* dynamically call TypeToken.isSupertypeOf with Guava 19 and later and
TypeToken.isAssignableFrom with Guava 18 and earlier
* consume or ignore values from methods with CheckReturnValue
* replace usage of removed Iterators.emptyIterator
Today the filesystem blobstore provider use the extended attributes of the
filesystem to store metadata. This is not always desirable as some filesystems
(such as NFSv3) do not support extended metadata. The current source code does
not allow to change this easily. getUserDefinedFileAttributeView could easily
be overriden for this purpose and allow extensions that customize where the
metadata should be stored.
See also JCLOUDS-658
This is a combination of 16 commits:
* First approach to ArbitraryCpuRamTemplateBuilderImpl
* Several fixes: refactoring some names, format, identation problems, some missing license headers and generateId method
* Refactored parse utility
* Added GoogleComputeEngineArbitraryCpuRamTemplateBuilderImpl to support GCE custom machine URI
* extracted hardware creation to automaticHardwareForCpuAndRam method
* Fixed ide automatic asterisk imports
* correcting WIP base case PR according to comments
* added machineTypeUriToHardware to set custom hardware in nodes
* fix checkstyle violations and other PR comments
* Set the providerId to custom machineType URI and fix adding node log
* Arbitrary hardware tests added to BaseTemplateBuilderLiveTest and GoogleComputeEngineTemplateBuilderLiveTest
* Added two more tests to BaseTemplateBuilderLiveTest
* Move repeated constants to TestUtils to reuse code
* Fix full path in the Hardware id and URI
* Add custom hardware tests to BaseComputeServiceLiveTest and GCEServiceLiveTest
* Change customHardware test to use buildTemplate and fix identation
Previously it mocked the container and the StateToStatus. It just
asserted that getter methods were called, rather than confirming the
resulting NodeMetadata had the right values.
By removing some of the mocks, it not only simplifies the code but
improves code-coverage. (e.g. if StateToStatus was being passed the
wrong value, the previous test would not have noticed).
When providing node ip in metadata jclouds was using the
ip of the docker manager which was fine for docker machine
but fails when using a swarm cluster. This uses the ip returned
from docker.
Previously we created a new builder instead of using the one the
method modified and did not preserve content encoding. Addresses
AWSS3BlobIntegrationLiveTest.testPutInputStream test failures.
Due to Swift's wonky MPU implementation we must try the multi-part
delete first followed by a single-part delete if it fails. Previously
delete would leave orphan subobjects. Fixesandrewgaul/s3proxy#92.
This commit refactors the ImageCacheSupplier to act as a
proper cache. It is used by the ImageExtesion and all operations
on the images are propagated to the cache.
A method has also been added to the TemplateBuilder to let users
force a cache refresh. There have been several requests to provide a way
to disable image caching in the compute abstraction, and this new method
should fix that.
AWS S3 MPU ETag are hashes of the part ETag headers but some
implementations, specifically S3Proxy with the filesystem provider,
represent multi-part objects as a single object. Remove these checks
since they add nothing.
These are simpler than the full XML API and better supported by
non-AWS S3 implementations, e.g., Ceph, S3Proxy. Further this makes
the provider more consistent when creating a bucket or object which
only supports setting canned ACLs.
Swift implements multi-part upload with user-visible parts and an
explicit manifest. When deleting an MPU blob it can delete only the
manifest or both the manifest and parts. For consistency with other
providers, we now do the latter in the portable abstraction. Swift
ignores the multipart-manifest=delete parameter for single-part
objects. Fixesandrewgaul/s3proxy#92.
Previously we only used the blob name and not the blob
name/slo/timestamp/part size cookie which yield extra parts when
listing an MPU with parts from a previous MPU. Listing using the
stricter prefix gives the expected results.
Fixesandrewgaul/s3proxy#91.
- bump api version to 1.21
- use `alpine/3.2` image for liveTests
- use `kwart/alpine-ext:3.2-ssh` image as ssh-able image
- assert request bodies created correctly in NetworkApiMockTest
On Windows, we need to avoid trailing spaces, as the test fails to
create the required blob. Specifically, we should not test blobs named
" " and "%20 ".
When constructing the expected HTTP requests in STS tests, STS API
should not pre-encode the strings. The form parameters are already
encoded and are otherwise encoded twice.
When constructing the query path, S3 does not properly handle encoded
paths. For example, if a blob named %20 is to be placed into the blob
store, S3 would end up placing blob named " " (what %20 represents).
This occurs because the S3 provider examines the URI's path portion
(which is presented in a decoded fasion to the caller). After
examining the path, it is not encoded again. Instead, we should call
getRawPath() to avoid this issue.
There are two issues on the decoding path:
1. Given a blob named " ", S3 API will throw a RuntimeException due to
a NULL check -- the key that it uses is NULL to represent the XML
content " " corresponding to the blob name.
2. Given a blob named "%20 ", S3 API will generate a URI for a blob
named "%20%20", which is also incorrect. The correct URI would be
"%2520%20" (escaping the first "%" and " " characters).
The first issue is due to the currentOrNull() helper, which calls
trim() on the string and then compares the string to an empty string.
This means that a blob named " " will be parsed as "" and then
converted to NULL as the result of that method. Passing "null" as the
key then fails in a number of places (notably, appendPath()).
The second issue is due to the appendPath() method in the jclouds Uris
class. The issue here is that appendPath() calls urlDecode() and
passes the result to path(). The path() method, in turn, also calls
urlDecode(). After these transformations, a properly encoded blob of
the form %2520%20 turns into "%20 " and then " " (two spaces). After
these transformations the path is encoded again, resulting in "%20%20"
(which is wrong).
deprecated hardware types in EC2 are flagged using this, according to latest AWS advice,
to fix situations where deprecated hardware types would be chosen when a non-deprecated alternative exists.
we also deprecate T2 because it requires a VPC.
also fixes semantics of ImagesToRegionAndIdMap to match the logic used in the TemplateBuilderImpl;
RegionAndName should contain the ID not the ProviderID. this is only really used in tests so no external impact.
(previously the cache only worked if provider ID matched the non-location segment of the image ID.)
however this does now assert that image id's are in the right format for AWS, as suggested by @nacx.
Move the prefix and delimiter tests from BaseContainerLiveTest to
BaseContainerIntegrationTest, as that class is subclassed in
Filesystem integration tests. This will make sure the file system blob
store will not regress.
The ListContainerTest is updated to _not_ use a file separator
character, as that leads to a possible creation of directories.
Instead, "-" is used as the delimiter for the delimiter option tests.
The patch adds the delimiter support in the openstack-swift API. As
part of the change, the subdirectory support in results is introduced.
This occurs when a prefix and delimiter options are set and there are
subdirectories present in the listing (i.e. multiple objects under the
same prefix/delimiter). In this case, Swift will return a list of
"subdir" objects (similar to CommonPrefixes in S3), which need to be
treated differently.
The patch adds delimiter option support in the generic blob store
interface. A live integration test is added to verify that jclouds
correctly lists objects separated by a delimiter.
Openstack Swift ContainerApiLiveTest testListWithOptions assumes that
there are no other containers and that ContainerApiLiveTest container
will _always_ be the first container.
The patch changes the behavior of the test to instead pull out the
container from the returned list.
There is no notion of marker blobs in the file system blob store,
which means that when listing results after issuing the
"createDirectory" request, the marker blob will not be in the result
set.
When listing blobs, Swift returns an array of "subdir" entries for
every common prefix. The openstack-swift provider needs to process
such entries and add them to the result set. The subdir array is an
array of strings, so additional metadata needs to be added to the
objects (bogus ETag, bogus LastModifiedDate, and so on).
When directory marker blobs are used, this means that potential _two_
entries are generated for every directory if:
1. the delimiter is set and matches the directory blob (e.g. dir/ and
delimiter /)
2. there are objects under the directory name (e.g. dir/blob), which
will result in results that include common prefixes (subdir)
In the above example, we should expect two results: "dir" and "dir/"
representing the directory marker blob and the common prefix,
respectively.
This is caught in the testDirectory integration test.
The patch changes the behavior of the Swift provider to correctly
handle the results in the subdir stanza and changes the test to expect
the directory marker to be returned in the list.
Plumbs the prefix option to the openstack-swift provider. In the
process, the support for the recursive option is modified to avoid
setting the _path_ parameter, but rather use the delimiter if required
(setting the delimiter is sufficient for a non-recursive listing).
- Tidy up DockerTemplateOptions and add tests for new options
- Make Dns a list of strings in HostConfig
- Change template option builder to remove use of optionals
- Update tests to verify HostConfig.Dns as list
When a filesystem blob does not have content metadata and when users
set jclouds.filesystem.auto-detect-content-type to tru, probe the file
type to return to clients. This is useful when using jclouds to serve
an existing filesystem.
Currently swift creates blob-1, blob-2... blob-n for multipart
upload segments, which are very common names that can easily
collide with normal objects. This changes the naming convention
to that of python-swiftclient (the `swift' command line).
python-swiftclient also uses a different segment container by
default, that's not implemented by this patch
Providers yield different messages so remove parsing from test. Also
narrow exception handling scope.TTP 416 into IllegalArgumentException
Providers yield different messages so remove parsing from test. Also
narrow exception handling scope.
Ignasi Barrera
Arvind Nadendla
Andrew Gaul
Zack Shoylev
Francois Rigault
Daniel Haeser Rech
Josef Cacek
Iván Lomba
Andrew Donald Kennedy
neonbunny
Andrea Turli
Duncan Grant
urban
Aled Sage
Jim Spring
Oleg
Fernando Ribeiro
IgnacioCalvo
rongallagher
Reijhanniel Jearl Campos
Zhao Jin
Rita Zhang
Graeme-Miller
Ka-Hing Cheung
Stuart Hendren
Josef Cacek
John McDonnell
googlielmo
Timur Alperovich
Irmo Manie
Oliver Gondža
Nate Rosenblum
Adrian Bravo
Marcus Wanner
Csaba Palfi
Alex Heneveld
Andrew Bayer
Iván López