2 = The Tahoe REST-ful Web API =
4 1. Enabling the web-API port
5 2. Basic Concepts: GET, PUT, DELETE, POST
6 3. URLs, Machine-Oriented Interfaces
7 4. Browser Operations: Human-Oriented Interfaces
8 5. Welcome / Debug / Status pages
9 6. Static Files in /public_html
10 7. Safety and security issues -- names vs. URIs
14 == Enabling the web-API port ==
16 Every Tahoe node is capable of running a built-in HTTP server. To enable
17 this, just write a port number into the "[node]web.port" line of your node's
18 tahoe.cfg file. For example, writing "web.port = 3456" into the "[node]"
19 section of $NODEDIR/tahoe.cfg will cause the node to run a webserver on port
22 This string is actually a Twisted "strports" specification, meaning you can
23 get more control over the interface to which the server binds by supplying
24 additional arguments. For more details, see the documentation on
25 twisted.application.strports:
26 http://twistedmatrix.com/documents/current/api/twisted.application.strports.html
28 Writing "tcp:3456:interface=127.0.0.1" into the web.port line does the same
29 but binds to the loopback interface, ensuring that only the programs on the
30 local host can connect. Using
31 "ssl:3456:privateKey=mykey.pem:certKey=cert.pem" runs an SSL server.
33 This webport can be set when the node is created by passing a --webport
34 option to the 'tahoe create-node' command. By default, the node listens on
35 port 3456, on the loopback (127.0.0.1) interface.
39 As described in architecture.txt, each file and directory in a Tahoe virtual
40 filesystem is referenced by an identifier that combines the designation of
41 the object with the authority to do something with it (such as read or modify
42 the contents). This identifier is called a "read-cap" or "write-cap",
43 depending upon whether it enables read-only or read-write access. These
44 "caps" are also referred to as URIs.
46 The Tahoe web-based API is "REST-ful", meaning it implements the concepts of
47 "REpresentational State Transfer": the original scheme by which the World
48 Wide Web was intended to work. Each object (file or directory) is referenced
49 by a URL that includes the read- or write- cap. HTTP methods (GET, PUT, and
50 DELETE) are used to manipulate these objects. You can think of the URL as a
51 noun, and the method as a verb.
53 In REST, the GET method is used to retrieve information about an object, or
54 to retrieve some representation of the object itself. When the object is a
55 file, the basic GET method will simply return the contents of that file.
56 Other variations (generally implemented by adding query parameters to the
57 URL) will return information about the object, such as metadata. GET
58 operations are required to have no side-effects.
60 PUT is used to upload new objects into the filesystem, or to replace an
61 existing object. DELETE it used to delete objects from the filesystem. Both
62 PUT and DELETE are required to be idempotent: performing the same operation
63 multiple times must have the same side-effects as only performing it once.
65 POST is used for more complicated actions that cannot be expressed as a GET,
66 PUT, or DELETE. POST operations can be thought of as a method call: sending
67 some message to the object referenced by the URL. In Tahoe, POST is also used
68 for operations that must be triggered by an HTML form (including upload and
69 delete), because otherwise a regular web browser has no way to accomplish
70 these tasks. In general, everything that can be done with a PUT or DELETE can
71 also be done with a POST.
73 Tahoe's web API is designed for two different consumers. The first is a
74 program that needs to manipulate the virtual file system. Such programs are
75 expected to use the RESTful interface described above. The second is a human
76 using a standard web browser to work with the filesystem. This user is given
77 a series of HTML pages with links to download files, and forms that use POST
78 actions to upload, rename, and delete files.
80 When an error occurs, the HTTP response code will be set to an appropriate
81 400-series code (like 404 for an unknown childname, or 400 Gone when a file
82 is unrecoverable due to insufficient shares), and the HTTP response body will
83 usually contain a few lines of explanation as to the cause of the error and
84 possible responses. Unusual exceptions may result in a 500 Internal Server
85 Error as a catch-all, with a default response body will contain a
86 Nevow-generated HTML-ized representation of the Python exception stack trace
87 that caused the problem. CLI programs which want to copy the response body to
88 stderr should provide an "Accept: text/plain" header to their requests to get
89 a plain text stack trace instead. If the Accept header contains */*, or
90 text/*, or text/html (or if there is no Accept header), HTML tracebacks will
95 Tahoe uses a variety of read- and write- caps to identify files and
96 directories. The most common of these is the "immutable file read-cap", which
97 is used for most uploaded files. These read-caps look like the following:
99 URI:CHK:ime6pvkaxuetdfah2p2f35pe54:4btz54xk3tew6nd4y2ojpxj4m6wxjqqlwnztgre6gnjgtucd5r4a:3:10:202
101 The next most common is a "directory write-cap", which provides both read and
102 write access to a directory, and look like this:
104 URI:DIR2:djrdkfawoqihigoett4g6auz6a:jx5mplfpwexnoqff7y5e4zjus4lidm76dcuarpct7cckorh2dpgq
106 There are also "directory read-caps", which start with "URI:DIR2-RO:", and
107 give read-only access to a directory. Finally there are also mutable file
108 read- and write- caps, which start with "URI:SSK", and give access to mutable
111 (later versions of Tahoe will make these strings shorter, and will remove the
112 unfortunate colons, which must be escaped when these caps are embedded in
115 To refer to any Tahoe object through the web API, you simply need to combine
116 a prefix (which indicates the HTTP server to use) with the cap (which
117 indicates which object inside that server to access). Since the default Tahoe
118 webport is 3456, the most common prefix is one that will use a local node
119 listening on this port:
121 http://127.0.0.1:3456/uri/ + $CAP
123 So, to access the directory named above (which happens to be the
124 publically-writable sample directory on the Tahoe test grid, described at
125 http://allmydata.org/trac/tahoe/wiki/TestGrid), the URL would be:
127 http://127.0.0.1:3456/uri/URI%3ADIR2%3Adjrdkfawoqihigoett4g6auz6a%3Ajx5mplfpwexnoqff7y5e4zjus4lidm76dcuarpct7cckorh2dpgq/
129 (note that the colons in the directory-cap are url-encoded into "%3A"
132 Likewise, to access the file named above, use:
134 http://127.0.0.1:3456/uri/URI%3ACHK%3Aime6pvkaxuetdfah2p2f35pe54%3A4btz54xk3tew6nd4y2ojpxj4m6wxjqqlwnztgre6gnjgtucd5r4a%3A3%3A10%3A202
136 In the rest of this document, we'll use "$DIRCAP" as shorthand for a read-cap
137 or write-cap that refers to a directory, and "$FILECAP" to abbreviate a cap
138 that refers to a file (whether mutable or immutable). So those URLs above can
141 http://127.0.0.1:3456/uri/$DIRCAP/
142 http://127.0.0.1:3456/uri/$FILECAP
144 The operation summaries below will abbreviate these further, by eliding the
145 server prefix. They will be displayed like this:
153 Tahoe directories contain named child entries, just like directories in a regular
154 local filesystem. These child entries, called "dirnodes", consist of a name,
155 metadata, a write slot, and a read slot. The write and read slots normally contain
156 a write-cap and read-cap referring to the same object, which can be either a file
157 or a subdirectory. The write slot may be empty (actually, both may be empty,
158 but that is unusual).
160 If you have a Tahoe URL that refers to a directory, and want to reference a
161 named child inside it, just append the child name to the URL. For example, if
162 our sample directory contains a file named "welcome.txt", we can refer to
165 http://127.0.0.1:3456/uri/$DIRCAP/welcome.txt
167 (or http://127.0.0.1:3456/uri/URI%3ADIR2%3Adjrdkfawoqihigoett4g6auz6a%3Ajx5mplfpwexnoqff7y5e4zjus4lidm76dcuarpct7cckorh2dpgq/welcome.txt)
169 Multiple levels of subdirectories can be handled this way:
171 http://127.0.0.1:3456/uri/$DIRCAP/tahoe-source/docs/webapi.txt
173 In this document, when we need to refer to a URL that references a file using
174 this child-of-some-directory format, we'll use the following string:
176 /uri/$DIRCAP/[SUBDIRS../]FILENAME
178 The "[SUBDIRS../]" part means that there are zero or more (optional)
179 subdirectory names in the middle of the URL. The "FILENAME" at the end means
180 that this whole URL refers to a file of some sort, rather than to a
183 When we need to refer specifically to a directory in this way, we'll write:
185 /uri/$DIRCAP/[SUBDIRS../]SUBDIR
188 Note that all components of pathnames in URLs are required to be UTF-8
189 encoded, so "resume.doc" (with an acute accent on both E's) would be accessed
192 http://127.0.0.1:3456/uri/$DIRCAP/r%C3%A9sum%C3%A9.doc
194 Also note that the filenames inside upload POST forms are interpreted using
195 whatever character set was provided in the conventional '_charset' field, and
196 defaults to UTF-8 if not otherwise specified. The JSON representation of each
197 directory contains native unicode strings. Tahoe directories are specified to
198 contain unicode filenames, and cannot contain binary strings that are not
199 representable as such.
201 All Tahoe operations that refer to existing files or directories must include
202 a suitable read- or write- cap in the URL: the wapi server won't add one
203 for you. If you don't know the cap, you can't access the file. This allows
204 the security properties of Tahoe caps to be extended across the wapi
207 == Slow Operations, Progress, and Cancelling ==
209 Certain operations can be expected to take a long time. The "t=deep-check",
210 described below, will recursively visit every file and directory reachable
211 from a given starting point, which can take minutes or even hours for
212 extremely large directory structures. A single long-running HTTP request is a
213 fragile thing: proxies, NAT boxes, browsers, and users may all grow impatient
214 with waiting and give up on the connection.
216 For this reason, long-running operations have an "operation handle", which
217 can be used to poll for status/progress messages while the operation
218 proceeds. This handle can also be used to cancel the operation. These handles
219 are created by the client, and passed in as a an "ophandle=" query argument
220 to the POST or PUT request which starts the operation. The following
221 operations can then be used to retrieve status:
223 GET /operations/$HANDLE?output=HTML (with or without t=status)
224 GET /operations/$HANDLE?output=JSON (same)
226 These two retrieve the current status of the given operation. Each operation
227 presents a different sort of information, but in general the page retrieved
230 * whether the operation is complete, or if it is still running
231 * how much of the operation is complete, and how much is left, if possible
233 Note that the final status output can be quite large: a deep-manifest of a
234 directory structure with 300k directories and 200k unique files is about
235 275MB of JSON, and might take two minutes to generate. For this reason, the
236 full status is not provided until the operation has completed.
238 The HTML form will include a meta-refresh tag, which will cause a regular
239 web browser to reload the status page about 60 seconds later. This tag will
240 be removed once the operation has completed.
242 There may be more status information available under
243 /operations/$HANDLE/$ETC : i.e., the handle forms the root of a URL space.
245 POST /operations/$HANDLE?t=cancel
247 This terminates the operation, and returns an HTML page explaining what was
248 cancelled. If the operation handle has already expired (see below), this
249 POST will return a 404, which indicates that the operation is no longer
250 running (either it was completed or terminated). The response body will be
251 the same as a GET /operations/$HANDLE on this operation handle, and the
252 handle will be expired immediately afterwards.
254 The operation handle will eventually expire, to avoid consuming an unbounded
255 amount of memory. The handle's time-to-live can be reset at any time, by
256 passing a retain-for= argument (with a count of seconds) to either the
257 initial POST that starts the operation, or the subsequent GET request which
258 asks about the operation. For example, if a 'GET
259 /operations/$HANDLE?output=JSON&retain-for=600' query is performed, the
260 handle will remain active for 600 seconds (10 minutes) after the GET was
263 In addition, if the GET includes a release-after-complete=True argument, and
264 the operation has completed, the operation handle will be released
267 If a retain-for= argument is not used, the default handle lifetimes are:
269 * handles will remain valid at least until their operation finishes
270 * uncollected handles for finished operations (i.e. handles for operations
271 which have finished but for which the GET page has not been accessed since
272 completion) will remain valid for one hour, or for the total time consumed
273 by the operation, whichever is greater.
274 * collected handles (i.e. the GET page has been retrieved at least once
275 since the operation completed) will remain valid for ten minutes.
277 Many "slow" operations can begin to use unacceptable amounts of memory when
278 operation on large directory structures. The memory usage increases when the
279 ophandle is polled, as the results must be copied into a JSON string, sent
280 over the wire, then parsed by a client. So, as an alternative, many "slow"
281 operations have streaming equivalents. These equivalents do not use operation
282 handles. Instead, they emit line-oriented status results immediately. Client
283 code can cancel the operation by simply closing the HTTP connection.
285 == Programmatic Operations ==
287 Now that we know how to build URLs that refer to files and directories in a
288 Tahoe virtual filesystem, what sorts of operations can we do with those URLs?
289 This section contains a catalog of GET, PUT, DELETE, and POST operations that
290 can be performed on these URLs. This set of operations are aimed at programs
291 that use HTTP to communicate with a Tahoe node. A later section describes
292 operations that are intended for web browsers.
294 === Reading A File ===
297 GET /uri/$DIRCAP/[SUBDIRS../]FILENAME
299 This will retrieve the contents of the given file. The HTTP response body
300 will contain the sequence of bytes that make up the file.
302 To view files in a web browser, you may want more control over the
303 Content-Type and Content-Disposition headers. Please see the next section
304 "Browser Operations", for details on how to modify these URLs for that
307 === Writing/Uploading A File ===
310 PUT /uri/$DIRCAP/[SUBDIRS../]FILENAME
312 Upload a file, using the data from the HTTP request body, and add whatever
313 child links and subdirectories are necessary to make the file available at
314 the given location. Once this operation succeeds, a GET on the same URL will
315 retrieve the same contents that were just uploaded. This will create any
316 necessary intermediate subdirectories.
318 To use the /uri/$FILECAP form, $FILECAP be a write-cap for a mutable file.
320 In the /uri/$DIRCAP/[SUBDIRS../]FILENAME form, if the target file is a
321 writable mutable file, that files contents will be overwritten in-place. If
322 it is a read-cap for a mutable file, an error will occur. If it is an
323 immutable file, the old file will be discarded, and a new one will be put in
326 When creating a new file, if "mutable=true" is in the query arguments, the
327 operation will create a mutable file instead of an immutable one.
329 This returns the file-cap of the resulting file. If a new file was created
330 by this method, the HTTP response code (as dictated by rfc2616) will be set
331 to 201 CREATED. If an existing file was replaced or modified, the response
334 Note that the 'curl -T localfile http://127.0.0.1:3456/uri/$DIRCAP/foo.txt'
335 command can be used to invoke this operation.
339 This uploads a file, and produces a file-cap for the contents, but does not
340 attach the file into the virtual drive. No directories will be modified by
341 this operation. The file-cap is returned as the body of the HTTP response.
343 If "mutable=true" is in the query arguments, the operation will create a
344 mutable file, and return its write-cap in the HTTP respose. The default is
345 to create an immutable file, returning the read-cap as a response.
347 === Creating A New Directory ===
352 Create a new empty directory and return its write-cap as the HTTP response
353 body. This does not make the newly created directory visible from the
354 virtual drive. The "PUT" operation is provided for backwards compatibility:
355 new code should use POST.
357 POST /uri?t=mkdir-with-children
359 Create a new directory, populated with a set of child nodes, and return its
360 write-cap as the HTTP response body. The new directory is not attached to
361 any other directory: the returned write-cap is the only reference to it.
363 Initial children are provided as the body of the POST form (this is more
364 efficient than doing separate mkdir and set_children operations). If the
365 body is empty, the new directory will be empty. If not empty, the body will
366 be interpreted as a UTF-8 JSON-encoded dictionary of children with which the
367 new directory should be populated, using the same format as would be
368 returned in the 'children' value of the t=json GET request, described below.
369 Each dictionary key should be a child name, and each value should be a list
370 of [TYPE, PROPDICT], where PROPDICT contains "rw_uri", "ro_uri", and
371 "metadata" keys (all others are ignored). For example, the PUT request body
375 "Fran\u00e7ais": [ "filenode", {
376 "ro_uri": "URI:CHK:...",
379 "ctime": 1202777696.7564139,
380 "mtime": 1202777696.7564139,
382 "linkcrtime": 1202777696.7564139,
383 "linkmotime": 1202777696.7564139,
385 "subdir": [ "dirnode", {
386 "rw_uri": "URI:DIR2:...",
387 "ro_uri": "URI:DIR2-RO:...",
389 "ctime": 1202778102.7589991,
390 "mtime": 1202778111.2160511,
392 "linkcrtime": 1202777696.7564139,
393 "linkmotime": 1202777696.7564139,
397 For forward-compatibility, a mutable directory can also contain caps in
398 a format that is unknown to the webapi server. When such caps are retrieved
399 from a mutable directory in a "ro_uri" field, they will be prefixed with
400 the string "ro.", indicating that they must not be decoded without
401 checking that they are read-only. The "ro." prefix must not be stripped
402 off without performing this check. (Future versions of the webapi server
403 will perform it where necessary.)
405 If both the "rw_uri" and "ro_uri" fields are present in a given PROPDICT,
406 and the webapi server recognizes the rw_uri as a write cap, then it will
407 reset the ro_uri to the corresponding read cap and discard the original
408 contents of ro_uri (in order to ensure that the two caps correspond to the
409 same object and that the ro_uri is in fact read-only). However this may not
410 happen for caps in a format unknown to the webapi server. Therefore, when
411 writing a directory the webapi client should ensure that the contents
412 of "rw_uri" and "ro_uri" for a given PROPDICT are a consistent
413 (write cap, read cap) pair if possible. If the webapi client only has
414 one cap and does not know whether it is a write cap or read cap, then
415 it is acceptable to set "rw_uri" to that cap and omit "ro_uri". The
416 client must not put a write cap into a "ro_uri" field.
418 Note that the webapi-using client application must not provide the
419 "Content-Type: multipart/form-data" header that usually accompanies HTML
420 form submissions, since the body is not formatted this way. Doing so will
421 cause a server error as the lower-level code misparses the request body.
423 Child file names should each be expressed as a unicode string, then used as
424 keys of the dictionary. The dictionary should then be converted into JSON,
425 and the resulting string encoded into UTF-8. This UTF-8 bytestring should
426 then be used as the POST body.
428 POST /uri?t=mkdir-immutable
430 Like t=mkdir-with-children above, but the new directory will be
431 deep-immutable. This means that the directory itself is immutable, and that
432 it can only contain objects that are treated as being deep-immutable, like
433 immutable files, literal files, and deep-immutable directories.
435 For forward-compatibility, a deep-immutable directory can also contain caps
436 in a format that is unknown to the webapi server. When such caps are retrieved
437 from a deep-immutable directory in a "ro_uri" field, they will be prefixed
438 with the string "imm.", indicating that they must not be decoded without
439 checking that they are immutable. The "imm." prefix must not be stripped
440 off without performing this check. (Future versions of the webapi server
441 will perform it where necessary.)
443 The cap for each child may be given either in the "rw_uri" or "ro_uri"
444 field of the PROPDICT (not both). If a cap is given in the "rw_uri" field,
445 then the webapi server will check that it is an immutable read-cap of a
446 *known* format, and give an error if it is not. If a cap is given in the
447 "ro_uri" field, then the webapi server will still check whether known
448 caps are immutable, but for unknown caps it will simply assume that the
449 cap can be stored, as described above. Note that an attacker would be
450 able to store any cap in an immutable directory, so this check when
451 creating the directory is only to help non-malicious clients to avoid
452 accidentally giving away more authority than intended.
454 A non-empty request body is mandatory, since after the directory is created,
455 it will not be possible to add more children to it.
457 POST /uri/$DIRCAP/[SUBDIRS../]SUBDIR?t=mkdir
458 PUT /uri/$DIRCAP/[SUBDIRS../]SUBDIR?t=mkdir
460 Create new directories as necessary to make sure that the named target
461 ($DIRCAP/SUBDIRS../SUBDIR) is a directory. This will create additional
462 intermediate mutable directories as necessary. If the named target directory
463 already exists, this will make no changes to it.
465 If the final directory is created, it will be empty.
467 This operation will return an error if a blocking file is present at any of
468 the parent names, preventing the server from creating the necessary parent
469 directory; or if it would require changing an immutable directory.
471 The write-cap of the new directory will be returned as the HTTP response
474 POST /uri/$DIRCAP/[SUBDIRS../]SUBDIR?t=mkdir-with-children
476 Like /uri?t=mkdir-with-children, but the final directory is created as a
477 child of an existing mutable directory. This will create additional
478 intermediate mutable directories as necessary. If the final directory is
479 created, it will be populated with initial children from the POST request
480 body, as described above.
482 This operation will return an error if a blocking file is present at any of
483 the parent names, preventing the server from creating the necessary parent
484 directory; or if it would require changing an immutable directory; or if
485 the immediate parent directory already has a a child named SUBDIR.
487 POST /uri/$DIRCAP/[SUBDIRS../]SUBDIR?t=mkdir-immutable
489 Like /uri?t=mkdir-immutable, but the final directory is created as a child
490 of an existing mutable directory. The final directory will be deep-immutable,
491 and will be populated with the children specified as a JSON dictionary in
492 the POST request body.
494 In Tahoe 1.6 this operation creates intermediate mutable directories if
495 necessary, but that behaviour should not be relied on; see ticket #920.
497 This operation will return an error if the parent directory is immutable,
498 or already has a child named SUBDIR.
500 POST /uri/$DIRCAP/[SUBDIRS../]?t=mkdir&name=NAME
502 Create a new empty mutable directory and attach it to the given existing
503 directory. This will create additional intermediate directories as necessary.
505 This operation will return an error if a blocking file is present at any of
506 the parent names, preventing the server from creating the necessary parent
507 directory, or if it would require changing any immutable directory.
509 The URL of this operation points to the parent of the bottommost new directory,
510 whereas the /uri/$DIRCAP/[SUBDIRS../]SUBDIR?t=mkdir operation above has a URL
511 that points directly to the bottommost new directory.
513 POST /uri/$DIRCAP/[SUBDIRS../]?t=mkdir-with-children&name=NAME
515 Like /uri/$DIRCAP/[SUBDIRS../]?t=mkdir&name=NAME, but the new directory will
516 be populated with initial children via the POST request body. This command
517 will create additional intermediate mutable directories as necessary.
519 This operation will return an error if a blocking file is present at any of
520 the parent names, preventing the server from creating the necessary parent
521 directory; or if it would require changing an immutable directory; or if
522 the immediate parent directory already has a a child named NAME.
524 Note that the name= argument must be passed as a queryarg, because the POST
525 request body is used for the initial children JSON.
527 POST /uri/$DIRCAP/[SUBDIRS../]?t=mkdir-immutable&name=NAME
529 Like /uri/$DIRCAP/[SUBDIRS../]?t=mkdir-with-children&name=NAME, but the
530 final directory will be deep-immutable. The children are specified as a
531 JSON dictionary in the POST request body. Again, the name= argument must be
532 passed as a queryarg.
534 In Tahoe 1.6 this operation creates intermediate mutable directories if
535 necessary, but that behaviour should not be relied on; see ticket #920.
537 This operation will return an error if the parent directory is immutable,
538 or already has a child named NAME.
540 === Get Information About A File Or Directory (as JSON) ===
542 GET /uri/$FILECAP?t=json
543 GET /uri/$DIRCAP?t=json
544 GET /uri/$DIRCAP/[SUBDIRS../]SUBDIR?t=json
545 GET /uri/$DIRCAP/[SUBDIRS../]FILENAME?t=json
547 This returns a machine-parseable JSON-encoded description of the given
548 object. The JSON always contains a list, and the first element of the list is
549 always a flag that indicates whether the referenced object is a file or a
550 directory. If it is a capability to a file, then the information includes
551 file size and URI, like this:
553 GET /uri/$FILECAP?t=json :
557 "verify_uri": verify_uri,
562 If it is a capability to a directory followed by a path from that directory
563 to a file, then the information also includes metadata from the link to the
564 file in the parent directory, like this:
566 GET /uri/$DIRCAP/[SUBDIRS../]FILENAME?t=json :
570 "verify_uri": verify_uri,
574 "ctime": 1202777696.7564139,
575 "mtime": 1202777696.7564139,
577 "linkcrtime": 1202777696.7564139,
578 "linkmotime": 1202777696.7564139,
581 If it is a directory, then it includes information about the children of
582 this directory, as a mapping from child name to a set of data about the
583 child (the same data that would appear in a corresponding GET?t=json of the
584 child itself). The child entries also include metadata about each child,
585 including link-creation- and link-change- timestamps. The output looks like
588 GET /uri/$DIRCAP?t=json :
589 GET /uri/$DIRCAP/[SUBDIRS../]SUBDIR?t=json :
592 "rw_uri": read_write_uri,
593 "ro_uri": read_only_uri,
594 "verify_uri": verify_uri,
597 "foo.txt": [ "filenode", {
601 "ctime": 1202777696.7564139,
602 "mtime": 1202777696.7564139,
604 "linkcrtime": 1202777696.7564139,
605 "linkmotime": 1202777696.7564139,
607 "subdir": [ "dirnode", {
611 "ctime": 1202778102.7589991,
612 "mtime": 1202778111.2160511,
614 "linkcrtime": 1202777696.7564139,
615 "linkmotime": 1202777696.7564139,
619 In the above example, note how 'children' is a dictionary in which the keys
620 are child names and the values depend upon whether the child is a file or a
621 directory. The value is mostly the same as the JSON representation of the
622 child object (except that directories do not recurse -- the "children"
623 entry of the child is omitted, and the directory view includes the metadata
624 that is stored on the directory edge).
626 Then the rw_uri field will be present in the information about a directory
627 if and only if you have read-write access to that directory. The verify_uri
628 field will be presend if and only if the object has a verify-cap
629 (non-distributed LIT files do not have verify-caps).
631 ==== About the metadata ====
633 The value of the 'mtime' key and of the 'tahoe':'linkmotime' is updated
634 whenever a link to a child is set. The value of the 'ctime' key and of the
635 'tahoe':'linkcrtime' key is updated whenever a link to a child is created --
636 i.e. when there was not previously a link under that name.
638 In Tahoe earlier than v1.4.0, only the 'mtime'/'ctime' keys were populated.
639 Starting in Tahoe v1.4.0, the 'linkmotime'/'linkcrtime' keys in the 'tahoe'
640 sub-dict are also populated.
642 The reason we added the new values in Tahoe v1.4.0 is that there is a
643 "set_children" API (described below) which you can use to overwrite the
644 values of the 'mtime'/'ctime' pair, and this API is used by the "tahoe
645 backup" command (both in Tahoe v1.3.0 and in Tahoe v1.4.0) to set the
646 'mtime' and 'ctime' values when backing up files from a local filesystem
647 into the Tahoe filesystem. As of Tahoe v1.4.0, the set_children API cannot
648 be used to set anything under the 'tahoe' key of the metadata dict -- if
649 you include 'tahoe' keys in your 'metadata' arguments then it will silently
652 Therefore, if the 'tahoe' sub-dict is present, you can rely on the
653 'linkcrtime' and 'linkmotime' values therein to have the semantics described
654 above. (This is assuming that only official Tahoe clients have been used to
655 write those links, and that their system clocks were set to what you expected
656 -- there is nothing preventing someone from editing their Tahoe client or
657 writing their own Tahoe client which would overwrite those values however
658 they like, and there is nothing to constrain their system clock from taking
661 The meaning of the 'ctime'/'mtime' fields are slightly more complex.
663 The meaning of the 'mtime' field is: whenever the edge is updated (by an HTTP
664 PUT or POST, as is done by the "tahoe cp" command), then the mtime is set to
665 the current time on the clock of the updating client. Whenever the edge is
666 updated by "tahoe backup" then the mtime is instead set to the value which
667 the updating client read from its local filesystem for the "mtime" of the
668 local file in question, which means the last time the contents of that file
669 were changed. Note however, that if the edge in the Tahoe filesystem points
670 to a mutable file and the contents of that mutable file is changed then the
671 "mtime" value on that edge will *not* be updated, since the edge itself
672 wasn't updated -- only the mutable file was.
674 The meaning of the 'ctime' field is even more complex. Whenever a new edge is
675 created (by an HTTP PUT or POST, as is done by "tahoe cp") then the ctime is
676 set to the current time on the clock of the updating client. Whenever the
677 edge is created *or updated* by "tahoe backup" then the ctime is instead set
678 to the value which the updating client read from its local filesystem. On
679 Windows, it reads the timestamp of when the local file was created and puts
680 that into the "ctime", and on other platforms it reads the timestamp of the
681 most recent time that either the contents or the metadata of the local file
682 was changed and puts that into the ctime. Again, if the edge points to a
683 mutable file and the content of that mutable file is changed then the ctime
684 will not be updated in any case.
686 Therefore there are several ways that the 'ctime' field could be confusing:
688 1. You might be confused about whether it reflects the time of the creation
689 of a link in the Tahoe filesystem or a timestamp copied in from a local
692 2. You might be confused about whether it is a copy of the file creation time
693 (if "tahoe backup" was run on a Windows system) or of the last
694 contents-or-metadata change (if "tahoe backup" was run on a different
697 3. You might be confused by the fact that changing the contents of a mutable
698 file in Tahoe don't have any effect on any links pointing at that file in any
699 directories, although "tahoe backup" sets the link 'ctime'/'mtime' to reflect
700 timestamps about the local file corresponding to the Tahoe file to which the
703 4. Also, quite apart from Tahoe, you might be confused about the meaning of
704 the 'ctime' in unix local filesystems, which people sometimes think means
705 file creation time, but which actually means, in unix local filesystems, the
706 most recent time that the file contents or the file metadata (such as owner,
707 permission bits, extended attributes, etc.) has changed. Note that although
708 'ctime' does not mean file creation time in Unix, it does mean link creation
709 time in Tahoe, unless the "tahoe backup" command has been used on that link,
710 in which case it means something about the local filesystem file which
711 corresponds to the Tahoe file which is pointed at by the link. It means
712 either file creation time of the local file (if "tahoe backup" was run on
713 Windows) or file-contents-or-metadata-update-time of the local file (if
714 "tahoe backup" was run on a different operating system).
717 === Attaching an existing File or Directory by its read- or write- cap ===
719 PUT /uri/$DIRCAP/[SUBDIRS../]CHILDNAME?t=uri
721 This attaches a child object (either a file or directory) to a specified
722 location in the virtual filesystem. The child object is referenced by its
723 read- or write- cap, as provided in the HTTP request body. This will create
724 intermediate directories as necessary.
726 This is similar to a UNIX hardlink: by referencing a previously-uploaded file
727 (or previously-created directory) instead of uploading/creating a new one,
728 you can create two references to the same object.
730 The read- or write- cap of the child is provided in the body of the HTTP
731 request, and this same cap is returned in the response body.
733 The default behavior is to overwrite any existing object at the same
734 location. To prevent this (and make the operation return an error instead
735 of overwriting), add a "replace=false" argument, as "?t=uri&replace=false".
736 With replace=false, this operation will return an HTTP 409 "Conflict" error
737 if there is already an object at the given location, rather than
738 overwriting the existing object. To allow the operation to overwrite a
739 file, but return an error when trying to overwrite a directory, use
740 "replace=only-files" (this behavior is closer to the traditional unix "mv"
741 command). Note that "true", "t", and "1" are all synonyms for "True", and
742 "false", "f", and "0" are synonyms for "False", and the parameter is
745 === Adding multiple files or directories to a parent directory at once ===
747 POST /uri/$DIRCAP/[SUBDIRS..]?t=set_children
748 POST /uri/$DIRCAP/[SUBDIRS..]?t=set-children (Tahoe >= v1.6)
750 This command adds multiple children to a directory in a single operation.
751 It reads the request body and interprets it as a JSON-encoded description
752 of the child names and read/write-caps that should be added.
754 The body should be a JSON-encoded dictionary, in the same format as the
755 "children" value returned by the "GET /uri/$DIRCAP?t=json" operation
756 described above. In this format, each key is a child names, and the
757 corresponding value is a tuple of (type, childinfo). "type" is ignored, and
758 "childinfo" is a dictionary that contains "rw_uri", "ro_uri", and
759 "metadata" keys. You can take the output of "GET /uri/$DIRCAP1?t=json" and
760 use it as the input to "POST /uri/$DIRCAP2?t=set_children" to make DIR2
761 look very much like DIR1 (except for any existing children of DIR2 that
762 were not overwritten, and any existing "tahoe" metadata keys as described
765 When the set_children request contains a child name that already exists in
766 the target directory, this command defaults to overwriting that child with
767 the new value (both child cap and metadata, but if the JSON data does not
768 contain a "metadata" key, the old child's metadata is preserved). The
769 command takes a boolean "overwrite=" query argument to control this
770 behavior. If you use "?t=set_children&overwrite=false", then an attempt to
771 replace an existing child will instead cause an error.
773 Any "tahoe" key in the new child's "metadata" value is ignored. Any
774 existing "tahoe" metadata is preserved. The metadata["tahoe"] value is
775 reserved for metadata generated by the tahoe node itself. The only two keys
776 currently placed here are "linkcrtime" and "linkmotime". For details, see
777 the section above entitled "Get Information About A File Or Directory (as
778 JSON)", in the "About the metadata" subsection.
780 Note that this command was introduced with the name "set_children", which
781 uses an underscore rather than a hyphen as other multi-word command names
782 do. The variant with a hyphen is now accepted, but clients that desire
783 backward compatibility should continue to use "set_children".
786 === Deleting a File or Directory ===
788 DELETE /uri/$DIRCAP/[SUBDIRS../]CHILDNAME
790 This removes the given name from its parent directory. CHILDNAME is the
791 name to be removed, and $DIRCAP/SUBDIRS.. indicates the directory that will
794 Note that this does not actually delete the file or directory that the name
795 points to from the tahoe grid -- it only removes the named reference from
796 this directory. If there are other names in this directory or in other
797 directories that point to the resource, then it will remain accessible
798 through those paths. Even if all names pointing to this object are removed
799 from their parent directories, then someone with possession of its read-cap
800 can continue to access the object through that cap.
802 The object will only become completely unreachable once 1: there are no
803 reachable directories that reference it, and 2: nobody is holding a read-
804 or write- cap to the object. (This behavior is very similar to the way
805 hardlinks and anonymous files work in traditional unix filesystems).
807 This operation will not modify more than a single directory. Intermediate
808 directories which were implicitly created by PUT or POST methods will *not*
809 be automatically removed by DELETE.
811 This method returns the file- or directory- cap of the object that was just
814 == Browser Operations ==
816 This section describes the HTTP operations that provide support for humans
817 running a web browser. Most of these operations use HTML forms that use POST
818 to drive the Tahoe node. This section is intended for HTML authors who want
819 to write web pages that contain forms and buttons which manipulate the Tahoe
822 Note that for all POST operations, the arguments listed can be provided
823 either as URL query arguments or as form body fields. URL query arguments are
824 separated from the main URL by "?", and from each other by "&". For example,
825 "POST /uri/$DIRCAP?t=upload&mutable=true". Form body fields are usually
826 specified by using <input type="hidden"> elements. For clarity, the
827 descriptions below display the most significant arguments as URL query args.
829 === Viewing A Directory (as HTML) ===
831 GET /uri/$DIRCAP/[SUBDIRS../]
833 This returns an HTML page, intended to be displayed to a human by a web
834 browser, which contains HREF links to all files and directories reachable
835 from this directory. These HREF links do not have a t= argument, meaning
836 that a human who follows them will get pages also meant for a human. It also
837 contains forms to upload new files, and to delete files and directories.
838 Those forms use POST methods to do their job.
840 === Viewing/Downloading a File ===
843 GET /uri/$DIRCAP/[SUBDIRS../]FILENAME
845 This will retrieve the contents of the given file. The HTTP response body
846 will contain the sequence of bytes that make up the file.
848 If you want the HTTP response to include a useful Content-Type header,
849 either use the second form (which starts with a $DIRCAP), or add a
850 "filename=foo" query argument, like "GET /uri/$FILECAP?filename=foo.jpg".
851 The bare "GET /uri/$FILECAP" does not give the Tahoe node enough information
852 to determine a Content-Type (since Tahoe immutable files are merely
853 sequences of bytes, not typed+named file objects).
855 If the URL has both filename= and "save=true" in the query arguments, then
856 the server to add a "Content-Disposition: attachment" header, along with a
857 filename= parameter. When a user clicks on such a link, most browsers will
858 offer to let the user save the file instead of displaying it inline (indeed,
859 most browsers will refuse to display it inline). "true", "t", "1", and other
860 case-insensitive equivalents are all treated the same.
862 Character-set handling in URLs and HTTP headers is a dubious art[1]. For
863 maximum compatibility, Tahoe simply copies the bytes from the filename=
864 argument into the Content-Disposition header's filename= parameter, without
865 trying to interpret them in any particular way.
868 GET /named/$FILECAP/FILENAME
870 This is an alternate download form which makes it easier to get the correct
871 filename. The Tahoe server will provide the contents of the given file, with
872 a Content-Type header derived from the given filename. This form is used to
873 get browsers to use the "Save Link As" feature correctly, and also helps
874 command-line tools like "wget" and "curl" use the right filename. Note that
875 this form can *only* be used with file caps; it is an error to use a
876 directory cap after the /named/ prefix.
878 === Get Information About A File Or Directory (as HTML) ===
880 GET /uri/$FILECAP?t=info
881 GET /uri/$DIRCAP/?t=info
882 GET /uri/$DIRCAP/[SUBDIRS../]SUBDIR/?t=info
883 GET /uri/$DIRCAP/[SUBDIRS../]FILENAME?t=info
885 This returns a human-oriented HTML page with more detail about the selected
886 file or directory object. This page contains the following items:
891 raw contents (text/plain)
892 access caps (URIs): verify-cap, read-cap, write-cap (for mutable objects)
893 check/verify/repair form
894 deep-check/deep-size/deep-stats/manifest (for directories)
895 replace-conents form (for mutable files)
897 === Creating a Directory ===
901 This creates a new empty directory, but does not attach it to the virtual
904 If a "redirect_to_result=true" argument is provided, then the HTTP response
905 will cause the web browser to be redirected to a /uri/$DIRCAP page that
906 gives access to the newly-created directory. If you bookmark this page,
907 you'll be able to get back to the directory again in the future. This is the
908 recommended way to start working with a Tahoe server: create a new unlinked
909 directory (using redirect_to_result=true), then bookmark the resulting
910 /uri/$DIRCAP page. There is a "create directory" button on the Welcome page
911 to invoke this action.
913 If "redirect_to_result=true" is not provided (or is given a value of
914 "false"), then the HTTP response body will simply be the write-cap of the
917 POST /uri/$DIRCAP/[SUBDIRS../]?t=mkdir&name=CHILDNAME
919 This creates a new empty directory as a child of the designated SUBDIR. This
920 will create additional intermediate directories as necessary.
922 If a "when_done=URL" argument is provided, the HTTP response will cause the
923 web browser to redirect to the given URL. This provides a convenient way to
924 return the browser to the directory that was just modified. Without a
925 when_done= argument, the HTTP response will simply contain the write-cap of
926 the directory that was just created.
929 === Uploading a File ===
933 This uploads a file, and produces a file-cap for the contents, but does not
934 attach the file into the virtual drive. No directories will be modified by
937 The file must be provided as the "file" field of an HTML encoded form body,
938 produced in response to an HTML form like this:
939 <form action="/uri" method="POST" enctype="multipart/form-data">
940 <input type="hidden" name="t" value="upload" />
941 <input type="file" name="file" />
942 <input type="submit" value="Upload Unlinked" />
945 If a "when_done=URL" argument is provided, the response body will cause the
946 browser to redirect to the given URL. If the when_done= URL has the string
947 "%(uri)s" in it, that string will be replaced by a URL-escaped form of the
948 newly created file-cap. (Note that without this substitution, there is no
949 way to access the file that was just uploaded).
951 The default (in the absence of when_done=) is to return an HTML page that
952 describes the results of the upload. This page will contain information
953 about which storage servers were used for the upload, how long each
956 If a "mutable=true" argument is provided, the operation will create a
957 mutable file, and the response body will contain the write-cap instead of
958 the upload results page. The default is to create an immutable file,
959 returning the upload results page as a response.
962 POST /uri/$DIRCAP/[SUBDIRS../]?t=upload
964 This uploads a file, and attaches it as a new child of the given directory,
965 which must be mutable. The file must be provided as the "file" field of an
966 HTML-encoded form body, produced in response to an HTML form like this:
967 <form action="." method="POST" enctype="multipart/form-data">
968 <input type="hidden" name="t" value="upload" />
969 <input type="file" name="file" />
970 <input type="submit" value="Upload" />
973 A "name=" argument can be provided to specify the new child's name,
974 otherwise it will be taken from the "filename" field of the upload form
975 (most web browsers will copy the last component of the original file's
976 pathname into this field). To avoid confusion, name= is not allowed to
979 If there is already a child with that name, and it is a mutable file, then
980 its contents are replaced with the data being uploaded. If it is not a
981 mutable file, the default behavior is to remove the existing child before
982 creating a new one. To prevent this (and make the operation return an error
983 instead of overwriting the old child), add a "replace=false" argument, as
984 "?t=upload&replace=false". With replace=false, this operation will return an
985 HTTP 409 "Conflict" error if there is already an object at the given
986 location, rather than overwriting the existing object. Note that "true",
987 "t", and "1" are all synonyms for "True", and "false", "f", and "0" are
988 synonyms for "False". the parameter is case-insensitive.
990 This will create additional intermediate directories as necessary, although
991 since it is expected to be triggered by a form that was retrieved by "GET
992 /uri/$DIRCAP/[SUBDIRS../]", it is likely that the parent directory will
995 If a "mutable=true" argument is provided, any new file that is created will
996 be a mutable file instead of an immutable one. <input type="checkbox"
997 name="mutable" /> will give the user a way to set this option.
999 If a "when_done=URL" argument is provided, the HTTP response will cause the
1000 web browser to redirect to the given URL. This provides a convenient way to
1001 return the browser to the directory that was just modified. Without a
1002 when_done= argument, the HTTP response will simply contain the file-cap of
1003 the file that was just uploaded (a write-cap for mutable files, or a
1004 read-cap for immutable files).
1006 POST /uri/$DIRCAP/[SUBDIRS../]FILENAME?t=upload
1008 This also uploads a file and attaches it as a new child of the given
1009 directory, which must be mutable. It is a slight variant of the previous
1010 operation, as the URL refers to the target file rather than the parent
1011 directory. It is otherwise identical: this accepts mutable= and when_done=
1014 POST /uri/$FILECAP?t=upload
1016 This modifies the contents of an existing mutable file in-place. An error is
1017 signalled if $FILECAP does not refer to a mutable file. It behaves just like
1018 the "PUT /uri/$FILECAP" form, but uses a POST for the benefit of HTML forms
1021 === Attaching An Existing File Or Directory (by URI) ===
1023 POST /uri/$DIRCAP/[SUBDIRS../]?t=uri&name=CHILDNAME&uri=CHILDCAP
1025 This attaches a given read- or write- cap "CHILDCAP" to the designated
1026 directory, with a specified child name. This behaves much like the PUT t=uri
1027 operation, and is a lot like a UNIX hardlink.
1029 This will create additional intermediate directories as necessary, although
1030 since it is expected to be triggered by a form that was retrieved by "GET
1031 /uri/$DIRCAP/[SUBDIRS../]", it is likely that the parent directory will
1034 This accepts the same replace= argument as POST t=upload.
1036 === Deleting A Child ===
1038 POST /uri/$DIRCAP/[SUBDIRS../]?t=delete&name=CHILDNAME
1040 This instructs the node to remove a child object (file or subdirectory) from
1041 the given directory, which must be mutable. Note that the entire subtree is
1042 unlinked from the parent. Unlike deleting a subdirectory in a UNIX local
1043 filesystem, the subtree need not be empty; if it isn't, then other references
1044 into the subtree will see that the child subdirectories are not modified by
1045 this operation. Only the link from the given directory to its child is severed.
1047 === Renaming A Child ===
1049 POST /uri/$DIRCAP/[SUBDIRS../]?t=rename&from_name=OLD&to_name=NEW
1051 This instructs the node to rename a child of the given directory, which must
1052 be mutable. This has a similar effect to removing the child, then adding the
1053 same child-cap under the new name, except that it preserves metadata. This
1054 operation cannot move the child to a different directory.
1056 This operation will replace any existing child of the new name, making it
1057 behave like the UNIX "mv -f" command.
1059 === Other Utilities ===
1063 This causes a redirect to /uri/$CAP, and retains any additional query
1064 arguments (like filename= or save=). This is for the convenience of web
1065 forms which allow the user to paste in a read- or write- cap (obtained
1066 through some out-of-band channel, like IM or email).
1068 Note that this form merely redirects to the specific file or directory
1069 indicated by the $CAP: unlike the GET /uri/$DIRCAP form, you cannot
1070 traverse to children by appending additional path segments to the URL.
1072 GET /uri/$DIRCAP/[SUBDIRS../]?t=rename-form&name=$CHILDNAME
1074 This provides a useful facility to browser-based user interfaces. It
1075 returns a page containing a form targetting the "POST $DIRCAP t=rename"
1076 functionality described above, with the provided $CHILDNAME present in the
1077 'from_name' field of that form. I.e. this presents a form offering to
1078 rename $CHILDNAME, requesting the new name, and submitting POST rename.
1080 GET /uri/$DIRCAP/[SUBDIRS../]CHILDNAME?t=uri
1082 This returns the file- or directory- cap for the specified object.
1084 GET /uri/$DIRCAP/[SUBDIRS../]CHILDNAME?t=readonly-uri
1086 This returns a read-only file- or directory- cap for the specified object.
1087 If the object is an immutable file, this will return the same value as
1090 === Debugging and Testing Features ===
1092 These URLs are less-likely to be helpful to the casual Tahoe user, and are
1093 mainly intended for developers.
1097 This triggers the FileChecker to determine the current "health" of the
1098 given file or directory, by counting how many shares are available. The
1099 page that is returned will display the results. This can be used as a "show
1100 me detailed information about this file" page.
1102 If a verify=true argument is provided, the node will perform a more
1103 intensive check, downloading and verifying every single bit of every share.
1105 If an add-lease=true argument is provided, the node will also add (or
1106 renew) a lease to every share it encounters. Each lease will keep the share
1107 alive for a certain period of time (one month by default). Once the last
1108 lease expires or is explicitly cancelled, the storage server is allowed to
1111 If an output=JSON argument is provided, the response will be
1112 machine-readable JSON instead of human-oriented HTML. The data is a
1113 dictionary with the following keys:
1115 storage-index: a base32-encoded string with the objects's storage index,
1116 or an empty string for LIT files
1117 summary: a string, with a one-line summary of the stats of the file
1118 results: a dictionary that describes the state of the file. For LIT files,
1119 this dictionary has only the 'healthy' key, which will always be
1120 True. For distributed files, this dictionary has the following
1122 count-shares-good: the number of good shares that were found
1123 count-shares-needed: 'k', the number of shares required for recovery
1124 count-shares-expected: 'N', the number of total shares generated
1125 count-good-share-hosts: the number of distinct storage servers with
1126 good shares. If this number is less than
1127 count-shares-good, then some shares are doubled
1128 up, increasing the correlation of failures. This
1129 indicates that one or more shares should be
1130 moved to an otherwise unused server, if one is
1132 count-wrong-shares: for mutable files, the number of shares for
1133 versions other than the 'best' one (highest
1134 sequence number, highest roothash). These are
1136 count-recoverable-versions: for mutable files, the number of
1137 recoverable versions of the file. For
1138 a healthy file, this will equal 1.
1139 count-unrecoverable-versions: for mutable files, the number of
1140 unrecoverable versions of the file.
1141 For a healthy file, this will be 0.
1142 count-corrupt-shares: the number of shares with integrity failures
1143 list-corrupt-shares: a list of "share locators", one for each share
1144 that was found to be corrupt. Each share locator
1145 is a list of (serverid, storage_index, sharenum).
1146 needs-rebalancing: (bool) True if there are multiple shares on a single
1147 storage server, indicating a reduction in reliability
1148 that could be resolved by moving shares to new
1150 servers-responding: list of base32-encoded storage server identifiers,
1151 one for each server which responded to the share
1153 healthy: (bool) True if the file is completely healthy, False otherwise.
1154 Healthy files have at least N good shares. Overlapping shares
1155 (indicated by count-good-share-hosts < count-shares-good) do not
1156 currently cause a file to be marked unhealthy. If there are at
1157 least N good shares, then corrupt shares do not cause the file to
1158 be marked unhealthy, although the corrupt shares will be listed
1159 in the results (list-corrupt-shares) and should be manually
1160 removed to wasting time in subsequent downloads (as the
1161 downloader rediscovers the corruption and uses alternate shares).
1162 sharemap: dict mapping share identifier to list of serverids
1163 (base32-encoded strings). This indicates which servers are
1164 holding which shares. For immutable files, the shareid is
1165 an integer (the share number, from 0 to N-1). For
1166 immutable files, it is a string of the form
1167 'seq%d-%s-sh%d', containing the sequence number, the
1168 roothash, and the share number.
1170 POST $URL?t=start-deep-check (must add &ophandle=XYZ)
1172 This initiates a recursive walk of all files and directories reachable from
1173 the target, performing a check on each one just like t=check. The result
1174 page will contain a summary of the results, including details on any
1175 file/directory that was not fully healthy.
1177 t=start-deep-check can only be invoked on a directory. An error (400
1178 BAD_REQUEST) will be signalled if it is invoked on a file. The recursive
1179 walker will deal with loops safely.
1181 This accepts the same verify= and add-lease= arguments as t=check.
1183 Since this operation can take a long time (perhaps a second per object),
1184 the ophandle= argument is required (see "Slow Operations, Progress, and
1185 Cancelling" above). The response to this POST will be a redirect to the
1186 corresponding /operations/$HANDLE page (with output=HTML or output=JSON to
1187 match the output= argument given to the POST). The deep-check operation
1188 will continue to run in the background, and the /operations page should be
1189 used to find out when the operation is done.
1191 Detailed check results for non-healthy files and directories will be
1192 available under /operations/$HANDLE/$STORAGEINDEX, and the HTML status will
1193 contain links to these detailed results.
1195 The HTML /operations/$HANDLE page for incomplete operations will contain a
1196 meta-refresh tag, set to 60 seconds, so that a browser which uses
1197 deep-check will automatically poll until the operation has completed.
1199 The JSON page (/options/$HANDLE?output=JSON) will contain a
1200 machine-readable JSON dictionary with the following keys:
1202 finished: a boolean, True if the operation is complete, else False. Some
1203 of the remaining keys may not be present until the operation
1205 root-storage-index: a base32-encoded string with the storage index of the
1206 starting point of the deep-check operation
1207 count-objects-checked: count of how many objects were checked. Note that
1208 non-distributed objects (i.e. small immutable LIT
1209 files) are not checked, since for these objects,
1210 the data is contained entirely in the URI.
1211 count-objects-healthy: how many of those objects were completely healthy
1212 count-objects-unhealthy: how many were damaged in some way
1213 count-corrupt-shares: how many shares were found to have corruption,
1214 summed over all objects examined
1215 list-corrupt-shares: a list of "share identifiers", one for each share
1216 that was found to be corrupt. Each share identifier
1217 is a list of (serverid, storage_index, sharenum).
1218 list-unhealthy-files: a list of (pathname, check-results) tuples, for
1219 each file that was not fully healthy. 'pathname' is
1220 a list of strings (which can be joined by "/"
1221 characters to turn it into a single string),
1222 relative to the directory on which deep-check was
1223 invoked. The 'check-results' field is the same as
1224 that returned by t=check&output=JSON, described
1226 stats: a dictionary with the same keys as the t=start-deep-stats command
1229 POST $URL?t=stream-deep-check
1231 This initiates a recursive walk of all files and directories reachable from
1232 the target, performing a check on each one just like t=check. For each
1233 unique object (duplicates are skipped), a single line of JSON is emitted to
1234 the HTTP response channel (or an error indication, see below). When the walk
1235 is complete, a final line of JSON is emitted which contains the accumulated
1236 file-size/count "deep-stats" data.
1238 This command takes the same arguments as t=start-deep-check.
1240 A CLI tool can split the response stream on newlines into "response units",
1241 and parse each response unit as JSON. Each such parsed unit will be a
1242 dictionary, and will contain at least the "type" key: a string, one of
1243 "file", "directory", or "stats".
1245 For all units that have a type of "file" or "directory", the dictionary will
1246 contain the following keys:
1248 "path": a list of strings, with the path that is traversed to reach the
1250 "cap": a write-cap URI for the file or directory, if available, else a
1252 "verifycap": a verify-cap URI for the file or directory
1253 "repaircap": an URI for the weakest cap that can still be used to repair
1255 "storage-index": a base32 storage index for the object
1256 "check-results": a copy of the dictionary which would be returned by
1257 t=check&output=json, with three top-level keys:
1258 "storage-index", "summary", and "results", and a variety
1259 of counts and sharemaps in the "results" value.
1261 Note that non-distributed files (i.e. LIT files) will have values of None
1262 for verifycap, repaircap, and storage-index, since these files can neither
1263 be verified nor repaired, and are not stored on the storage servers.
1264 Likewise the check-results dictionary will be limited: an empty string for
1265 storage-index, and a results dictionary with only the "healthy" key.
1267 The last unit in the stream will have a type of "stats", and will contain
1268 the keys described in the "start-deep-stats" operation, below.
1270 If any errors occur during the traversal (specifically if a directory is
1271 unrecoverable, such that further traversal is not possible), an error
1272 indication is written to the response body, instead of the usual line of
1273 JSON. This error indication line will begin with the string "ERROR:" (in all
1274 caps), and contain a summary of the error on the rest of the line. The
1275 remaining lines of the response body will be a python exception. The client
1276 application should look for the ERROR: and stop processing JSON as soon as
1277 it is seen. Note that neither a file being unrecoverable nor a directory
1278 merely being unhealthy will cause traversal to stop. The line just before
1279 the ERROR: will describe the directory that was untraversable, since the
1280 unit is emitted to the HTTP response body before the child is traversed.
1283 POST $URL?t=check&repair=true
1285 This performs a health check of the given file or directory, and if the
1286 checker determines that the object is not healthy (some shares are missing
1287 or corrupted), it will perform a "repair". During repair, any missing
1288 shares will be regenerated and uploaded to new servers.
1290 This accepts the same verify=true and add-lease= arguments as t=check. When
1291 an output=JSON argument is provided, the machine-readable JSON response
1292 will contain the following keys:
1294 storage-index: a base32-encoded string with the objects's storage index,
1295 or an empty string for LIT files
1296 repair-attempted: (bool) True if repair was attempted
1297 repair-successful: (bool) True if repair was attempted and the file was
1298 fully healthy afterwards. False if no repair was
1299 attempted, or if a repair attempt failed.
1300 pre-repair-results: a dictionary that describes the state of the file
1301 before any repair was performed. This contains exactly
1302 the same keys as the 'results' value of the t=check
1303 response, described above.
1304 post-repair-results: a dictionary that describes the state of the file
1305 after any repair was performed. If no repair was
1306 performed, post-repair-results and pre-repair-results
1307 will be the same. This contains exactly the same keys
1308 as the 'results' value of the t=check response,
1311 POST $URL?t=start-deep-check&repair=true (must add &ophandle=XYZ)
1313 This triggers a recursive walk of all files and directories, performing a
1314 t=check&repair=true on each one.
1316 Like t=start-deep-check without the repair= argument, this can only be
1317 invoked on a directory. An error (400 BAD_REQUEST) will be signalled if it
1318 is invoked on a file. The recursive walker will deal with loops safely.
1320 This accepts the same verify= and add-lease= arguments as
1321 t=start-deep-check. It uses the same ophandle= mechanism as
1322 start-deep-check. When an output=JSON argument is provided, the response
1323 will contain the following keys:
1325 finished: (bool) True if the operation has completed, else False
1326 root-storage-index: a base32-encoded string with the storage index of the
1327 starting point of the deep-check operation
1328 count-objects-checked: count of how many objects were checked
1330 count-objects-healthy-pre-repair: how many of those objects were completely
1331 healthy, before any repair
1332 count-objects-unhealthy-pre-repair: how many were damaged in some way
1333 count-objects-healthy-post-repair: how many of those objects were completely
1334 healthy, after any repair
1335 count-objects-unhealthy-post-repair: how many were damaged in some way
1337 count-repairs-attempted: repairs were attempted on this many objects.
1338 count-repairs-successful: how many repairs resulted in healthy objects
1339 count-repairs-unsuccessful: how many repairs resulted did not results in
1340 completely healthy objects
1341 count-corrupt-shares-pre-repair: how many shares were found to have
1342 corruption, summed over all objects
1343 examined, before any repair
1344 count-corrupt-shares-post-repair: how many shares were found to have
1345 corruption, summed over all objects
1346 examined, after any repair
1347 list-corrupt-shares: a list of "share identifiers", one for each share
1348 that was found to be corrupt (before any repair).
1349 Each share identifier is a list of (serverid,
1350 storage_index, sharenum).
1351 list-remaining-corrupt-shares: like list-corrupt-shares, but mutable shares
1352 that were successfully repaired are not
1353 included. These are shares that need
1354 manual processing. Since immutable shares
1355 cannot be modified by clients, all corruption
1356 in immutable shares will be listed here.
1357 list-unhealthy-files: a list of (pathname, check-results) tuples, for
1358 each file that was not fully healthy. 'pathname' is
1359 relative to the directory on which deep-check was
1360 invoked. The 'check-results' field is the same as
1361 that returned by t=check&repair=true&output=JSON,
1363 stats: a dictionary with the same keys as the t=start-deep-stats command
1366 POST $URL?t=stream-deep-check&repair=true
1368 This triggers a recursive walk of all files and directories, performing a
1369 t=check&repair=true on each one. For each unique object (duplicates are
1370 skipped), a single line of JSON is emitted to the HTTP response channel (or
1371 an error indication). When the walk is complete, a final line of JSON is
1372 emitted which contains the accumulated file-size/count "deep-stats" data.
1374 This emits the same data as t=stream-deep-check (without the repair=true),
1375 except that the "check-results" field is replaced with a
1376 "check-and-repair-results" field, which contains the keys returned by
1377 t=check&repair=true&output=json (i.e. repair-attempted, repair-successful,
1378 pre-repair-results, and post-repair-results). The output does not contain
1379 the summary dictionary that is provied by t=start-deep-check&repair=true
1380 (the one with count-objects-checked and list-unhealthy-files), since the
1381 receiving client is expected to calculate those values itself from the
1382 stream of per-object check-and-repair-results.
1384 Note that the "ERROR:" indication will only be emitted if traversal stops,
1385 which will only occur if an unrecoverable directory is encountered. If a
1386 file or directory repair fails, the traversal will continue, and the repair
1387 failure will be indicated in the JSON data (in the "repair-successful" key).
1389 POST $DIRURL?t=start-manifest (must add &ophandle=XYZ)
1391 This operation generates a "manfest" of the given directory tree, mostly
1392 for debugging. This is a table of (path, filecap/dircap), for every object
1393 reachable from the starting directory. The path will be slash-joined, and
1394 the filecap/dircap will contain a link to the object in question. This page
1395 gives immediate access to every object in the virtual filesystem subtree.
1397 This operation uses the same ophandle= mechanism as deep-check. The
1398 corresponding /operations/$HANDLE page has three different forms. The
1399 default is output=HTML.
1401 If output=text is added to the query args, the results will be a text/plain
1402 list. The first line is special: it is either "finished: yes" or "finished:
1403 no"; if the operation is not finished, you must periodically reload the
1404 page until it completes. The rest of the results are a plaintext list, with
1405 one file/dir per line, slash-separated, with the filecap/dircap separated
1408 If output=JSON is added to the queryargs, then the results will be a
1409 JSON-formatted dictionary with six keys. Note that because large directory
1410 structures can result in very large JSON results, the full results will not
1411 be available until the operation is complete (i.e. until output["finished"]
1414 finished (bool): if False then you must reload the page until True
1415 origin_si (base32 str): the storage index of the starting point
1416 manifest: list of (path, cap) tuples, where path is a list of strings.
1417 verifycaps: list of (printable) verify cap strings
1418 storage-index: list of (base32) storage index strings
1419 stats: a dictionary with the same keys as the t=start-deep-stats command
1422 POST $DIRURL?t=start-deep-size (must add &ophandle=XYZ)
1424 This operation generates a number (in bytes) containing the sum of the
1425 filesize of all directories and immutable files reachable from the given
1426 directory. This is a rough lower bound of the total space consumed by this
1427 subtree. It does not include space consumed by mutable files, nor does it
1428 take expansion or encoding overhead into account. Later versions of the
1429 code may improve this estimate upwards.
1431 The /operations/$HANDLE status output consists of two lines of text:
1436 POST $DIRURL?t=start-deep-stats (must add &ophandle=XYZ)
1438 This operation performs a recursive walk of all files and directories
1439 reachable from the given directory, and generates a collection of
1440 statistics about those objects.
1442 The result (obtained from the /operations/$OPHANDLE page) is a
1443 JSON-serialized dictionary with the following keys (note that some of these
1444 keys may be missing until 'finished' is True):
1446 finished: (bool) True if the operation has finished, else False
1447 count-immutable-files: count of how many CHK files are in the set
1448 count-mutable-files: same, for mutable files (does not include directories)
1449 count-literal-files: same, for LIT files (data contained inside the URI)
1450 count-files: sum of the above three
1451 count-directories: count of directories
1452 count-unknown: count of unrecognized objects (perhaps from the future)
1453 size-immutable-files: total bytes for all CHK files in the set, =deep-size
1454 size-mutable-files (TODO): same, for current version of all mutable files
1455 size-literal-files: same, for LIT files
1456 size-directories: size of directories (includes size-literal-files)
1457 size-files-histogram: list of (minsize, maxsize, count) buckets,
1458 with a histogram of filesizes, 5dB/bucket,
1459 for both literal and immutable files
1460 largest-directory: number of children in the largest directory
1461 largest-immutable-file: number of bytes in the largest CHK file
1463 size-mutable-files is not implemented, because it would require extra
1464 queries to each mutable file to get their size. This may be implemented in
1467 Assuming no sharing, the basic space consumed by a single root directory is
1468 the sum of size-immutable-files, size-mutable-files, and size-directories.
1469 The actual disk space used by the shares is larger, because of the
1470 following sources of overhead:
1473 expansion due to erasure coding
1474 share management data (leases)
1475 backend (ext3) minimum block size
1477 POST $URL?t=stream-manifest
1479 This operation performs a recursive walk of all files and directories
1480 reachable from the given starting point. For each such unique object
1481 (duplicates are skipped), a single line of JSON is emitted to the HTTP
1482 response channel (or an error indication, see below). When the walk is
1483 complete, a final line of JSON is emitted which contains the accumulated
1484 file-size/count "deep-stats" data.
1486 A CLI tool can split the response stream on newlines into "response units",
1487 and parse each response unit as JSON. Each such parsed unit will be a
1488 dictionary, and will contain at least the "type" key: a string, one of
1489 "file", "directory", or "stats".
1491 For all units that have a type of "file" or "directory", the dictionary will
1492 contain the following keys:
1494 "path": a list of strings, with the path that is traversed to reach the
1496 "cap": a write-cap URI for the file or directory, if available, else a
1498 "verifycap": a verify-cap URI for the file or directory
1499 "repaircap": an URI for the weakest cap that can still be used to repair
1501 "storage-index": a base32 storage index for the object
1503 Note that non-distributed files (i.e. LIT files) will have values of None
1504 for verifycap, repaircap, and storage-index, since these files can neither
1505 be verified nor repaired, and are not stored on the storage servers.
1507 The last unit in the stream will have a type of "stats", and will contain
1508 the keys described in the "start-deep-stats" operation, below.
1510 If any errors occur during the traversal (specifically if a directory is
1511 unrecoverable, such that further traversal is not possible), an error
1512 indication is written to the response body, instead of the usual line of
1513 JSON. This error indication line will begin with the string "ERROR:" (in all
1514 caps), and contain a summary of the error on the rest of the line. The
1515 remaining lines of the response body will be a python exception. The client
1516 application should look for the ERROR: and stop processing JSON as soon as
1517 it is seen. The line just before the ERROR: will describe the directory that
1518 was untraversable, since the manifest entry is emitted to the HTTP response
1519 body before the child is traversed.
1521 == Other Useful Pages ==
1523 The portion of the web namespace that begins with "/uri" (and "/named") is
1524 dedicated to giving users (both humans and programs) access to the Tahoe
1525 virtual filesystem. The rest of the namespace provides status information
1526 about the state of the Tahoe node.
1528 GET / (the root page)
1530 This is the "Welcome Page", and contains a few distinct sections:
1532 Node information: library versions, local nodeid, services being provided.
1534 Filesystem Access Forms: create a new directory, view a file/directory by
1535 URI, upload a file (unlinked), download a file by
1538 Grid Status: introducer information, helper information, connected storage
1543 This page lists all active uploads and downloads, and contains a short list
1544 of recent upload/download operations. Each operation has a link to a page
1545 that describes file sizes, servers that were involved, and the time consumed
1546 in each phase of the operation.
1548 A GET of /status/?t=json will contain a machine-readable subset of the same
1549 data. It returns a JSON-encoded dictionary. The only key defined at this
1550 time is "active", with a value that is a list of operation dictionaries, one
1551 for each active operation. Once an operation is completed, it will no longer
1552 appear in data["active"] .
1554 Each op-dict contains a "type" key, one of "upload", "download",
1555 "mapupdate", "publish", or "retrieve" (the first two are for immutable
1556 files, while the latter three are for mutable files and directories).
1558 The "upload" op-dict will contain the following keys:
1560 type (string): "upload"
1561 storage-index-string (string): a base32-encoded storage index
1562 total-size (int): total size of the file
1563 status (string): current status of the operation
1564 progress-hash (float): 1.0 when the file has been hashed
1565 progress-ciphertext (float): 1.0 when the file has been encrypted.
1566 progress-encode-push (float): 1.0 when the file has been encoded and
1567 pushed to the storage servers. For helper
1568 uploads, the ciphertext value climbs to 1.0
1569 first, then encoding starts. For unassisted
1570 uploads, ciphertext and encode-push progress
1571 will climb at the same pace.
1573 The "download" op-dict will contain the following keys:
1575 type (string): "download"
1576 storage-index-string (string): a base32-encoded storage index
1577 total-size (int): total size of the file
1578 status (string): current status of the operation
1579 progress (float): 1.0 when the file has been fully downloaded
1581 Front-ends which want to report progress information are advised to simply
1582 average together all the progress-* indicators. A slightly more accurate
1583 value can be found by ignoring the progress-hash value (since the current
1584 implementation hashes synchronously, so clients will probably never see
1585 progress-hash!=1.0).
1589 This page provides a basic tool to predict the likely storage and bandwidth
1590 requirements of a large Tahoe grid. It provides forms to input things like
1591 total number of users, number of files per user, average file size, number
1592 of servers, expansion ratio, hard drive failure rate, etc. It then provides
1593 numbers like how many disks per server will be needed, how many read
1594 operations per second should be expected, and the likely MTBF for files in
1595 the grid. This information is very preliminary, and the model upon which it
1596 is based still needs a lot of work.
1600 If the node is running a helper (i.e. if [helper]enabled is set to True in
1601 tahoe.cfg), then this page will provide a list of all the helper operations
1602 currently in progress. If "?t=json" is added to the URL, it will return a
1603 JSON-formatted list of helper statistics, which can then be used to produce
1604 graphs to indicate how busy the helper is.
1608 This page provides "node statistics", which are collected from a variety of
1611 load_monitor: every second, the node schedules a timer for one second in
1612 the future, then measures how late the subsequent callback
1613 is. The "load_average" is this tardiness, measured in
1614 seconds, averaged over the last minute. It is an indication
1615 of a busy node, one which is doing more work than can be
1616 completed in a timely fashion. The "max_load" value is the
1617 highest value that has been seen in the last 60 seconds.
1619 cpu_monitor: every minute, the node uses time.clock() to measure how much
1620 CPU time it has used, and it uses this value to produce
1621 1min/5min/15min moving averages. These values range from 0%
1622 (0.0) to 100% (1.0), and indicate what fraction of the CPU
1623 has been used by the Tahoe node. Not all operating systems
1624 provide meaningful data to time.clock(): they may report 100%
1625 CPU usage at all times.
1627 uploader: this counts how many immutable files (and bytes) have been
1628 uploaded since the node was started
1630 downloader: this counts how many immutable files have been downloaded
1631 since the node was started
1633 publishes: this counts how many mutable files (including directories) have
1634 been modified since the node was started
1636 retrieves: this counts how many mutable files (including directories) have
1637 been read since the node was started
1639 There are other statistics that are tracked by the node. The "raw stats"
1640 section shows a formatted dump of all of them.
1642 By adding "?t=json" to the URL, the node will return a JSON-formatted
1643 dictionary of stats values, which can be used by other tools to produce
1644 graphs of node behavior. The misc/munin/ directory in the source
1645 distribution provides some tools to produce these graphs.
1647 GET / (introducer status)
1649 For Introducer nodes, the welcome page displays information about both
1650 clients and servers which are connected to the introducer. Servers make
1651 "service announcements", and these are listed in a table. Clients will
1652 subscribe to hear about service announcements, and these subscriptions are
1653 listed in a separate table. Both tables contain information about what
1654 version of Tahoe is being run by the remote node, their advertised and
1655 outbound IP addresses, their nodeid and nickname, and how long they have
1658 By adding "?t=json" to the URL, the node will return a JSON-formatted
1659 dictionary of stats values, which can be used to produce graphs of connected
1660 clients over time. This dictionary has the following keys:
1662 ["subscription_summary"] : a dictionary mapping service name (like
1663 "storage") to an integer with the number of
1664 clients that have subscribed to hear about that
1666 ["announcement_summary"] : a dictionary mapping service name to an integer
1667 with the number of servers which are announcing
1669 ["announcement_distinct_hosts"] : a dictionary mapping service name to an
1670 integer which represents the number of
1671 distinct hosts that are providing that
1672 service. If two servers have announced
1673 FURLs which use the same hostnames (but
1674 different ports and tubids), they are
1675 considered to be on the same host.
1678 == Static Files in /public_html ==
1680 The wapi server will take any request for a URL that starts with /static
1681 and serve it from a configurable directory which defaults to
1682 $BASEDIR/public_html . This is configured by setting the "[node]web.static"
1683 value in $BASEDIR/tahoe.cfg . If this is left at the default value of
1684 "public_html", then http://localhost:3456/static/subdir/foo.html will be
1685 served with the contents of the file $BASEDIR/public_html/subdir/foo.html .
1687 This can be useful to serve a javascript application which provides a
1688 prettier front-end to the rest of the Tahoe wapi.
1691 == safety and security issues -- names vs. URIs ==
1693 Summary: use explicit file- and dir- caps whenever possible, to reduce the
1694 potential for surprises when the filesystem structure is changed.
1696 Tahoe provides a mutable filesystem, but the ways that the filesystem can
1697 change are limited. The only thing that can change is that the mapping from
1698 child names to child objects that each directory contains can be changed by
1699 adding a new child name pointing to an object, removing an existing child name,
1700 or changing an existing child name to point to a different object.
1702 Obviously if you query Tahoe for information about the filesystem and then act
1703 to change the filesystem (such as by getting a listing of the contents of a
1704 directory and then adding a file to the directory), then the filesystem might
1705 have been changed after you queried it and before you acted upon it. However,
1706 if you use the URI instead of the pathname of an object when you act upon the
1707 object, then the only change that can happen is if the object is a directory
1708 then the set of child names it has might be different. If, on the other hand,
1709 you act upon the object using its pathname, then a different object might be in
1710 that place, which can result in more kinds of surprises.
1712 For example, suppose you are writing code which recursively downloads the
1713 contents of a directory. The first thing your code does is fetch the listing
1714 of the contents of the directory. For each child that it fetched, if that
1715 child is a file then it downloads the file, and if that child is a directory
1716 then it recurses into that directory. Now, if the download and the recurse
1717 actions are performed using the child's name, then the results might be
1718 wrong, because for example a child name that pointed to a sub-directory when
1719 you listed the directory might have been changed to point to a file (in which
1720 case your attempt to recurse into it would result in an error and the file
1721 would be skipped), or a child name that pointed to a file when you listed the
1722 directory might now point to a sub-directory (in which case your attempt to
1723 download the child would result in a file containing HTML text describing the
1726 If your recursive algorithm uses the uri of the child instead of the name of
1727 the child, then those kinds of mistakes just can't happen. Note that both the
1728 child's name and the child's URI are included in the results of listing the
1729 parent directory, so it isn't any harder to use the URI for this purpose.
1731 The read and write caps in a given directory node are separate URIs, and
1732 can't be assumed to point to the same object even if they were retrieved in
1733 the same operation (although the webapi server attempts to ensure this
1734 in most cases). If you need to rely on that property, you should explicitly
1735 verify it. More generally, you should not make assumptions about the
1736 internal consistency of the contents of mutable directories. As a result
1737 of the signatures on mutable object versions, it is guaranteed that a given
1738 version was written in a single update, but -- as in the case of a file --
1739 the contents may have been chosen by a malicious writer in a way that is
1740 designed to confuse applications that rely on their consistency.
1742 In general, use names if you want "whatever object (whether file or
1743 directory) is found by following this name (or sequence of names) when my
1744 request reaches the server". Use URIs if you want "this particular object".
1746 == Concurrency Issues ==
1748 Tahoe uses both mutable and immutable files. Mutable files can be created
1749 explicitly by doing an upload with ?mutable=true added, or implicitly by
1750 creating a new directory (since a directory is just a special way to
1751 interpret a given mutable file).
1753 Mutable files suffer from the same consistency-vs-availability tradeoff that
1754 all distributed data storage systems face. It is not possible to
1755 simultaneously achieve perfect consistency and perfect availability in the
1756 face of network partitions (servers being unreachable or faulty).
1758 Tahoe tries to achieve a reasonable compromise, but there is a basic rule in
1759 place, known as the Prime Coordination Directive: "Don't Do That". What this
1760 means is that if write-access to a mutable file is available to several
1761 parties, then those parties are responsible for coordinating their activities
1762 to avoid multiple simultaneous updates. This could be achieved by having
1763 these parties talk to each other and using some sort of locking mechanism, or
1764 by serializing all changes through a single writer.
1766 The consequences of performing uncoordinated writes can vary. Some of the
1767 writers may lose their changes, as somebody else wins the race condition. In
1768 many cases the file will be left in an "unhealthy" state, meaning that there
1769 are not as many redundant shares as we would like (reducing the reliability
1770 of the file against server failures). In the worst case, the file can be left
1771 in such an unhealthy state that no version is recoverable, even the old ones.
1772 It is this small possibility of data loss that prompts us to issue the Prime
1773 Coordination Directive.
1775 Tahoe nodes implement internal serialization to make sure that a single Tahoe
1776 node cannot conflict with itself. For example, it is safe to issue two
1777 directory modification requests to a single tahoe node's wapi server at the
1778 same time, because the Tahoe node will internally delay one of them until
1779 after the other has finished being applied. (This feature was introduced in
1780 Tahoe-1.1; back with Tahoe-1.0 the web client was responsible for serializing
1781 web requests themselves).
1783 For more details, please see the "Consistency vs Availability" and "The Prime
1784 Coordination Directive" sections of mutable.txt, in the same directory as
1788 [1]: URLs and HTTP and UTF-8, Oh My
1790 HTTP does not provide a mechanism to specify the character set used to
1791 encode non-ascii names in URLs (rfc2396#2.1). We prefer the convention that
1792 the filename= argument shall be a URL-encoded UTF-8 encoded unicode object.
1793 For example, suppose we want to provoke the server into using a filename of
1794 "f i a n c e-acute e" (i.e. F I A N C U+00E9 E). The UTF-8 encoding of this
1795 is 0x66 0x69 0x61 0x6e 0x63 0xc3 0xa9 0x65 (or "fianc\xC3\xA9e", as python's
1796 repr() function would show). To encode this into a URL, the non-printable
1797 characters must be escaped with the urlencode '%XX' mechansim, giving us
1798 "fianc%C3%A9e". Thus, the first line of the HTTP request will be "GET
1799 /uri/CAP...?save=true&filename=fianc%C3%A9e HTTP/1.1". Not all browsers
1800 provide this: IE7 uses the Latin-1 encoding, which is fianc%E9e.
1802 The response header will need to indicate a non-ASCII filename. The actual
1803 mechanism to do this is not clear. For ASCII filenames, the response header
1806 Content-Disposition: attachment; filename="english.txt"
1808 If Tahoe were to enforce the utf-8 convention, it would need to decode the
1809 URL argument into a unicode string, and then encode it back into a sequence
1810 of bytes when creating the response header. One possibility would be to use
1811 unencoded utf-8. Developers suggest that IE7 might accept this:
1813 #1: Content-Disposition: attachment; filename="fianc\xC3\xA9e"
1814 (note, the last four bytes of that line, not including the newline, are
1815 0xC3 0xA9 0x65 0x22)
1817 RFC2231#4 (dated 1997): suggests that the following might work, and some
1818 developers (http://markmail.org/message/dsjyokgl7hv64ig3) have reported that
1819 it is supported by firefox (but not IE7):
1821 #2: Content-Disposition: attachment; filename*=utf-8''fianc%C3%A9e
1823 My reading of RFC2616#19.5.1 (which defines Content-Disposition) says that
1824 the filename= parameter is defined to be wrapped in quotes (presumeably to
1825 allow spaces without breaking the parsing of subsequent parameters), which
1828 #3: Content-Disposition: attachment; filename*=utf-8''"fianc%C3%A9e"
1830 However this is contrary to the examples in the email thread listed above.
1832 Developers report that IE7 (when it is configured for UTF-8 URL encoding,
1833 which is not the default in asian countries), will accept:
1835 #4: Content-Disposition: attachment; filename=fianc%C3%A9e
1837 However, for maximum compatibility, Tahoe simply copies bytes from the URL
1838 into the response header, rather than enforcing the utf-8 convention. This
1839 means it does not try to decode the filename from the URL argument, nor does
1840 it encode the filename into the response header.