ruby-manta 1.1.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/LICENSE +19 -0
- data/README.md +722 -0
- data/example.rb +78 -0
- data/lib/ruby-manta.rb +925 -0
- data/lib/version.rb +3 -0
- data/ruby-manta.gemspec +27 -0
- data/tests/test_ruby-manta.rb +653 -0
- metadata +80 -0
checksums.yaml
ADDED
@@ -0,0 +1,7 @@
|
|
1
|
+
---
|
2
|
+
SHA1:
|
3
|
+
metadata.gz: 5526525955be143def0033e1f1e72e5a57fe3aef
|
4
|
+
data.tar.gz: b4d4c6a1d8976375027b705e1484419cb4a5daf7
|
5
|
+
SHA512:
|
6
|
+
metadata.gz: 931e4b54947627df97a45131e712d10bc641138800d17b6c9fd57c1c137646816cb0709807289284aab9682b841bcb50db70ad58b2a4d31ac89dc27b0dfc4c1a
|
7
|
+
data.tar.gz: 2211861bc3ce04d358aeeb530ec7382e06de454b662f424e64ead06f2bc38c64159fd2fc674bab40534eeb4f753eea52ce209481d4cd179b9d12193e13d168c2
|
data/LICENSE
ADDED
@@ -0,0 +1,19 @@
|
|
1
|
+
The MIT License (MIT) Copyright (c) 2012 Joyent
|
2
|
+
|
3
|
+
Permission is hereby granted, free of charge, to any person obtaining a copy of
|
4
|
+
this software and associated documentation files (the "Software"), to deal in
|
5
|
+
the Software without restriction, including without limitation the rights to
|
6
|
+
use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
|
7
|
+
of the Software, and to permit persons to whom the Software is furnished to do
|
8
|
+
so, subject to the following conditions:
|
9
|
+
|
10
|
+
The above copyright notice and this permission notice shall be included in all
|
11
|
+
copies or substantial portions of the Software.
|
12
|
+
|
13
|
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
14
|
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
15
|
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
16
|
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
17
|
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
18
|
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
19
|
+
SOFTWARE.
|
data/README.md
ADDED
@@ -0,0 +1,722 @@
|
|
1
|
+
ruby-manta
|
2
|
+
==========
|
3
|
+
|
4
|
+
|
5
|
+
|
6
|
+
What's ruby-manta?
|
7
|
+
------------------
|
8
|
+
|
9
|
+
ruby-manta is a client for communicating with Manta. Manta is a RESTful service,
|
10
|
+
so ruby-manta is effectively an HTTP(S) wrapper which handles required HTTP
|
11
|
+
headers and performs some sanity checks. ruby-manta seeks to expose all of
|
12
|
+
Manta's features in a thin low-abstraction client.
|
13
|
+
|
14
|
+
|
15
|
+
|
16
|
+
What's Manta?
|
17
|
+
-------------
|
18
|
+
|
19
|
+
Manta is a RESTful replicated object store with a directory structure,
|
20
|
+
emphasizing Consistency and Partition Tolerance (C & P) in Brewer's CAP,
|
21
|
+
which differs from the usual Dynamo-derivative choice of Availability and
|
22
|
+
Partition Tolerance (A & P). This makes reasoning about Manta simpler (reads
|
23
|
+
after writes will _always_ return the newest version), and it supports CAS
|
24
|
+
semantics, but it also means that HTTP 500s may temporarily be returned for
|
25
|
+
some objects on rare occasion.
|
26
|
+
|
27
|
+
Manta also provides a map-reduce service operating over objects stored in
|
28
|
+
Manta. The most notable feature here is that map and reduce phases operate
|
29
|
+
fully within a UNIX environment. Standard utilities, language environments
|
30
|
+
(e.g. node.js, Python, Ruby, Perl), precompiled binaries (e.g. LuxRender,
|
31
|
+
BLAST, Postgres, your own custom binaries) and anything you can run in a
|
32
|
+
standard UNIX environment can be used.
|
33
|
+
|
34
|
+
|
35
|
+
|
36
|
+
Streaming Large Objects
|
37
|
+
-----------------------
|
38
|
+
|
39
|
+
One important limitation of ruby-manta is that it is not designed to handle
|
40
|
+
large objects. Specifically, when used to upload or download a large object,
|
41
|
+
that entire object will be loaded into Ruby's heap. If you're trying to move
|
42
|
+
multi-gigabyte objects using a sub-gigabyte VPS or zone, that won't work.
|
43
|
+
This leads to the following observations:
|
44
|
+
|
45
|
+
* don't upload or download large objects using ruby-manta
|
46
|
+
* if you must move large objects, consider compressing them first
|
47
|
+
* if they're still large, consider using node-manta instead.
|
48
|
+
|
49
|
+
Unlike ruby-manta, node-manta (the Node.js API for Manta) streams, so object
|
50
|
+
size is not a limitation. If you intend to work with large objects, use
|
51
|
+
node-manta instead. An slight alternative is to use node-manta for uploading
|
52
|
+
and downloading objects, and ruby-manta for everything else.
|
53
|
+
|
54
|
+
|
55
|
+
|
56
|
+
Example
|
57
|
+
-------
|
58
|
+
|
59
|
+
If you're like the author, examples are worth reams of explanation. Here,
|
60
|
+
hurried friend, is an example demonstrating some of ruby-manta's usage:
|
61
|
+
|
62
|
+
````` ruby
|
63
|
+
require 'ruby-manta'
|
64
|
+
|
65
|
+
# You'll need to provide these four environment variables to run this
|
66
|
+
# example. E.g.:
|
67
|
+
# USER=john KEY=~/.ssh/john HOST=https://us-east.manta.joyent.com DIR=. \
|
68
|
+
# ruby example.rb
|
69
|
+
host = ENV['HOST']
|
70
|
+
user = ENV['USER']
|
71
|
+
priv_key = ENV['KEY' ]
|
72
|
+
upload_dir = ENV['DIR' ]
|
73
|
+
|
74
|
+
# Read in private key, create a MantaClient instance. MantaClient is
|
75
|
+
# thread-safe and provides persistent connections with pooling, so you'll
|
76
|
+
# only ever need a single instance of this in a program.
|
77
|
+
priv_key_data = File.read(priv_key)
|
78
|
+
client = MantaClient.new(host, user, priv_key_data,
|
79
|
+
:disable_ssl_verification => true)
|
80
|
+
|
81
|
+
# Create an directory in Manta solely for this example run.
|
82
|
+
dir_path = '/' + user + '/stor/ruby-manta-example'
|
83
|
+
client.put_directory(dir_path)
|
84
|
+
|
85
|
+
# Upload files in a local directory to the Manta directory.
|
86
|
+
file_paths = Dir[upload_dir + '/*'].select { |p| File.file? p }
|
87
|
+
file_paths.each do |file_path|
|
88
|
+
file_name = File.basename(file_path)
|
89
|
+
# Be careful about binary files and file encodings in Ruby 1.9. If you don't
|
90
|
+
# use ASCII-8BIT (forced by 'rb' below), expect timeouts while PUTing an
|
91
|
+
# object.
|
92
|
+
file_data = File.open(file_path, 'rb') { |f| f.read }
|
93
|
+
client.put_object(dir_path + '/' + file_name, file_data)
|
94
|
+
end
|
95
|
+
|
96
|
+
# This example job runs the wc UNIX command on every object for the
|
97
|
+
# map phase, then uses awk during reduce to sum up the three numbers each wc
|
98
|
+
# returned.
|
99
|
+
job_details = {
|
100
|
+
:name => 'total word count',
|
101
|
+
:phases => [ {
|
102
|
+
:exec => 'wc'
|
103
|
+
}, {
|
104
|
+
:type => 'reduce',
|
105
|
+
:exec => "awk '{ l += $1; w += $2; c += $3 } END { print l, w, c }'"
|
106
|
+
} ]
|
107
|
+
}
|
108
|
+
|
109
|
+
# Create the job, then add the objects the job should operate on.
|
110
|
+
job_path, _ = client.create_job(job_details)
|
111
|
+
|
112
|
+
entries, _ = client.list_directory(dir_path)
|
113
|
+
obj_paths = entries.select { |e| e['type'] == 'object' }.
|
114
|
+
map { |e| dir_path + '/' + e['name'] }
|
115
|
+
|
116
|
+
client.add_job_keys(job_path, obj_paths)
|
117
|
+
|
118
|
+
# Tell Manta we're done adding objects to the job. Manta doesn't need this
|
119
|
+
# to start running a job -- you can see map results without it, for
|
120
|
+
# example -- but reduce phases in particular depend on all mapping
|
121
|
+
# finishing.
|
122
|
+
client.end_job_input(job_path)
|
123
|
+
|
124
|
+
# Poll until Manta finishes the job.
|
125
|
+
begin
|
126
|
+
sleep 1
|
127
|
+
job, _ = client.get_job(job_path)
|
128
|
+
end while job['state'] != 'done'
|
129
|
+
|
130
|
+
# We know in this case there will be only one result. Fetch it and
|
131
|
+
# display it.
|
132
|
+
results, _ = client.get_job_output(job_path)
|
133
|
+
data, _ = client.get_object(results[0])
|
134
|
+
puts data
|
135
|
+
|
136
|
+
# Clean up; remove objects and directory.
|
137
|
+
obj_paths.each do |obj_path|
|
138
|
+
client.delete_object(obj_path)
|
139
|
+
end
|
140
|
+
|
141
|
+
client.delete_directory(dir_path)
|
142
|
+
`````
|
143
|
+
|
144
|
+
|
145
|
+
NB: there's no catching of exceptions above! Real production code should
|
146
|
+
be prepared for the exceptional -- see "a note on semantics" below.
|
147
|
+
|
148
|
+
If you see a PEM pass phrase request, that's because you're using an
|
149
|
+
encrypted private key. In production on a server, you'd presumably use an
|
150
|
+
unencrypted key.
|
151
|
+
|
152
|
+
|
153
|
+
|
154
|
+
Installation
|
155
|
+
------------
|
156
|
+
|
157
|
+
If you're one of the chaps using Ruby 1.9.*, life is easy:
|
158
|
+
|
159
|
+
gem install ruby-manta-1.0.1.gem
|
160
|
+
|
161
|
+
Done.
|
162
|
+
|
163
|
+
Ruby 1.8.7 was end-of-life'd on June, 2013. As a result, ruby-manta no longer
|
164
|
+
supports it either.
|
165
|
+
|
166
|
+
|
167
|
+
|
168
|
+
Public and Private spaces
|
169
|
+
-------------------------
|
170
|
+
|
171
|
+
ruby-manta operates with Manta paths. An example of a path is:
|
172
|
+
|
173
|
+
/john/stor/image.png
|
174
|
+
|
175
|
+
This object, image.png, belongs to the user "john". It's in his private space,
|
176
|
+
"stor". But what if John wants to let the ~~unwashed hoi poll~~general public to
|
177
|
+
see his image.png? In this case there is also the "public" space:
|
178
|
+
|
179
|
+
/john/public/image.png
|
180
|
+
|
181
|
+
|
182
|
+
|
183
|
+
Signed URLs
|
184
|
+
-----------
|
185
|
+
|
186
|
+
Objects put in the public space are accessible by everyone. Objects in the
|
187
|
+
private space are only accessible by individuals authenticated and authorized
|
188
|
+
by Manta. Manta also supports temporary signed URLs that allow unauthenticated
|
189
|
+
individuals to operate on private objects, until the link expires. See
|
190
|
+
gen_signed_url() below for the details.
|
191
|
+
|
192
|
+
|
193
|
+
|
194
|
+
Map/Reduce Jobs
|
195
|
+
---------------
|
196
|
+
|
197
|
+
Alas, this is beyond the scope of this document. Please refer to Manta's
|
198
|
+
documentation for details on how to construct a job. ruby-manta passes job
|
199
|
+
details directly to Manta, so what you see is what you'll get.
|
200
|
+
|
201
|
+
Short summary: create a job, add paths to objects the job should operate on,
|
202
|
+
then close the job. Poll until Manta tells you the job is finished, then peek
|
203
|
+
at the resulting objects.
|
204
|
+
|
205
|
+
|
206
|
+
|
207
|
+
The API
|
208
|
+
=======
|
209
|
+
|
210
|
+
A note on sematics
|
211
|
+
------------------
|
212
|
+
|
213
|
+
All methods throw exceptions upon failure. If a method doesn't throw, the
|
214
|
+
returned results are valid. The most common category of failure you'll see
|
215
|
+
inherit from the MantaClient::MantaClientError class. If you feed ruby-manta
|
216
|
+
an argument it doesn't like, it'll throw ArgumentError. You might also see
|
217
|
+
Errno::ECONNREFUSED and HTTPClient::TimeoutError exceptions from the underlying
|
218
|
+
HTTPClient class.
|
219
|
+
|
220
|
+
Most methods take paths to Manta objects or jobs. Object (or directory) paths
|
221
|
+
are typically of the forms:
|
222
|
+
|
223
|
+
/<user>/stor/<directory>*/<object>
|
224
|
+
/<user>/public/<directory>*/<object>
|
225
|
+
/<user>/jobs/.../stor/...
|
226
|
+
|
227
|
+
The last one is a path for an intermediate or final object generated by a job.
|
228
|
+
Job paths are simpler:
|
229
|
+
|
230
|
+
/<user>/jobs/<job UUID>
|
231
|
+
|
232
|
+
MantaClient methods perform some basic sanity checks to prevent you from
|
233
|
+
using malformed paths, or mixing object/directory paths and job paths.
|
234
|
+
ArgumentError exceptions are immediately thrown if a bad path is provided,
|
235
|
+
otherwise you'll likely receive a MantaClient::ResourceNotFound exception
|
236
|
+
(inherits from MantaClient::MantaClientError) after the call fails in Manta.
|
237
|
+
|
238
|
+
All method calls, except for get_signed_url(), can take an optional :attempts.
|
239
|
+
By default they use the :attempts from the constructor (which default to three
|
240
|
+
tries), but the number of times a Manta call is attempted upon certain failures
|
241
|
+
can be overridden on an individual basis.
|
242
|
+
|
243
|
+
Lastly, recall that due to Manta's semantics you may see 500 errors on occasion.
|
244
|
+
When that happens, try again after a minute or three.
|
245
|
+
|
246
|
+
|
247
|
+
|
248
|
+
Conditional requests
|
249
|
+
--------------------
|
250
|
+
|
251
|
+
Operations on objects support conditional requests. Pass in as an optional
|
252
|
+
argument :if_modified_since, :if_unmodified_since, :if_match, or :if_none_match
|
253
|
+
to the method. For example, to conditionally get an object with etag
|
254
|
+
"e346dce6-22f3-4ed5-8191-6b059b3684de":
|
255
|
+
|
256
|
+
````` ruby
|
257
|
+
|
258
|
+
client.get_object(path, :if_match => 'e346dce6-22f3-4ed5-8191-6b059b3684de')
|
259
|
+
`````
|
260
|
+
|
261
|
+
You can get the Etag or Last-Modified from the headers returned by most
|
262
|
+
methods.
|
263
|
+
|
264
|
+
The methods follow the RFC2616 semantics in the following manner: where 304
|
265
|
+
would return instead of data, a nil returns instead. Where 412 would occur,
|
266
|
+
a MantaClient::PreconditionFailed is thrown.
|
267
|
+
|
268
|
+
Conditional requests allow many good things, so you're advised to use them
|
269
|
+
where applicable. You can conditionally download an object only if it has
|
270
|
+
changed, update an object with CAS semantics, create snaplinks correctly in the
|
271
|
+
face of concurrent updates, and so forth.
|
272
|
+
|
273
|
+
|
274
|
+
|
275
|
+
Cross-origin resource sharing
|
276
|
+
-----------------------------
|
277
|
+
|
278
|
+
Browsers do not allow cross-domain requests due to the same-origin policy.
|
279
|
+
Cross-Origin Resource Sharing (CORS) headers provide a mechanism by which a
|
280
|
+
browser can safely loosen this policy.
|
281
|
+
|
282
|
+
ruby-manta and Manta support all headers specified by the W3C working draft,
|
283
|
+
by passing in optional arguments to put_object() or put_directory():
|
284
|
+
|
285
|
+
:access_control_allow_credentials, :access_control_allow_headers,
|
286
|
+
:access_control_allow_methods, :access_control_allow_origin,
|
287
|
+
:access_control_expose_headers, :access_control_max_age
|
288
|
+
|
289
|
+
You can also pass in :origin to most object- and directory-related methods.
|
290
|
+
|
291
|
+
|
292
|
+
|
293
|
+
initialize(manta_host, user, priv_key, _options_)
|
294
|
+
-------------------------------------------------
|
295
|
+
|
296
|
+
Construct a new MantaClient instance.
|
297
|
+
|
298
|
+
priv_key_data is data read directly from an SSH private key (i.e. RFC 4716
|
299
|
+
format). The method can also accept several optional args: :connect_timeout,
|
300
|
+
:send_timeout, :receive_timeout, :disable_ssl_verification and :attempts.
|
301
|
+
The timeouts are in seconds, and :attempts determines the default number of
|
302
|
+
attempts each method will make upon receiving recoverable errors.
|
303
|
+
|
304
|
+
Will throw an exception if given a key whose format it doesn't understand.
|
305
|
+
|
306
|
+
MantaClient is thread-safe (in theory, anyway), and uses an HTTP client that
|
307
|
+
pools connections. You should only need to initialize a single MantaClient
|
308
|
+
object per process.
|
309
|
+
|
310
|
+
Example:
|
311
|
+
|
312
|
+
````` ruby
|
313
|
+
|
314
|
+
priv_key_data = File.read('/home/john/.ssh/john')
|
315
|
+
client = MantaClient.new('https://manta.joyentcloud.com', 'john',
|
316
|
+
priv_key_data, :disable_ssl_verification => true)
|
317
|
+
`````
|
318
|
+
|
319
|
+
|
320
|
+
|
321
|
+
put_object(object path, file data, _options_)
|
322
|
+
---------------------------------------------
|
323
|
+
|
324
|
+
Uploads object data to Manta to the given path, along with a computed MD5
|
325
|
+
hash.
|
326
|
+
|
327
|
+
The path must be a valid object path. Data can be any sequence of octets.
|
328
|
+
The HTTP Content-Type stored on Manta can be set with an optional
|
329
|
+
:content_type argument; the default is application/octet-stream. The
|
330
|
+
number of distributed replicates of an object stored in Manta can be set
|
331
|
+
with an optional :durability_level; the default is 2. Supports CORS optional
|
332
|
+
arguments.
|
333
|
+
|
334
|
+
Returns true along with received HTTP headers.
|
335
|
+
|
336
|
+
Examples:
|
337
|
+
|
338
|
+
````` ruby
|
339
|
+
|
340
|
+
obj_path, headers = client.put_object('/john/stor/area51_map.png',
|
341
|
+
binary_file_data,
|
342
|
+
:content_type => 'image/png',
|
343
|
+
:durability_level => 1,
|
344
|
+
:if_unmodified_since => Time.now - 300,
|
345
|
+
:access_control_allow_origin => 'http://example.com')
|
346
|
+
|
347
|
+
obj_path, _ = client.put_object('/john/public/pass.txt', 'illuminati 4evah')
|
348
|
+
`````
|
349
|
+
|
350
|
+
|
351
|
+
|
352
|
+
get_object(object path, _options_)
|
353
|
+
----------------------------------
|
354
|
+
|
355
|
+
Get an object from Manta at a given path, and checks it's uncorrupted.
|
356
|
+
|
357
|
+
The object path must point at an actual Manta object. :head => true can
|
358
|
+
optionally be passed in to do a HEAD instead of a GET.
|
359
|
+
|
360
|
+
Returns the retrieved data along with received HTTP headers.
|
361
|
+
|
362
|
+
Examples:
|
363
|
+
|
364
|
+
````` ruby
|
365
|
+
|
366
|
+
_, headers = client.get_object('/john/stor/area51_map.png',
|
367
|
+
:head => true,
|
368
|
+
:origin => 'https://illuminati.org')
|
369
|
+
|
370
|
+
file_data, headers = client.get_object('/john/stor/area51_map.png')
|
371
|
+
`````
|
372
|
+
|
373
|
+
|
374
|
+
|
375
|
+
delete_object(object path, _options_)
|
376
|
+
-------------------------------------
|
377
|
+
|
378
|
+
Deletes an object off Manta at a given path.
|
379
|
+
|
380
|
+
The object path must point at an actual object.
|
381
|
+
|
382
|
+
Returns true along with received HTTP headers.
|
383
|
+
|
384
|
+
Examples:
|
385
|
+
|
386
|
+
````` ruby
|
387
|
+
|
388
|
+
client.delete_object('/john/stor/area51_map.png')
|
389
|
+
|
390
|
+
_, headers = client.delete_object('/john/public/pass.txt')
|
391
|
+
`````
|
392
|
+
|
393
|
+
|
394
|
+
|
395
|
+
put_directory(dir path, _options_)
|
396
|
+
----------------------------------
|
397
|
+
|
398
|
+
Creates a directory on Manta at a given path. Supports CORS optional arguments.
|
399
|
+
|
400
|
+
Returns true along with received HTTP headers.
|
401
|
+
|
402
|
+
Example:
|
403
|
+
|
404
|
+
````` ruby
|
405
|
+
|
406
|
+
client.put_directory('/john/stor/plans-world-domination')
|
407
|
+
|
408
|
+
client.put_directory('/john/public/honeypot',
|
409
|
+
:access_control_allow_methods => 'GET, PUT, DELETE',
|
410
|
+
:access_control_allow_origin => '*')
|
411
|
+
`````
|
412
|
+
|
413
|
+
|
414
|
+
|
415
|
+
list_directory(dir_path, _options_)
|
416
|
+
-----------------------------------
|
417
|
+
|
418
|
+
Gets a lexicographically sorted directory listing on Manta at a given path,
|
419
|
+
|
420
|
+
The path must be a valid directory path and point at an actual directory.
|
421
|
+
:limit optionally changes the maximum number of entries; the default is 1000.
|
422
|
+
If given :marker, an object name in the directory, returned directory entries
|
423
|
+
will begin from that point. :head => true can optionally be passed in to do a
|
424
|
+
HEAD instead of a GET.
|
425
|
+
|
426
|
+
Returns an array of hash objects, each object representing a directory
|
427
|
+
entry. Also returns the received HTTP headers.
|
428
|
+
|
429
|
+
Examples:
|
430
|
+
|
431
|
+
````` ruby
|
432
|
+
|
433
|
+
dir_entries, _ = client.list_directory('/john/stor/plans-world-domination',
|
434
|
+
:limit => 50,
|
435
|
+
:marker => 'take_over_pentagon.txt')
|
436
|
+
|
437
|
+
_, headers = client.list_directory('/john/stor/plans-world-domination',
|
438
|
+
:head => true)
|
439
|
+
`````
|
440
|
+
|
441
|
+
|
442
|
+
|
443
|
+
delete_directory(dir_path, _options_)
|
444
|
+
-------------------------------------
|
445
|
+
|
446
|
+
Removes a directory from Manta at a given path.
|
447
|
+
|
448
|
+
The path must be a valid directory path, and point at an actual directory.
|
449
|
+
The directory must be empty.
|
450
|
+
|
451
|
+
Returns true along with received HTTP headers.
|
452
|
+
|
453
|
+
Example:
|
454
|
+
|
455
|
+
````` ruby
|
456
|
+
|
457
|
+
client.delete_directory('/john/stor/plans-world-domination')
|
458
|
+
`````
|
459
|
+
|
460
|
+
|
461
|
+
|
462
|
+
put_snaplink(orig_path, link_path, _options_)
|
463
|
+
-----------------------------------------
|
464
|
+
|
465
|
+
Creates a snaplink from on object in Manta at a given path to a different path.
|
466
|
+
This effectively creates another reference to the same object. Since objects
|
467
|
+
are immutable, PUTting over that object reduces the refcount on the object;
|
468
|
+
other references (i.e. snaplinks) continue to see the original version.
|
469
|
+
|
470
|
+
Both paths should be valid object paths. orig_path should point at an existing
|
471
|
+
object.
|
472
|
+
|
473
|
+
Returns true along with received HTTP headers.
|
474
|
+
|
475
|
+
Example:
|
476
|
+
|
477
|
+
````` ruby
|
478
|
+
|
479
|
+
client.put_snaplink('/john/stor/character_assassination.txt',
|
480
|
+
'/john/public/media_consultation.txt')
|
481
|
+
`````
|
482
|
+
|
483
|
+
|
484
|
+
|
485
|
+
create_job(job_description, _options_)
|
486
|
+
--------------------------------------
|
487
|
+
|
488
|
+
Creates a job in Manta.
|
489
|
+
|
490
|
+
The job must be a hash, containing at minimum a :phases key. See README.md
|
491
|
+
or the Manta docs to see the format and options for setting up a job on
|
492
|
+
Manta; this method effectively just converts the job hash to JSON and sends
|
493
|
+
to the Manta service.
|
494
|
+
|
495
|
+
Returns the path for the new job, along with received HTTP headers.
|
496
|
+
|
497
|
+
Example:
|
498
|
+
|
499
|
+
````` ruby
|
500
|
+
|
501
|
+
job_desc = { :phases => [{ :exec => 'grep skynet' }] }
|
502
|
+
job_path, _ = client.create_job(job_desc)
|
503
|
+
`````
|
504
|
+
|
505
|
+
|
506
|
+
|
507
|
+
get_job(job_path, _options_)
|
508
|
+
----------------------------
|
509
|
+
|
510
|
+
Gets various information about a job in Manta at a given job path.
|
511
|
+
|
512
|
+
The path must point at an actual job. :head => true can optionally be passed
|
513
|
+
in to do a HEAD instead of a GET.
|
514
|
+
|
515
|
+
Returns a hash with job information, along with received HTTP headers.
|
516
|
+
|
517
|
+
Example:
|
518
|
+
|
519
|
+
````` ruby
|
520
|
+
|
521
|
+
job_path = '/john/jobs/80e481c4-8567-47e7-bdba-f0c5705af1c7'
|
522
|
+
job_info, _ = client.get_job(job_path)
|
523
|
+
`````
|
524
|
+
|
525
|
+
|
526
|
+
|
527
|
+
get_job_errors(job_path, _options_)
|
528
|
+
-----------------------------------
|
529
|
+
|
530
|
+
Gets errors that occured during the execution of a job in Manta at a given
|
531
|
+
job path.
|
532
|
+
|
533
|
+
The must point at an actual job. :head => true can optionally be passed in to
|
534
|
+
do a HEAD instead of a GET.
|
535
|
+
|
536
|
+
Returns an array of hashes, each hash containing information about an
|
537
|
+
error; this information is best-effort by Manta, so it may not be complete.
|
538
|
+
Also returns received HTTP headers.
|
539
|
+
|
540
|
+
Examples:
|
541
|
+
|
542
|
+
````` ruby
|
543
|
+
|
544
|
+
job_path = '/john/jobs/80e481c4-8567-47e7-bdba-f0c5705af1c7'
|
545
|
+
job_errors, _ = client.get_job_errors(job_path)
|
546
|
+
|
547
|
+
_, headers = client.get_job_errors(job_path, :head => true)
|
548
|
+
`````
|
549
|
+
|
550
|
+
|
551
|
+
|
552
|
+
cancel_job(job_path, _options_)
|
553
|
+
-------------------------------
|
554
|
+
|
555
|
+
Cancels a running job in Manta at a given path.
|
556
|
+
|
557
|
+
The job path must point at an actual job.
|
558
|
+
|
559
|
+
Returns true along with received HTTP headers.
|
560
|
+
|
561
|
+
Example:
|
562
|
+
|
563
|
+
````` ruby
|
564
|
+
|
565
|
+
client.cancel_job('/john/jobs/80e481c4-8567-47e7-bdba-f0c5705af1c7')
|
566
|
+
`````
|
567
|
+
|
568
|
+
|
569
|
+
|
570
|
+
add_job_keys(job_path, object_keys, _options_)
|
571
|
+
----------------------------------------------
|
572
|
+
|
573
|
+
Adds objects for a running job in Manta to process.
|
574
|
+
|
575
|
+
The job_path must point at an actual running job. The obj_paths must be an
|
576
|
+
array of object paths pointing at actual objects.
|
577
|
+
|
578
|
+
Returns true, along with received HTTP headers.
|
579
|
+
|
580
|
+
Example:
|
581
|
+
|
582
|
+
````` ruby
|
583
|
+
|
584
|
+
client.add_job_keys('/john/jobs/80e481c4-8567-47e7-bdba-f0c5705af1c7',
|
585
|
+
['/john/stor/skynet_plans.txt',
|
586
|
+
'/john/stor/the_matrix.txt'])
|
587
|
+
`````
|
588
|
+
|
589
|
+
|
590
|
+
|
591
|
+
end_job_input(job_path, _options_)
|
592
|
+
----------------------------------
|
593
|
+
|
594
|
+
Inform Manta that no more objects will be added for processing by a job,
|
595
|
+
and that the job should finish all phases and terminate.
|
596
|
+
|
597
|
+
The job path must point at an actual running job.
|
598
|
+
|
599
|
+
Returns true, along with received HTTP headers.
|
600
|
+
|
601
|
+
Example:
|
602
|
+
|
603
|
+
````` ruby
|
604
|
+
|
605
|
+
client.end_job_input('/john/jobs/80e481c4-8567-47e7-bdba-f0c5705af1c7')
|
606
|
+
`````
|
607
|
+
|
608
|
+
|
609
|
+
|
610
|
+
|
611
|
+
get_job_input(job_path, _options_)
|
612
|
+
----------------------------------
|
613
|
+
|
614
|
+
Get a list of objects that have been given to a Manta job for processing.
|
615
|
+
|
616
|
+
The job path must point at an actual running job.
|
617
|
+
|
618
|
+
Returns an array of object paths, along with received HTTP headers.
|
619
|
+
|
620
|
+
Example:
|
621
|
+
|
622
|
+
````` ruby
|
623
|
+
|
624
|
+
job_path = '/john/jobs/80e481c4-8567-47e7-bdba-f0c5705af1c7'
|
625
|
+
obj_paths, _ = client.get_job_input(job_path)
|
626
|
+
`````
|
627
|
+
|
628
|
+
|
629
|
+
|
630
|
+
get_job_output(job_path, _options_)
|
631
|
+
-----------------------------------
|
632
|
+
|
633
|
+
Get a list of objects that contain the intermediate and/or final results of a
|
634
|
+
running Manta job.
|
635
|
+
|
636
|
+
The job_path must point at an actual running job.
|
637
|
+
|
638
|
+
Returns an array of object paths, along with received HTTP headers.
|
639
|
+
|
640
|
+
Example:
|
641
|
+
|
642
|
+
````` ruby
|
643
|
+
|
644
|
+
job_path = '/john/jobs/80e481c4-8567-47e7-bdba-f0c5705af1c7'
|
645
|
+
obj_paths, _ = client.get_job_output(job_path)
|
646
|
+
`````
|
647
|
+
|
648
|
+
|
649
|
+
|
650
|
+
get_job_failures(job_path, _options_)
|
651
|
+
-------------------------------------
|
652
|
+
|
653
|
+
Get a list of objects that had failures during processing in a Manta job.
|
654
|
+
|
655
|
+
The job path must point at an actual running job.
|
656
|
+
|
657
|
+
Returns an array of object paths, along with received HTTP headers.
|
658
|
+
|
659
|
+
Example:
|
660
|
+
|
661
|
+
````` ruby
|
662
|
+
|
663
|
+
job_path = '/john/jobs/80e481c4-8567-47e7-bdba-f0c5705af1c7'
|
664
|
+
obj_failures, _ = client.get_job_failures(job_path)
|
665
|
+
`````
|
666
|
+
|
667
|
+
|
668
|
+
|
669
|
+
list_jobs(state, _options_)
|
670
|
+
---------------------------
|
671
|
+
|
672
|
+
Get a list of Manta jobs.
|
673
|
+
|
674
|
+
The state indicates which kind of jobs to return. :running is for jobs
|
675
|
+
that are currently processing, :done and :all should be obvious. Be careful
|
676
|
+
of the latter two if you've run a lot of jobs -- the list could be quite
|
677
|
+
long.
|
678
|
+
|
679
|
+
Returns an array of hashes, each hash containing some information about a job.
|
680
|
+
Also returns received HTTP headers.
|
681
|
+
|
682
|
+
Example:
|
683
|
+
|
684
|
+
````` ruby
|
685
|
+
|
686
|
+
running_jobs, _ = client.list_jobs(:running)
|
687
|
+
`````
|
688
|
+
|
689
|
+
|
690
|
+
|
691
|
+
gen_signed_url(expiry_date, http_method, path, _query_args_)
|
692
|
+
----------------------------------------------------------
|
693
|
+
|
694
|
+
Generates a signed URL which can be used by unauthenticated users to
|
695
|
+
make a request to Manta at the given path. This is typically used to GET
|
696
|
+
an object.
|
697
|
+
|
698
|
+
expires is a Time object or integer representing time after epoch; this
|
699
|
+
determines how long the signed URL will be valid for. The method is the HTTP
|
700
|
+
method (:get, :put, :post, :delete) the signed URL is allowed to be used
|
701
|
+
for. The path must be a valid object (or directory?) path. Lastly, the
|
702
|
+
optional args is an array containing pairs of query args that will be
|
703
|
+
appended at the end of the URL.
|
704
|
+
|
705
|
+
The returned URL is signed, and can be used either over HTTP or HTTPS until
|
706
|
+
it reaches the expiry date.
|
707
|
+
|
708
|
+
Example:
|
709
|
+
|
710
|
+
````` ruby
|
711
|
+
|
712
|
+
url = client.gen_signed_url(Time.now + 5000, :get, '/john/stor/pass.txt')
|
713
|
+
`````
|
714
|
+
|
715
|
+
|
716
|
+
|
717
|
+
License
|
718
|
+
=======
|
719
|
+
|
720
|
+
(c) 2012 Joyent, licensed under MIT. See LICENSE for details, you legal geek
|
721
|
+
you.
|
722
|
+
|