Commit graph

186 commits

Author SHA1 Message Date
Kyle Marsh
76956d86e4 S3 Fuzzer: finish describing bucket operations
Finishing touches on decision graph for bucket operations.
2011-09-12 12:59:53 -07:00
Kyle Marsh
e12f124686 S3 Fuzzer: Change how random data works
Remove SpecialVariables dict subclass in favor of RepeatExpandingFormatter
string.Formatter subclass.
2011-09-12 12:58:57 -07:00
Kyle Marsh
14288ad2f6 S3 Fuzzer: Incorporate Tv's suggestions
Tv looked over the fuzzer and had some idiomatic and design suggestions.

Adds several tests and reworks how expansion happens in addition to idiom
changes.
2011-09-12 12:57:43 -07:00
Kyle Marsh
bb7111a0d1 S3 Fuzzer: Write Decision Graph
Start writing the decision graph describing the fuzzer's attack surface in
earnest.
2011-09-12 12:55:32 -07:00
Kyle Marsh
62bd05a390 S3 Fuzzer: Handle null choices
Sometimes you might want to have your current node terminate the descent or
set something to the empty string.
2011-09-12 12:53:18 -07:00
Kyle Marsh
d7b49713f7 S3 Fuzzer: Implmented headers and made random safe
Random can sometimes include } or { which will confuse the string formatter.
Formatter escapes those values when they're doubled: }}, {{ but this
required some slight hacking to the expander.
2011-09-12 12:53:18 -07:00
Kyle Marsh
3f1314f7c8 S3 Fuzzer: set values can be weighted lists now 2011-09-12 12:53:18 -07:00
Kyle Marsh
4737652fc1 S3 Fuzzer: Added binary mode to random data generator 2011-09-12 12:53:18 -07:00
Kyle Marsh
195571b555 S3 Fuzzer: support weights for choices 2011-09-12 12:53:18 -07:00
Kyle Marsh
f5bb3f9c15 S3 Fuzzer: implemented decision expansion
Decision reached by descending the graph describing the attack surface can
be expanded to fill in all the placeholders
2011-09-12 12:53:17 -07:00
Kyle Marsh
7d9ec02686 S3 Fuzzer: Added SpecialVariables dict subclass
Helper class to catch sentinal keys and turn them into random values.  This
will be used to generate garbage data when expanding a decision.

Also add unit tests for expand_decision and assemble_decision
2011-09-12 12:52:37 -07:00
Kyle Marsh
a9a41a2891 S3 Fuzzer: began writing graph descent
still missing headers and choice weights
2011-09-12 12:52:37 -07:00
Kyle Marsh
fc93c02963 S3 Fuzzer: Change direction towards decision tree
Fuzzer now builds requests based on a DAG that describes the request space
and attack surface.
2011-09-12 12:51:01 -07:00
Kyle Marsh
691955935d S3 Fuzzer: add authorization to FuzzyRequest 2011-09-12 12:50:43 -07:00
Kyle Marsh
31adf6afa5 S3 Fuzzer: Begin writing unit tests for S3 Fuzzer 2011-09-12 12:50:14 -07:00
Kyle Marsh
93cf99aed9 S3 Fuzzer: Add skeleton for a targeted S3 Fuzzer 2011-09-12 12:49:12 -07:00
Stephon Striplin
2efa622751 Add atomic read tests 2011-08-29 13:14:18 -07:00
Stephon Striplin
f232f43d0b update attr decorators for dho
* passes test_object_raw_put_authenticated
* fails test_atomic_write_bucket_gone
2011-08-29 12:47:17 -07:00
Stephon Striplin
eeaeec573b don't delete key in atomic gone test
The key hasn't been written to the server so it doesn't need a delete
2011-08-29 12:36:00 -07:00
Stephon Striplin
3d775a02f4 add raw PUT tests 2011-08-26 13:45:37 -07:00
Stephon Striplin
76f78c0bb8 refactor raw GET tests
urllib2 is no longer necessary. We will use urlparse instead to get the
path and query components of the uri.
2011-08-26 13:42:31 -07:00
Stephon Striplin
e4d78eccbb add test for missing Content-Length header bucket creation 2011-08-25 17:46:25 -07:00
Stephon Striplin
aa700f1c8a add an atomic put test for removing bucket 2011-08-25 17:20:11 -07:00
Stephon Striplin
18d0b7d250 Add raw GET tests 2011-08-25 11:39:50 -07:00
Stephon Striplin
ea3f73ef90 add bucket list data vs object data comparison tests
* test_bucket_list_return_data
* test_bucket_list_object_time
2011-08-11 15:58:26 -07:00
Stephon Striplin
fd68aa2b5b remove superfluous code in test_bucket_list_maxkeys_zero 2011-08-11 15:52:16 -07:00
Stephon Striplin
16d84101e4 remove fails_on_dho on passing tests
* test_object_set_get_metadata_empty_to_unreadable_prefix
* test_object_set_get_metadata_empty_to_unreadable_suffix
* test_object_set_get_metadata_empty_to_unreadable_infix
* test_object_set_get_metadata_overwrite_to_unreadable_prefix
* test_object_set_get_metadata_overwrite_to_unreadable_suffix
* test_object_set_get_metadata_overwrite_to_unreadable_infix
2011-08-11 13:42:10 -07:00
Stephon Striplin
5e7792e0bf Switch test_bucket_list_maxkeys_unreadble to use '\n' instead of '\a'.
This avoids the SAX parser error. It also gave a different error code
from Amazon.
2011-08-11 11:23:43 -07:00
Stephon Striplin
80d332346f Switch tests using '\a' to '\n'
As noted in ce3d448d64, This will avoid
SAX parser errors.
2011-08-11 11:21:17 -07:00
Stephon Striplin
adcfe6f681 remove contentlength zero header test
This test effectively tells the server that this request will contain no
data and any data in the body should be interpreted to be part of the
next request. By luck, we close the tcp connection beforehand so we
don't see this additional data. However, the issues caused here would
not be at the rgw/dho level.
2011-08-11 11:01:56 -07:00
Tommi Virtanen
230ec11852 Flag test_object_create_bad_contenttype_unreadable as failing on both RGW and DHO. 2011-08-11 09:35:06 -07:00
Tommi Virtanen
76faa58bf7 DHO does not support bucket logging currently. 2011-08-11 09:31:20 -07:00
Tommi Virtanen
ce3d448d64 Switch test_object_create_unreadable to use '\n' instead of '\a'.
This avoids a SAX/boto interaction bug that is not of much interest,
yet still seems "evil" enough that it fulfills the original intent
of the test.

Add a link to the relevant bug report.
2011-08-11 09:30:26 -07:00
Tommi Virtanen
48e078a758 Flag test_bucket_list_maxkeys_unreadable as failing on dho & rgw. 2011-08-11 09:28:48 -07:00
Tommi Virtanen
6577c8266c Flag some fails_on_dho tests as fails_on_rgw too. 2011-08-11 08:47:25 -07:00
Tommi Virtanen
7da111aa7e Test that S3 object write replaces the metadata.
See http://tracker.newdream.net/issues/1352
2011-08-10 15:04:19 -07:00
Tommi Virtanen
199b8b46c5 Test S3 api for handling of Content-MD5 that does not match.
The existing test_object_create_bad_md5_invalid is not valid
base64, and hence triggers a different code path. This time,
submit valid base64, but have the result not match the expected
md5sum.
2011-08-10 14:55:34 -07:00
Stephon Striplin
0e9935a01e refactor/add additional prefix/delimiter bucket list tests 2011-08-10 12:05:28 -07:00
Stephon Striplin
f855beff73 add marker bucket list tests 2011-08-09 18:42:06 -07:00
Stephon Striplin
1f2810557f add maxkeys bucket list tests 2011-08-09 18:35:18 -07:00
Stephon Striplin
e7f66f178e add test for no bucket list delimiter 2011-08-09 15:02:54 -07:00
Stephon Striplin
881c09f626 add test for checking bucket list with no prefix 2011-08-08 17:39:02 -07:00
Stephon Striplin
0e698734b6 add non-existant prefix bucket list test 2011-08-08 17:24:40 -07:00
Stephon Striplin
c89404ba68 remove fails_on_dho/rgw on test_bucket_list_many 2011-08-08 16:40:14 -07:00
Stephon Striplin
854dcc5b10 create additional parameterized bucket list tests
* Test on prefix and delimiter, including slash and non-slash
  delimiters.
2011-08-08 15:16:15 -07:00
Stephon Striplin
a1102aaaad add Authorization header tests 2011-07-28 16:46:53 -07:00
Stephon Striplin
1dedf26b95 refactor header update logic 2011-07-28 11:34:52 -07:00
Tommi Virtanen
82a645c625 Demonstrate rgw bucket listing bug.
Discovered by stumbling on the "deleted_cnt" workaround
in s3tests.common.

TODO: Add tests for rest of the functionality in
http://docs.amazonwebservices.com/AmazonS3/latest/API/index.html?RESTBucketGET.html
2011-07-28 11:00:47 -07:00
Tommi Virtanen
a9c47e4951 RandomContentFile.start_time is never used, remove. 2011-07-28 10:20:47 -07:00
Stephon Striplin
832b297cec support multiple boto versions for header writing
Depending on the version of boto, different methods are necessary for
manipulating headers. We now support 2.0rc1 and 2.0 by sub-classing
boto.s3.connection.S3Connection and overloading
boto.connection.HTTPRequest.authorize, respectfully.
2011-07-27 20:04:33 -07:00
Kyle Marsh
68c8f4b87c s3-tests: Object download response code test
If a client specifies a range when requesting an object the server should
respond with 206 Partial Content when answering a request to get an object,
not 200 OK.  It should also respond with only those bytes requested.

This commit introduces a test, test_ranged_request_response_code, that
checks those criteria.
2011-07-27 15:01:13 -07:00
Stephon Striplin
ade49565db change utf8 -> unreadable on header tests 2011-07-27 11:49:18 -07:00
Tommi Virtanen
561dbe8f0b Measure and record duration of write and read stages in roundtrip test. 2011-07-27 11:33:15 -07:00
Stephon Striplin
3b15fc7547 add Date header tests 2011-07-27 10:31:00 -07:00
Stephon Striplin
28c4b432c4 add a Content-Type header test 2011-07-27 10:30:24 -07:00
Stephon Striplin
ab7b4eb03d update Content-Length header tests
* Add more tests
* Add nose.with_setup on some tests
2011-07-27 10:30:00 -07:00
Stephon Striplin
73a2f8fd7f add another Expect header test 2011-07-27 10:29:37 -07:00
Stephon Striplin
2b63bfc18d update MD5 header tests
* Rename a test
* Add more tests
2011-07-27 10:29:06 -07:00
Stephon Striplin
405f33e10c add User-Agent header tests 2011-07-27 10:28:03 -07:00
Stephon Striplin
82ac4f0138 subclass S3Connection class for headers
This makes it possible to overwrite all headers
2011-07-27 10:27:45 -07:00
Tommi Virtanen
7f4fcecaef Remove dead code. 2011-07-26 16:29:16 -07:00
Tommi Virtanen
8a43540eea Convert s3tests.common back to a module.
Code in __init__.py is evil.
2011-07-26 16:26:51 -07:00
Tommi Virtanen
3f49775ce3 Remove dead code. 2011-07-26 16:26:09 -07:00
Tommi Virtanen
700dd3f880 Rewrite roundtrip test based on the readwrite one.
They probably should share more code in the future, but it's only
200 lines of python, so I'm not all that worried.
2011-07-26 16:22:40 -07:00
Tommi Virtanen
f3718da14c Let config file control random number generator seeds in detail. 2011-07-26 15:17:35 -07:00
Tommi Virtanen
7debb53e77 No shebang lines for setup.py-managed scripts. 2011-07-26 14:26:28 -07:00
Tommi Virtanen
513d5fabf2 Make readwrite test not use unittest setup/teardown logic. 2011-07-26 14:20:28 -07:00
Tommi Virtanen
f58641d0bf Refactor bucket nuking in cleanup. 2011-07-26 14:20:28 -07:00
Tommi Virtanen
f30f9d0aca Refactor S3 connection opening. 2011-07-26 14:20:28 -07:00
Tommi Virtanen
367fd3981a Refactor config file reading. 2011-07-26 14:20:28 -07:00
Tommi Virtanen
7a98a496e4 Rename rand_readwrite to readwrite.
Simpler, and all the benchmarks are pretty much as pseudorandom,
in the end; that's not a relevant characteristic. The mix of read
and write operations is.

Re-run ./bootstrap to get the s3tests-test-readwrite script
regenerated to refer to the new module name.
2011-07-26 14:20:28 -07:00
Tommi Virtanen
e17c56a714 Say ".yaml" not ".yml". This is not MS-DOS, and YML is a Yahoo XML dialect. 2011-07-26 14:20:28 -07:00
Tommi Virtanen
3eb53db2b3 Don't bother with fractional nanoseconds.
<1 nanosecond is well below the threshold we can measure,
everything becomes simpler with integers, and they're easier
to skim.
2011-07-26 14:20:28 -07:00
Tommi Virtanen
f21b410154 Use a constant for nanoseconds-in-a-second.
State it in terms of 1e9 (as int not float). My eyes get
blurry from that many zeros.
2011-07-26 14:20:28 -07:00
Tommi Virtanen
37950ac5f2 Remove unused import. 2011-07-26 14:20:28 -07:00
Tommi Virtanen
1993c33ee3 Avoid built-in "file" as variable name. 2011-07-26 14:20:28 -07:00
Tommi Virtanen
3b204bfbdc Revert "adding max_amount to realistic.names generator"
This is now unused code.

This reverts commit 52ba1d4fbf9b3547f8fd27199cf2c4f2c59dbd45.
2011-07-26 14:20:28 -07:00
Tommi Virtanen
91e3918ef5 Use itertools.islice instead of special-casing generator slicing. 2011-07-26 14:20:28 -07:00
Steven Berler
55a72f002b moved most readwrite tool command-line opts to config file
The only remaining command-line options are setting the random seed
and --no-cleanup.
2011-07-26 14:20:28 -07:00
Steven Berler
a6c16990df remove extraneous #TODO 2011-07-26 14:20:28 -07:00
Steven Berler
0c0c590d47 use stderr for non-yml output in readwrite tool
stdout should now only contain the yml output.
2011-07-26 14:20:28 -07:00
Steven Berler
33876b0dae fix yml format in readwrite tool
Now will print the chunks out one line per chunk instead of splitting
them accross multiple lines.
2011-07-26 14:20:28 -07:00
Steven Berler
1e991bc4bc adding chunks to readwrite tool output 2011-07-26 14:20:28 -07:00
Steven Berler
8cf9c57012 fix readwrite tool file sizes
Makes the file sizes specified correctly in KB instead of B
2011-07-26 14:20:28 -07:00
Steven Berler
15cf598b12 readwrite tool improvements
Now generates a list of filenames to use duing the setup, and passes this
list to the workers rather than having the workers list the bucket contents
or generate file names themselves.

Added a "warmup" stage that first writes out all the files so that content
exists for each filename.

Added setting the random seed as a command line parameter and uses this seed
for both generating the file names and the file content generator.
2011-07-26 14:20:28 -07:00
Steven Berler
b465596063 adding max_amount to realistic.names generator
Set max_amount to limit the number of names yeilded by the generator.
2011-07-26 14:20:28 -07:00
Stephon Striplin
a775f5da04 add Content-Type object tests 2011-07-25 14:12:50 -07:00
Stephon Striplin
2e351e9eb8 add header-manipulation tests
These tests are based on explicitly modifying headers in some way.
2011-07-25 11:25:39 -07:00
Stephon Striplin
ee28900620 add test_object_create_unreadable 2011-07-25 11:25:39 -07:00
Yehuda Sadeh
9b3cdf753f test consistency of bucket acls, exercises cache 2011-07-21 14:10:05 -07:00
Stephon Striplin
338728ecd2 add test_bucket_acl_no_grants
When the owner has no grants on their bucket, they should not be able to
write to it, but still read and write the ACL. Amazon allows users to
read the bucket, and our behavior is consistent with this.
2011-07-20 12:30:23 -07:00
Tommi Virtanen
b54e1cbe99 Handle and report errors in readwrite test workers. 2011-07-20 11:32:52 -07:00
Tommi Virtanen
f3b5ed6ccc Say "worker_id" instead of name, as I keep confusing it with filenames. 2011-07-20 11:15:06 -07:00
Tommi Virtanen
1eb38e09e4 Link worker greenlets to parent, so we abort if they have an unhandled exception. 2011-07-20 11:06:30 -07:00
Tommi Virtanen
9ecaf63fbe Code layout cleanup. 2011-07-20 11:06:30 -07:00
Tommi Virtanen
deb8be9957 Make readwrite test put objects one by one, so it can report keys. 2011-07-20 11:06:30 -07:00
Tommi Virtanen
9e707b897d Remove the concept of batches from readwrite test.
That's an internal implementation detail caused by
the generate_objects api.
2011-07-20 11:06:30 -07:00
Tommi Virtanen
4cd4131dd7 Don't eat exceptions, that's a bad habit. 2011-07-20 11:06:30 -07:00
Tommi Virtanen
648ddbb761 Log readwrite test results in yaml.
This temporarily removes the statistics calculation; that
belongs somewhere else anyway, so it can be run on aggregated
results etc.
2011-07-20 11:06:30 -07:00
Tommi Virtanen
ed966af0a7 Use more idiomatic Python. 2011-07-20 11:06:30 -07:00