reporting. (#681, #837,
290  #939)
291- The timeouts for operation handles have been greatly increased, so
292  that you can view the results of an operation up to 4 days after it
293  has completed. After viewing them for the first time, the results
294  are retained for a further day. (#577)
295
296
297Release 1.6.0 (2010-02-01)
298--------------------------
299
300New Features
301''''''''''''
302
303- Immutable Directories
304
305  Tahoe-LAFS can now create and handle immutable directories. (#607,
306  #833, #931) These are read just like normal directories, but are
307  "deep-immutable", meaning that all their children (and everything
308  reachable from those children) must be immutable objects
309  (i.e. immutable or literal files, and other immutable directories).
310
311  These directories must be created in a single webapi call that
312  provides all of the children at once. (Since they cannot be changed
313  after creation, the usual create/add/add sequence cannot be used.)
314  They have URIs that start with "URI:DIR2-CHK:" or "URI:DIR2-LIT:",
315  and are described on the human-facing web interface (aka the "WUI")
316  with a "DIR-IMM" abbreviation (as opposed to "DIR" for the usual
317  read-write directories and "DIR-RO" for read-only directories).
318
319  Tahoe-LAFS releases before 1.6.0 cannot read the contents of an
320  immutable directory. 1.5.0 will tolerate their presence in a
321  directory listing (and display it as "unknown"). 1.4.1 and earlier
322  cannot tolerate them: a DIR-IMM child in any directory will prevent
323  the listing of that directory.
324
325  Immutable directories are repairable, just like normal immutable
326  files.
327
328  The webapi "POST t=mkdir-immutable" call is used to create immutable
329  directories. See docs/frontends/webapi.txt for details.
330
331- "tahoe backup" now creates immutable directories, backupdb has
332  dircache
333
334  The "tahoe backup" command has been enhanced to create immutable
335  directories (in previous releases, it created read-only mutable
336  directories) (#828). This is significantly faster, since it does not
337  need to create an RSA keypair for each new directory. Also "DIR-IMM"
338  immutable directories are repairable, unlike "DIR-RO" read-only
339  mutable directories at present. (A future Tahoe-LAFS release should
340  also be able to repair DIR-RO.)
341
342  In addition, the backupdb (used by "tahoe backup" to remember what
343  it has already copied) has been enhanced to store information about
344  existing immutable directories. This allows it to re-use directories
345  that have moved but still contain identical contents, or that have
346  been deleted and later replaced. (The 1.5.0 "tahoe backup" command
347  could only re-use directories that were in the same place as they
348  were in the immediately previous backup.)  With this change, the
349  backup process no longer needs to read the previous snapshot out of
350  the Tahoe-LAFS grid, reducing the network load considerably. (#606)
351
352  A "null backup" (in which nothing has changed since the previous
353  backup) will require only two Tahoe-side operations: one to add an
354  Archives/$TIMESTAMP entry, and a second to update the Latest/
355  link. On the local disk side, it will readdir() all your local
356  directories and stat() all your local files.
357
358  If you've been using "tahoe backup" for a while, you will notice
359  that your first use of it after upgrading to 1.6.0 may take a long
360  time: it must create proper immutable versions of all the old
361  read-only mutable directories. This process won't take as long as
362  the initial backup (where all the file contents had to be uploaded
363  too): it will require time proportional to the number and size of
364  your directories. After this initial pass, all subsequent passes
365  should take a tiny fraction of the time.
366
367  As noted above, Tahoe-LAFS versions earlier than 1.5.0 cannot list a
368  directory containing an immutable subdirectory. Tahoe-LAFS versions
369  earlier than 1.6.0 cannot read the contents of an immutable
370  directory.
371
372  The "tahoe backup" command has been improved to skip over unreadable
373  objects (like device files, named pipes, and files with permissions
374  that prevent the command from reading their contents), instead of
375  throwing an exception and terminating the backup process. It also
376  skips over symlinks, because these cannot be represented faithfully
377  in the Tahoe-side filesystem. A warning message will be emitted each
378  time something is skipped. (#729, #850, #641)
379
380- "create-node" command added, "create-client" now implies
381  --no-storage
382
383  The basic idea behind Tahoe-LAFS's client+server and client-only
384  processes is that you are creating a general-purpose Tahoe-LAFS
385  "node" process, which has several components that can be
386  activated. Storage service is one of these optional components, as
387  is the Helper, FTP server, and SFTP server. Web gateway
388  functionality is nominally on this list, but it is always active; a
389  future release will make it optional. There are three special
390  purpose servers that can't currently be run as a component in a
391  node: introducer, key-generator, and stats-gatherer.
392
393  So now "tahoe create-node" will create a Tahoe-LAFS node process,
394  and after creation you can edit its tahoe.cfg to enable or disable
395  the desired services. It is a more general-purpose replacement for
396  "tahoe create-client".  The default configuration has storage
397  service enabled. For convenience, the "--no-storage" argument makes
398  a tahoe.cfg file that disables storage service. (#760)
399
400  "tahoe create-client" has been changed to create a Tahoe-LAFS node
401  without a storage service. It is equivalent to "tahoe create-node
402  --no-storage". This helps to reduce the confusion surrounding the
403  use of a command with "client" in its name to create a storage
404  *server*. Use "tahoe create-client" to create a purely client-side
405  node. If you want to offer storage to the grid, use "tahoe
406  create-node" instead.
407
408  In the future, other services will be added to the node, and they
409  will be controlled through options in tahoe.cfg . The most important
410  of these services may get additional --enable-XYZ or --disable-XYZ
411  arguments to "tahoe create-node".
412
413- Performance Improvements
414
415  Download of immutable files begins as soon as the downloader has
416  located the K necessary shares (#928, #287). In both the previous
417  and current releases, a downloader will first issue queries to all
418  storage servers on the grid to locate shares before it begins
419  downloading the shares. In previous releases of Tahoe-LAFS, download
420  would not begin until all storage servers on the grid had replied to
421  the query, at which point K shares would be chosen for download from
422  among the shares that were located. In this release, download begins
423  as soon as any K shares are located. This means that downloads start
424  sooner, which is particularly important if there is a server on the
425  grid that is extremely slow or even hung in such a way that it will
426  never respond. In previous releases such a server would have a
427  negative impact on all downloads from that grid. In this release,
428  such a server will have no impact on downloads, as long as K shares
429  can be found on other, quicker, servers.  This also means that
430  downloads now use the "best-alacrity" servers that they talk to, as
431  measured by how quickly the servers reply to the initial query. This
432  might cause downloads to go faster, especially on grids with
433  heterogeneous servers or geographical dispersion.
434
435Minor Changes
436'''''''''''''
437
438- The webapi acquired a new "t=mkdir-with-children" command, to create
439  and populate a directory in a single call. This is significantly
440  faster than using separate "t=mkdir" and "t=set-children" operations
441  (it uses one gateway-to-grid roundtrip, instead of three or
442  four). (#533)
443
444- The t=set-children (note the hyphen) operation is now documented in
445  docs/frontends/webapi.txt, and is the new preferred spelling of the
446  old t=set_children (with an underscore). The underscore version
447  remains for backwards compatibility. (#381, #927)
448
449- The tracebacks produced by errors in CLI tools should now be in
450  plain text, instead of HTML (which is unreadable outside of a
451  browser). (#646)
452
453- The [storage]reserved_space configuration knob (which causes the
454  storage server to refuse shares when available disk space drops
455  below a threshold) should work on Windows now, not just UNIX. (#637)
456
457- "tahoe cp" should now exit with status "1" if it cannot figure out a
458  suitable target filename, such as when you copy from a bare
459  filecap. (#761)
460
461- "tahoe get" no longer creates a zero-length file upon error. (#121)
462
463- "tahoe ls" can now list single files. (#457)
464
465- "tahoe deep-check --repair" should tolerate repair failures now,
466  instead of halting traversal. (#874, #786)
467
468- "tahoe create-alias" no longer corrupts the aliases file if it had
469  previously been edited to have no trailing newline. (#741)
470
471- Many small packaging improvements were made to facilitate the
472  "tahoe-lafs" package being included in Ubuntu. Several mac/win32
473  binary libraries were removed, some figleaf code-coverage files were
474  removed, a bundled copy of darcsver-1.2.1 was removed, and
475  additional licensing text was added.
476
477- Several DeprecationWarnings for python2.6 were silenced. (#859)
478
479- The checker --add-lease option would sometimes fail for shares
480  stored on old (Tahoe v1.2.0) servers. (#875)
481
482- The documentation for installing on Windows (docs/quickstart.rst)
483  has been improved. (#773)
484
485For other changes not mentioned here, see
486<http://tahoe-lafs.org/trac/tahoe/query?milestone=1.6.0&keywords=!~news-done>.
487To include the tickets mentioned above, go to
488<http://tahoe-lafs.org/trac/tahoe/query?milestone=1.6.0>.
489
490
491Release 1.5.0 (2009-08-01)
492--------------------------
493
494Improvements
495''''''''''''
496
497- Uploads of immutable files now use pipelined writes, improving
498  upload speed slightly (10%) over high-latency connections. (#392)
499
500- Processing large directories has been sped up, by removing a O(N^2)
501  algorithm from the dirnode decoding path and retaining unmodified
502  encrypted entries.  (#750, #752)
503
504- The human-facing web interface (aka the "WUI") received a
505  significant CSS makeover by Kevin Reid, making it much prettier and
506  easier to read. The WUI "check" and "deep-check" forms now include a
507  "Renew Lease" checkbox, mirroring the CLI --add-lease option, so
508  leases can be added or renewed from the web interface.
509
510- The CLI "tahoe mv" command now refuses to overwrite
511  directories. (#705)
512
513- The CLI "tahoe webopen" command, when run without arguments, will
514  now bring up the "Welcome Page" (node status and mkdir/upload
515  forms).
516
517- The 3.5MB limit on mutable files was removed, so it should be
518  possible to upload arbitrarily-sized mutable files. Note, however,
519  that the data format and algorithm remains the same, so using
520  mutable files still requires bandwidth, computation, and RAM in
521  proportion to the size of the mutable file.  (#694)
522
523- This version of Tahoe-LAFS will tolerate directory entries that
524  contain filecap formats which it does not recognize: files and
525  directories from the future.  This should improve the user
526  experience (for 1.5.0 users) when we add new cap formats in the
527  future. Previous versions would fail badly, preventing the user from
528  seeing or editing anything else in those directories. These
529  unrecognized objects can be renamed and deleted, but obviously not
530  read or written. Also they cannot generally be copied. (#683)
531
532Bugfixes
533''''''''
534
535- deep-check-and-repair now tolerates read-only directories, such as
536  the ones produced by the "tahoe backup" CLI command. Read-only
537  directories and mutable files are checked, but not
538  repaired. Previous versions threw an exception when attempting the
539  repair and failed to process the remaining contents. We cannot yet
540  repair these read-only objects, but at least this version allows the
541  rest of the check+repair to proceed. (#625)
542
543- A bug in 1.4.1 which caused a server to be listed multiple times
544  (and frequently broke all connections to that server) was
545  fixed. (#653)
546
547- The plaintext-hashing code was removed from the Helper interface,
548  removing the Helper's ability to mount a
549  partial-information-guessing attack. (#722)
550
551Platform/packaging changes
552''''''''''''''''''''''''''
553
554- Tahoe-LAFS now runs on NetBSD, OpenBSD, ArchLinux, and NixOS, and on
555  an embedded system based on an ARM CPU running at 266 MHz.
556
557- Unit test timeouts have been raised to allow the tests to complete
558  on extremely slow platforms like embedded ARM-based NAS boxes, which
559  may take several hours to run the test suite. An ARM-specific
560  data-corrupting bug in an older version of Crypto++ (5.5.2) was
561  identified: ARM-users are encouraged to use recent
562  Crypto++/pycryptopp which avoids this problem.
563
564- Tahoe-LAFS now requires a SQLite library, either the sqlite3 that
565  comes built-in with python2.5/2.6, or the add-on pysqlite2 if you're
566  using python2.4. In the previous release, this was only needed for
567  the "tahoe backup" command: now it is mandatory.
568
569- Several minor documentation updates were made.
570
571- To help get Tahoe-LAFS into Linux distributions like Fedora and
572  Debian, packaging improvements are being made in both Tahoe-LAFS and
573  related libraries like pycryptopp and zfec.
574
575- The Crypto++ library included in the pycryptopp package has been
576  upgraded to version 5.6.0 of Crypto++, which includes a more
577  efficient implementation of SHA-256 in assembly for x86 or amd64
578  architectures.
579
580dependency updates
581''''''''''''''''''
582
583- foolscap-0.4.1
584- no python-2.4.0 or 2.4.1 (2.4.2 is good) (they contained a bug in base64.b32decode)
585- avoid python-2.6 on windows with mingw: compiler issues
586- python2.4 requires pysqlite2 (2.5,2.6 does not)
587- no python-3.x
588- pycryptopp-0.5.15
589
590
591Release 1.4.1 (2009-04-13)
592--------------------------
593
594Garbage Collection
595''''''''''''''''''
596
597- The big feature for this release is the implementation of garbage
598  collection, allowing Tahoe storage servers to delete shares for old
599  deleted files. When enabled, this uses a "mark and sweep" process:
600  clients are responsible for updating the leases on their shares
601  (generally by running "tahoe deep-check --add-lease"), and servers
602  are allowed to delete any share which does not have an up-to-date
603  lease. The process is described in detail in
604  docs/garbage-collection.txt .
605
606  The server must be configured to enable garbage-collection, by
607  adding directives to the [storage] section that define an age limit
608  for shares. The default configuration will not delete any shares.
609
610  Both servers and clients should be upgraded to this release to make
611  the garbage-collection as pleasant as possible. 1.2.0 servers have
612  code to perform the update-lease operation but it suffers from a
613  fatal bug, while 1.3.0 servers have update-lease but will return an
614  exception for unknown storage indices, causing clients to emit an
615  Incident for each exception, slowing the add-lease process down to a
616  crawl. 1.1.0 servers did not have the add-lease operation at all.
617
618Security/Usability Problems Fixed
619'''''''''''''''''''''''''''''''''
620
621- A super-linear algorithm in the Merkle Tree code was fixed, which
622  previously caused e.g. download of a 10GB file to take several hours
623  before the first byte of plaintext could be produced. The new
624  "alacrity" is about 2 minutes. A future release should reduce this
625  to a few seconds by fixing ticket #442.
626
627- The previous version permitted a small timing attack (due to our use
628  of strcmp) against the write-enabler and lease-renewal/cancel
629  secrets. An attacker who could measure response-time variations of
630  approximatly 3ns against a very noisy background time of about 15ms
631  might be able to guess these secrets. We do not believe this attack
632  was actually feasible. This release closes the attack by first
633  hashing the two strings to be compared with a random secret.
634
635webapi changes
636''''''''''''''
637
638- In most cases, HTML tracebacks will only be sent if an "Accept:
639  text/html" header was provided with the HTTP request. This will
640  generally cause browsers to get an HTMLized traceback but send
641  regular text/plain tracebacks to non-browsers (like the CLI
642  clients). More errors have been mapped to useful HTTP error codes.
643
644- The streaming webapi operations (deep-check and manifest) now have a
645  way to indicate errors (an output line that starts with "ERROR"
646  instead of being legal JSON). See docs/frontends/webapi.txt for
647  details.
648
649- The storage server now has its own status page (at /storage), linked
650  from the Welcome page. This page shows progress and results of the
651  two new share-crawlers: one which merely counts shares (to give an
652  estimate of how many files/directories are being stored in the
653  grid), the other examines leases and reports how much space would be
654  freed if GC were enabled. The page also shows how much disk space is
655  present, used, reserved, and available for the Tahoe server, and
656  whether the server is currently running in "read-write" mode or
657  "read-only" mode.
658
659- When a directory node cannot be read (perhaps because of insufficent
660  shares), a minimal webapi page is created so that the "more-info"
661  links (including a Check/Repair operation) will still be accessible.
662
663- A new "reliability" page was added, with the beginnings of work on a
664  statistical loss model. You can tell this page how many servers you
665  are using and their independent failure probabilities, and it will
666  tell you the likelihood that an arbitrary file will survive each
667  repair period. The "numpy" package must be installed to access this
668  page. A partial paper, written by Shawn Willden, has been added to
669  docs/proposed/lossmodel.lyx .
670
671CLI changes
672'''''''''''
673
674- "tahoe check" and "tahoe deep-check" now accept an "--add-lease"
675  argument, to update a lease on all shares. This is the "mark" side
676  of garbage collection.
677
678- In many cases, CLI error messages have been improved: the ugly
679  HTMLized traceback has been replaced by a normal python traceback.
680
681- "tahoe deep-check" and "tahoe manifest" now have better error
682  reporting.  "tahoe cp" is now non-verbose by default.
683
684- "tahoe backup" now accepts several "--exclude" arguments, to ignore
685  certain files (like editor temporary files and version-control
686  metadata) during backup.
687
688- On windows, the CLI now accepts local paths like "c:\dir\file.txt",
689  which previously was interpreted as a Tahoe path using a "c:" alias.
690
691- The "tahoe restart" command now uses "--force" by default (meaning
692  it will start a node even if it didn't look like there was one
693  already running).
694
695- The "tahoe debug consolidate" command was added. This takes a series
696  of independent timestamped snapshot directories (such as those
697  created by the allmydata.com windows backup program, or a series of
698  "tahoe cp -r" commands) and creates new snapshots that used shared
699  read-only directories whenever possible (like the output of "tahoe
700  backup"). In the most common case (when the snapshots are fairly
701  similar), the result will use significantly fewer directories than
702  the original, allowing "deep-check" and similar tools to run much
703  faster. In some cases, the speedup can be an order of magnitude or
704  more.  This tool is still somewhat experimental, and only needs to
705  be run on large backups produced by something other than "tahoe
706  backup", so it was placed under the "debug" category.
707
708- "tahoe cp -r --caps-only tahoe:dir localdir" is a diagnostic tool
709  which, instead of copying the full contents of files into the local
710  directory, merely copies their filecaps. This can be used to verify
711  the results of a "consolidation" operation.
712
713other fixes
714'''''''''''
715
716- The codebase no longer rauses RuntimeError as a kind of
717  assert(). Specific exception classes were created for each previous
718  instance of RuntimeError.
719
720 -Many unit tests were changed to use a non-network test harness,
721  speeding them up considerably.
722
723- Deep-traversal operations (manifest and deep-check) now walk
724  individual directories in alphabetical order. Occasional turn breaks
725  are inserted to prevent a stack overflow when traversing directories
726  with hundreds of entries.
727
728- The experimental SFTP server had its path-handling logic changed
729  slightly, to accomodate more SFTP clients, although there are still
730  issues (#645).
731
732
733Release 1.3.0 (2009-02-13)
734--------------------------
735
736Checker/Verifier/Repairer
737'''''''''''''''''''''''''
738
739- The primary focus of this release has been writing a checker /
740  verifier / repairer for files and directories.  "Checking" is the
741  act of asking storage servers whether they have a share for the
742  given file or directory: if there are not enough shares available,
743  the file or directory will be unrecoverable. "Verifying" is the act
744  of downloading and cryptographically asserting that the server's
745  share is undamaged: it requires more work (bandwidth and CPU) than
746  checking, but can catch problems that simple checking
747  cannot. "Repair" is the act of replacing missing or damaged shares
748  with new ones.
749
750- This release includes a full checker, a partial verifier, and a
751  partial repairer. The repairer is able to handle missing shares: new
752  shares are generated and uploaded to make up for the missing
753  ones. This is currently the best application of the repairer: to
754  replace shares that were lost because of server departure or
755  permanent drive failure.
756
757- The repairer in this release is somewhat able to handle corrupted
758  shares. The limitations are:
759
760 - Immutable verifier is incomplete: not all shares are used, and not
761   all fields of those shares are verified. Therefore the immutable
762   verifier has only a moderate chance of detecting corrupted shares.
763 - The mutable verifier is mostly complete: all shares are examined,
764   and most fields of the shares are validated.
765 - The storage server protocol offers no way for the repairer to
766   replace or delete immutable shares. If corruption is detected, the
767   repairer will upload replacement shares to other servers, but the
768   corrupted shares will be left in place.
769 - read-only directories and read-only mutable files must be repaired
770   by someone who holds the write-cap: the read-cap is
771   insufficient. Moreover, the deep-check-and-repair operation will
772   halt with an error if it attempts to repair one of these read-only
773   objects.
774 - Some forms of corruption can cause both download and repair
775   operations to fail. A future release will fix this, since download
776   should be tolerant of any corruption as long as there are at least
777   'k' valid shares, and repair should be able to fix any file that is
778   downloadable.
779
780- If the downloader, verifier, or repairer detects share corruption,
781  the servers which provided the bad shares will be notified (via a
782  file placed in the BASEDIR/storage/corruption-advisories directory)
783  so their operators can manually delete the corrupted shares and
784  investigate the problem. In addition, the "incident gatherer"
785  mechanism will automatically report share corruption to an incident
786  gatherer service, if one is configured. Note that corrupted shares
787  indicate hardware failures, serious software bugs, or malice on the
788  part of the storage server operator, so a corrupted share should be
789  considered highly unusual.
790
791- By periodically checking/repairing all files and directories,
792  objects in the Tahoe filesystem remain resistant to recoverability
793  failures due to missing and/or broken servers.
794
795- This release includes a wapi mechanism to initiate checks on
796  individual files and directories (with or without verification, and
797  with or without automatic repair). A related mechanism is used to
798  initiate a "deep-check" on a directory: recursively traversing the
799  directory and its children, checking (and/or verifying/repairing)
800  everything underneath. Both mechanisms can be run with an
801  "output=JSON" argument, to obtain machine-readable check/repair
802  status results. These results include a copy of the filesystem
803  statistics from the "deep-stats" operation (including total number
804  of files, size histogram, etc). If repair is possible, a "Repair"
805  button will appear on the results page.
806
807- The client web interface now features some extra buttons to initiate
808  check and deep-check operations. When these operations finish, they
809  display a results page that summarizes any problems that were
810  encountered. All long-running deep-traversal operations, including
811  deep-check, use a start-and-poll mechanism, to avoid depending upon
812  a single long-lived HTTP connection. docs/frontends/webapi.txt has
813  details.
814
815Efficient Backup
816''''''''''''''''
817
818- The "tahoe backup" command is new in this release, which creates
819  efficient versioned backups of a local directory. Given a local
820  pathname and a target Tahoe directory, this will create a read-only
821  snapshot of the local directory in $target/Archives/$timestamp. It
822  will also create $target/Latest, which is a reference to the latest
823  such snapshot. Each time you run "tahoe backup" with the same source
824  and target, a new $timestamp snapshot will be added. These snapshots
825  will share directories that have not changed since the last backup,
826  to speed up the process and minimize storage requirements. In
827  addition, a small database is used to keep track of which local
828  files have been uploaded already, to avoid uploading them a second
829  time. This drastically reduces the work needed to do a "null backup"
830  (when nothing has changed locally), making "tahoe backup' suitable
831  to run from a daily cronjob.
832
833  Note that the "tahoe backup" CLI command must be used in conjunction
834  with a 1.3.0-or-newer Tahoe client node; there was a bug in the
835  1.2.0 webapi implementation that would prevent the last step (create
836  $target/Latest) from working.
837
838Large Files
839'''''''''''
840
841- The 12GiB (approximate) immutable-file-size limitation is
842  lifted. This release knows how to handle so-called "v2 immutable
843  shares", which permit immutable files of up to about 18 EiB (about
844  3*10^14). These v2 shares are created if the file to be uploaded is
845  too large to fit into v1 shares. v1 shares are created if the file
846  is small enough to fit into them, so that files created with
847  tahoe-1.3.0 can still be read by earlier versions if they are not
848  too large. Note that storage servers also had to be changed to
849  support larger files, and this release is the first release in which
850  they are able to do that. Clients will detect which servers are
851  capable of supporting large files on upload and will not attempt to
852  upload shares of a large file to a server which doesn't support it.
853
854FTP/SFTP Server
855'''''''''''''''
856
857- Tahoe now includes experimental FTP and SFTP servers. When
858  configured with a suitable method to translate username+password
859  into a root directory cap, it provides simple access to the virtual
860  filesystem. Remember that FTP is completely unencrypted: passwords,
861  filenames, and file contents are all sent over the wire in
862  cleartext, so FTP should only be used on a local (127.0.0.1)
863  connection. This feature is still in development: there are no unit
864  tests yet, and behavior with respect to Unicode filenames is
865  uncertain. Please see docs/frontends/FTP-and-SFTP.txt for
866  configuration details. (#512, #531)
867
868CLI Changes
869'''''''''''
870
871- This release adds the 'tahoe create-alias' command, which is a
872  combination of 'tahoe mkdir' and 'tahoe add-alias'. This also allows
873  you to start using a new tahoe directory without exposing its URI in
874  the argv list, which is publicly visible (through the process table)
875  on most unix systems.  Thanks to Kevin Reid for bringing this issue
876  to our attention.
877
878- The single-argument form of "tahoe put" was changed to create an
879  unlinked file. I.e. "tahoe put bar.txt" will take the contents of a
880  local "bar.txt" file, upload them to the grid, and print the
881  resulting read-cap; the file will not be attached to any
882  directories. This seemed a bit more useful than the previous
883  behavior (copy stdin, upload to the grid, attach the resulting file
884  into your default tahoe: alias in a child named 'bar.txt').
885
886- "tahoe put" was also fixed to handle mutable files correctly: "tahoe
887  put bar.txt URI:SSK:..." will read the contents of the local bar.txt
888  and use them to replace the contents of the given mutable file.
889
890- The "tahoe webopen" command was modified to accept aliases. This
891  means "tahoe webopen tahoe:" will cause your web browser to open to
892  a "wui" page that gives access to the directory associated with the
893  default "tahoe:" alias. It should also accept leading slashes, like
894  "tahoe webopen tahoe:/stuff".
895
896- Many esoteric debugging commands were moved down into a "debug"
897  subcommand:
898
899 - tahoe debug dump-cap
900 - tahoe debug dump-share
901 - tahoe debug find-shares
902 - tahoe debug catalog-shares
903 - tahoe debug corrupt-share
904
905   The last command ("tahoe debug corrupt-share") flips a random bit
906   of the given local sharefile. This is used to test the file
907   verifying/repairing code, and obviously should not be used on user
908   data.
909
910The cli might not correctly handle arguments which contain non-ascii
911characters in Tahoe v1.3 (although depending on your platform it
912might, especially if your platform can be configured to pass such
913characters on the command-line in utf-8 encoding).  See
914http://tahoe-lafs.org/trac/tahoe/ticket/565 for details.
915
916Web changes
917'''''''''''
918
919- The "default webapi port", used when creating a new client node (and
920  in the getting-started documentation), was changed from 8123 to
921  3456, to reduce confusion when Tahoe accessed through a Firefox
922  browser on which the "Torbutton" extension has been installed. Port
923  8123 is occasionally used as a Tor control port, so Torbutton adds
924  8123 to Firefox's list of "banned ports" to avoid CSRF attacks
925  against Tor. Once 8123 is banned, it is difficult to diagnose why
926  you can no longer reach a Tahoe node, so the Tahoe default was
927  changed. Note that 3456 is reserved by IANA for the "vat" protocol,
928  but there are argueably more Torbutton+Tahoe users than vat users
929  these days. Note that this will only affect newly-created client
930  nodes. Pre-existing client nodes, created by earlier versions of
931  tahoe, may still be listening on 8123.
932
933- All deep-traversal operations (start-manifest, start-deep-size,
934  start-deep-stats, start-deep-check) now use a start-and-poll
935  approach, instead of using a single (fragile) long-running
936  synchronous HTTP connection. All these "start-" operations use POST
937  instead of GET. The old "GET manifest", "GET deep-size", and "POST
938  deep-check" operations have been removed.
939
940- The new "POST start-manifest" operation, when it finally completes,
941  results in a table of (path,cap), instead of the list of verifycaps
942  produced by the old "GET manifest". The table is available in
943  several formats: use output=html, output=text, or output=json to
944  choose one. The JSON output also includes stats, and a list of
945  verifycaps and storage-index strings. The "return_to=" and
946  "when_done=" arguments have been removed from the t=check and
947  deep-check operations.
948
949- The top-level status page (/status) now has a machine-readable form,
950  via "/status/?t=json". This includes information about the
951  currently-active uploads and downloads, which may be useful for
952  frontends that wish to display progress information. There is no
953  easy way to correlate the activities displayed here with recent wapi
954  requests, however.
955
956- Any files in BASEDIR/public_html/ (configurable) will be served in
957  response to requests in the /static/ portion of the URL space. This
958  will simplify the deployment of javascript-based frontends that can
959  still access wapi calls by conforming to the (regrettable)
960  "same-origin policy".
961
962- The welcome page now has a "Report Incident" button, which is tied
963  into the "Incident Gatherer" machinery. If the node is attached to
964  an incident gatherer (via log_gatherer.furl), then pushing this
965  button will cause an Incident to be signalled: this means recent log
966  events are aggregated and sent in a bundle to the gatherer. The user
967  can push this button after something strange takes place (and they
968  can provide a short message to go along with it), and the relevant
969  data will be delivered to a centralized incident-gatherer for later
970  processing by operations staff.
971
972- The "HEAD" method should now work correctly, in addition to the
973  usual "GET", "PUT", and "POST" methods. "HEAD" is supposed to return
974  exactly the same headers as "GET" would, but without any of the
975  actual response body data. For mutable files, this now does a brief
976  mapupdate (to figure out the size of the file that would be
977  returned), without actually retrieving the file's contents.
978
979- The "GET" operation on files can now support the HTTP "Range:"
980  header, allowing requests for partial content. This allows certain
981  media players to correctly stream audio and movies out of a Tahoe
982  grid. The current implementation uses a disk-based cache in
983  BASEDIR/private/cache/download , which holds the plaintext of the
984  files being downloaded. Future implementations might not use this
985  cache. GET for immutable files now returns an ETag header.
986
987- Each file and directory now has a "Show More Info" web page, which
988  contains much of the information that was crammed into the directory
989  page before. This includes readonly URIs, storage index strings,
990  object type, buttons to control checking/verifying/repairing, and
991  deep-check/deep-stats buttons (for directories). For mutable files,
992  the "replace contents" upload form has been moved here too. As a
993  result, the directory page is now much simpler and cleaner, and
994  several potentially-misleading links (like t=uri) are now gone.
995
996- Slashes are discouraged in Tahoe file/directory names, since they
997  cause problems when accessing the filesystem through the
998  wapi. However, there are a couple of accidental ways to generate
999  such names. This release tries to make it easier to correct such
1000  mistakes by escaping slashes in several places, allowing slashes in
1001  the t=info and t=delete commands, and in the source (but not the
1002  target) of a t=rename command.
1003
1004Packaging
1005'''''''''
1006
1007- Tahoe's dependencies have been extended to require the
1008  "[secure_connections]" feature from Foolscap, which will cause
1009  pyOpenSSL to be required and/or installed. If OpenSSL and its
1010  development headers are already installed on your system, this can
1011  occur automatically. Tahoe now uses pollreactor (instead of the
1012  default selectreactor) to work around a bug between pyOpenSSL and
1013  the most recent release of Twisted (8.1.0). This bug only affects
1014  unit tests (hang during shutdown), and should not impact regular
1015  use.
1016
1017- The Tahoe source code tarballs now come in two different forms:
1018  regular and "sumo". The regular tarball contains just Tahoe, nothing
1019  else. When building from the regular tarball, the build process will
1020  download any unmet dependencies from the internet (starting with the
1021  index at PyPI) so it can build and install them. The "sumo" tarball
1022  contains copies of all the libraries that Tahoe requires (foolscap,
1023  twisted, zfec, etc), so using the "sumo" tarball should not require
1024  any internet access during the build process. This can be useful if
1025  you want to build Tahoe while on an airplane, a desert island, or
1026  other bandwidth-limited environments.
1027
1028- Similarly, tahoe-lafs.org now hosts a "tahoe-deps" tarball which
1029  contains the latest versions of all these dependencies. This
1030  tarball, located at
1031  http://tahoe-lafs.org/source/tahoe/deps/tahoe-deps.tar.gz, can be
1032  unpacked in the tahoe source tree (or in its parent directory), and
1033  the build process should satisfy its downloading needs from it
1034  instead of reaching out to PyPI.  This can be useful if you want to
1035  build Tahoe from a darcs checkout while on that airplane or desert
1036  island.
1037
1038- Because of the previous two changes ("sumo" tarballs and the
1039  "tahoe-deps" bundle), most of the files have been removed from
1040  misc/dependencies/ . This brings the regular Tahoe tarball down to
1041  2MB (compressed), and the darcs checkout (without history) to about
1042  7.6MB. A full darcs checkout will still be fairly large (because of
1043  the historical patches which included the dependent libraries), but
1044  a 'lazy' one should now be small.
1045
1046- The default "make" target is now an alias for "setup.py build",
1047  which itself is an alias for "setup.py develop --prefix support",
1048  with some extra work before and after (see setup.cfg). Most of the
1049  complicated platform-dependent code in the Makefile was rewritten in
1050  Python and moved into setup.py, simplifying things considerably.
1051
1052- Likewise, the "make test" target now delegates most of its work to
1053  "setup.py test", which takes care of getting PYTHONPATH configured
1054  to access the tahoe code (and dependencies) that gets put in
1055  support/lib/ by the build_tahoe step. This should allow unit tests
1056  to be run even when trial (which is part of Twisted) wasn't already
1057  installed (in this case, trial gets installed to support/bin because
1058  Twisted is a dependency of Tahoe).
1059
1060- Tahoe is now compatible with the recently-released Python 2.6 ,
1061  although it is recommended to use Tahoe on Python 2.5, on which it
1062  has received more thorough testing and deployment.
1063
1064- Tahoe is now compatible with simplejson-2.0.x . The previous release
1065  assumed that simplejson.loads always returned unicode strings, which
1066  is no longer the case in 2.0.x .
1067
1068Grid Management Tools
1069'''''''''''''''''''''
1070
1071- Several tools have been added or updated in the misc/ directory,
1072  mostly munin plugins that can be used to monitor a storage grid.
1073
1074 - The misc/spacetime/ directory contains a "disk watcher" daemon
1075   (startable with 'tahoe start'), which can be configured with a set
1076   of HTTP URLs (pointing at the wapi '/statistics' page of a bunch of
1077   storage servers), and will periodically fetch
1078   disk-used/disk-available information from all the servers. It keeps
1079   this information in an Axiom database (a sqlite-based library
1080   available from divmod.org). The daemon computes time-averaged rates
1081   of disk usage, as well as a prediction of how much time is left
1082   before the grid is completely full.
1083
1084 - The misc/munin/ directory contains a new set of munin plugins
1085   (tahoe_diskleft, tahoe_diskusage, tahoe_doomsday) which talk to the
1086   disk-watcher and provide graphs of its calculations.
1087
1088 - To support the disk-watcher, the Tahoe statistics component
1089   (visible through the wapi at the /statistics/ URL) now includes
1090   disk-used and disk-available information. Both are derived through
1091   an equivalent of the unix 'df' command (i.e. they ask the kernel
1092   for the number of free blocks on the partition that encloses the
1093   BASEDIR/storage directory). In the future, the disk-available
1094   number will be further influenced by the local storage policy: if
1095   that policy says that the server should refuse new shares when less
1096   than 5GB is left on the partition, then "disk-available" will
1097   report zero even though the kernel sees 5GB remaining.
1098
1099 - The 'tahoe_overhead' munin plugin interacts with an
1100   allmydata.com-specific server which reports the total of the
1101   'deep-size' reports for all active user accounts, compares this
1102   with the disk-watcher data, to report on overhead percentages. This
1103   provides information on how much space could be recovered once
1104   Tahoe implements some form of garbage collection.
1105
1106Configuration Changes: single INI-format tahoe.cfg file
1107'''''''''''''''''''''''''''''''''''''''''''''''''''''''
1108
1109- The Tahoe node is now configured with a single INI-format file,
1110  named "tahoe.cfg", in the node's base directory. Most of the
1111  previous multiple-separate-files are still read for backwards
1112  compatibility (the embedded SSH debug server and the
1113  advertised_ip_addresses files are the exceptions), but new
1114  directives will only be added to tahoe.cfg . The "tahoe
1115  create-client" command will create a tahoe.cfg for you, with sample
1116  values commented out. (ticket #518)
1117
1118- tahoe.cfg now has controls for the foolscap "keepalive" and
1119  "disconnect" timeouts (#521).
1120
1121- tahoe.cfg now has controls for the encoding parameters:
1122  "shares.needed" and "shares.total" in the "[client]" section. The
1123  default parameters are still 3-of-10.
1124
1125- The inefficient storage 'sizelimit' control (which established an
1126  upper bound on the amount of space that a storage server is allowed
1127  to consume) has been replaced by a lightweight 'reserved_space'
1128  control (which establishes a lower bound on the amount of remaining
1129  space). The storage server will reject all writes that would cause
1130  the remaining disk space (as measured by a '/bin/df' equivalent) to
1131  drop below this value. The "[storage]reserved_space=" tahoe.cfg
1132  parameter controls this setting. (note that this only affects
1133  immutable shares: it is an outstanding bug that reserved_space does
1134  not prevent the allocation of new mutable shares, nor does it
1135  prevent the growth of existing mutable shares).
1136
1137Other Changes
1138'''''''''''''
1139
1140- Clients now declare which versions of the protocols they
1141  support. This is part of a new backwards-compatibility system:
1142  http://tahoe-lafs.org/trac/tahoe/wiki/Versioning .
1143
1144- The version strings for human inspection (as displayed on the
1145  Welcome web page, and included in logs) now includes a platform
1146  identifer (frequently including a linux distribution name, processor
1147  architecture, etc).
1148
1149- Several bugs have been fixed, including one that would cause an
1150  exception (in the logs) if a wapi download operation was cancelled
1151  (by closing the TCP connection, or pushing the "stop" button in a
1152  web browser).
1153
1154- Tahoe now uses Foolscap "Incidents", writing an "incident report"
1155  file to logs/incidents/ each time something weird occurs. These
1156  reports are available to an "incident gatherer" through the flogtool
1157  command. For more details, please see the Foolscap logging
1158  documentation. An incident-classifying plugin function is provided
1159  in misc/incident-gatherer/classify_tahoe.py .
1160
1161- If clients detect corruption in shares, they now automatically
1162  report it to the server holding that share, if it is new enough to
1163  accept the report.  These reports are written to files in
1164  BASEDIR/storage/corruption-advisories .
1165
1166- The 'nickname' setting is now defined to be a UTF-8 -encoded string,
1167  allowing non-ascii nicknames.
1168
1169- The 'tahoe start' command will now accept a --syslog argument and
1170  pass it through to twistd, making it easier to launch non-Tahoe
1171  nodes (like the cpu-watcher) and have them log to syslogd instead of
1172  a local file. This is useful when running a Tahoe node out of a USB
1173  flash drive.
1174
1175- The Mac GUI in src/allmydata/gui/ has been improved.
1176
1177
1178Release 1.2.0 (2008-07-21)
1179--------------------------
1180
1181Security
1182''''''''
1183
1184- This release makes the immutable-file "ciphertext hash tree"
1185  mandatory.  Previous releases allowed the uploader to decide whether
1186  their file would have an integrity check on the ciphertext or not. A
1187  malicious uploader could use this to create a readcap that would
1188  download as one file or a different one, depending upon which shares
1189  the client fetched first, with no errors raised. There are other
1190  integrity checks on the shares themselves, preventing a storage
1191  server or other party from violating the integrity properties of the
1192  read-cap: this failure was only exploitable by the uploader who
1193  gives you a carefully constructed read-cap. If you download the file
1194  with Tahoe 1.2.0 or later, you will not be vulnerable to this
1195  problem. #491
1196
1197  This change does not introduce a compatibility issue, because all
1198  existing versions of Tahoe will emit the ciphertext hash tree in
1199  their shares.
1200
1201Dependencies
1202''''''''''''
1203
1204- Tahoe now requires Foolscap-0.2.9 . It also requires pycryptopp 0.5
1205  or newer, since earlier versions had a bug that interacted with
1206  specific compiler versions that could sometimes result in incorrect
1207  encryption behavior. Both packages are included in the Tahoe source
1208  tarball in misc/dependencies/ , and should be built automatically
1209  when necessary.
1210
1211Web API
1212'''''''
1213
1214- Web API directory pages should now contain properly-slash-terminated
1215  links to other directories. They have also stopped using absolute
1216  links in forms and pages (which interfered with the use of a
1217  front-end load-balancing proxy).
1218
1219- The behavior of the "Check This File" button changed, in conjunction
1220  with larger internal changes to file checking/verification. The
1221  button triggers an immediate check as before, but the outcome is
1222  shown on its own page, and does not get stored anywhere. As a
1223  result, the web directory page no longer shows historical checker
1224  results.
1225
1226- A new "Deep-Check" button has been added, which allows a user to
1227  initiate a recursive check of the given directory and all files and
1228  directories reachable from it. This can cause quite a bit of work,
1229  and has no intermediate progress information or feedback about the
1230  process. In addition, the results of the deep-check are extremely
1231  limited. A later release will improve this behavior.
1232
1233- The web server's behavior with respect to non-ASCII (unicode)
1234  filenames in the "GET save=true" operation has been improved. To
1235  achieve maximum compatibility with variously buggy web browsers, the
1236  server does not try to figure out the character set of the inbound
1237  filename. It just echoes the same bytes back to the browser in the
1238  Content-Disposition header. This seems to make both IE7 and Firefox
1239  work correctly.
1240
1241Checker/Verifier/Repairer
1242'''''''''''''''''''''''''
1243
1244- Tahoe is slowly acquiring convenient tools to check up on file
1245  health, examine existing shares for errors, and repair files that
1246  are not fully healthy. This release adds a mutable
1247  checker/verifier/repairer, although testing is very limited, and
1248  there are no web interfaces to trigger repair yet. The "Check"
1249  button next to each file or directory on the wapi page will perform
1250  a file check, and the "deep check" button on each directory will
1251  recursively check all files and directories reachable from there
1252  (which may take a very long time).
1253
1254  Future releases will improve access to this functionality.
1255
1256Operations/Packaging
1257''''''''''''''''''''
1258
1259- A "check-grid" script has been added, along with a Makefile
1260  target. This is intended (with the help of a pre-configured node
1261  directory) to check upon the health of a Tahoe grid, uploading and
1262  downloading a few files. This can be used as a monitoring tool for a
1263  deployed grid, to be run periodically and to signal an error if it
1264  ever fails. It also helps with compatibility testing, to verify that
1265  the latest Tahoe code is still able to handle files created by an
1266  older version.
1267
1268- The munin plugins from misc/munin/ are now copied into any generated
1269  debian packages, and are made executable (and uncompressed) so they
1270  can be symlinked directly from /etc/munin/plugins/ .
1271
1272- Ubuntu "Hardy" was added as a supported debian platform, with a
1273  Makefile target to produce hardy .deb packages. Some notes have been
1274  added to docs/debian.txt about building Tahoe on a debian/ubuntu
1275  system.
1276
1277- Storage servers now measure operation rates and
1278  latency-per-operation, and provides results through the /statistics
1279  web page as well as the stats gatherer. Munin plugins have been
1280  added to match.
1281
1282Other
1283'''''
1284
1285- Tahoe nodes now use Foolscap "incident logging" to record unusual
1286  events to their NODEDIR/logs/incidents/ directory. These incident
1287  files can be examined by Foolscap logging tools, or delivered to an
1288  external log-gatherer for further analysis. Note that Tahoe now
1289  requires Foolscap-0.2.9, since 0.2.8 had a bug that complained about
1290  "OSError: File exists" when trying to create the incidents/
1291  directory for a second time.
1292
1293- If no servers are available when retrieving a mutable file (like a
1294  directory), the node now reports an error instead of hanging
1295  forever. Earlier releases would not only hang (causing the wapi
1296  directory listing to get stuck half-way through), but the internal
1297  dirnode serialization would cause all subsequent attempts to
1298  retrieve or modify the same directory to hang as well. #463
1299
1300- A minor internal exception (reported in logs/twistd.log, in the
1301  "stopProducing" method) was fixed, which complained about
1302  "self._paused_at not defined" whenever a file download was stopped
1303  from the web browser end.
1304
1305
1306Release 1.1.0 (2008-06-11)
1307--------------------------
1308
1309CLI: new "alias" model
1310''''''''''''''''''''''
1311
1312- The new CLI code uses an scp/rsync -like interface, in which
1313  directories in the Tahoe storage grid are referenced by a
1314  colon-suffixed alias. The new commands look like:
1315
1316 - tahoe cp local.txt tahoe:virtual.txt
1317 - tahoe ls work:subdir
1318
1319- More functionality is available through the CLI: creating unlinked
1320  files and directories, recursive copy in or out of the storage grid,
1321  hardlinks, and retrieving the raw read- or write- caps through the
1322  'ls' command. Please read docs/CLI.txt for complete details.
1323
1324wapi: new pages, new commands
1325'''''''''''''''''''''''''''''
1326
1327- Several new pages were added to the web API:
1328
1329 - /helper_status : to describe what a Helper is doing
1330 - /statistics : reports node uptime, CPU usage, other stats
1331 - /file : for easy file-download URLs, see #221
1332 - /cap == /uri : future compatibility
1333
1334- The localdir=/localfile= and t=download operations were
1335  removed. These required special configuration to enable anyways, but
1336  this feature was a security problem, and was mostly obviated by the
1337  new "cp -r" command.
1338
1339- Several new options to the GET command were added:
1340
1341 -  t=deep-size : add up the size of all immutable files reachable from the directory
1342 -  t=deep-stats : return a JSON-encoded description of number of files, size
1343                distribution, total size, etc
1344
1345- POST is now preferred over PUT for most operations which cause
1346  side-effects.
1347
1348- Most wapi calls now accept overwrite=, and default to overwrite=true
1349
1350- "POST /uri/DIRCAP/parent/child?t=mkdir" is now the preferred API to
1351  create multiple directories at once, rather than ...?t=mkdir-p .
1352
1353- PUT to a mutable file ("PUT /uri/MUTABLEFILECAP", "PUT
1354  /uri/DIRCAP/child") will modify the file in-place.
1355
1356- more munin graphs in misc/munin/
1357
1358 - tahoe-introstats
1359 - tahoe-rootdir-space
1360 - tahoe_estimate_files
1361 - mutable files published/retrieved
1362 - tahoe_cpu_watcher
1363 - tahoe_spacetime
1364
1365New Dependencies
1366''''''''''''''''
1367-  zfec 1.1.0
1368-  foolscap 0.2.8
1369-  pycryptopp 0.5
1370-  setuptools (now required at runtime)
1371
1372New Mutable-File Code
1373'''''''''''''''''''''
1374
1375- The mutable-file handling code (mostly used for directories) has
1376  been completely rewritten. The new scheme has a better API (with a
1377  modify() method) and is less likely to lose data when several
1378  uncoordinated writers change a file at the same time.
1379
1380- In addition, a single Tahoe process will coordinate its own
1381  writes. If you make two concurrent directory-modifying wapi calls to
1382  a single tahoe node, it will internally make one of them wait for
1383  the other to complete. This prevents auto-collision (#391).
1384
1385- The new mutable-file code also detects errors during publish
1386  better. Earlier releases might believe that a mutable file was
1387  published when in fact it failed.
1388
1389other features
1390''''''''''''''
1391
1392- The node now monitors its own CPU usage, as a percentage, measured
1393  every 60 seconds. 1/5/15 minute moving averages are available on the
1394  /statistics web page and via the stats-gathering interface.
1395
1396- Clients now accelerate reconnection to all servers after being
1397  offline (#374). When a client is offline for a long time, it scales
1398  back reconnection attempts to approximately once per hour, so it may
1399  take a while to make the first attempt, but once any attempt
1400  succeeds, the other server connections will be retried immediately.
1401
1402- A new "offloaded KeyGenerator" facility can be configured, to move
1403  RSA key generation out from, say, a wapi node, into a separate
1404  process. RSA keys can take several seconds to create, and so a wapi
1405  node which is being used for directory creation will be unavailable
1406  for anything else during this time. The Key Generator process will
1407  pre-compute a small pool of keys, to speed things up further. This
1408  also takes better advantage of multi-core CPUs, or SMP hosts.
1409
1410- The node will only use a potentially-slow "du -s" command at startup
1411  (to measure how much space has been used) if the "sizelimit"
1412  parameter has been configured (to limit how much space is
1413  used). Large storage servers should turn off sizelimit until a later
1414  release improves the space-management code, since "du -s" on a
1415  terabyte filesystem can take hours.
1416
1417- The Introducer now allows new announcements to replace old ones, to
1418  avoid buildups of obsolete announcements.
1419
1420- Immutable files are limited to about 12GiB (when using the default
1421  3-of-10 encoding), because larger files would be corrupted by the
1422  four-byte share-size field on the storage servers (#439). A later
1423  release will remove this limit. Earlier releases would allow >12GiB
1424  uploads, but the resulting file would be unretrievable.
1425
1426- The docs/ directory has been rearranged, with old docs put in
1427  docs/historical/ and not-yet-implemented ones in docs/proposed/ .
1428
1429- The Mac OS-X FUSE plugin has a significant bug fix: earlier versions
1430  would corrupt writes that used seek() instead of writing the file in
1431  linear order.  The rsync tool is known to perform writes in this
1432  order. This has been fixed.