| Age | Commit message (Collapse) | Author |
|---|
|
|
|
Due to the way the azure boot works, where we identify the hostname to the
fabric, we were inadvertently overwriting hostname that the user had set.
LP: #1375252
|
|
|
|
|
|
handling of cloud-config-archive input would fail in fully_decoded_payload.
part.get_charset() would return a Charset object, but
get_charset.input_codec is a string suitable for passing to decode.
This handles that correctly, and is more careful about binary data inside
input.
The test added verifies that cloud-config inside a cloud-config-archive
is handled correctly and also that binary data there is ignored without
exceptions raised.
LP: #1445143
|
|
|
|
|
|
Including minor refactoring to make mocking considerably easier.
|
|
It was believed that to install a package with config the command was:
snappy install --config=config-file <package>
Instead, what was implemented in snappy was:
snappy install <package> [<config-file>]
This modifies cloud-init to invoke the latter and changes the tests
appropriately.
LP: #1438836
|
|
|
|
On fspath installs, look for .config files harder.
Given a file named:
pkg.namespace_0.version_arch.snap
We'll search for config files named:
pkg.namespace_0.version_arch.config
pkg.namespace.config
pkg.config
|
|
|
|
the input to 'snappy config <packagename>' is expected to have
config:
<packagename>:
content:
So here we pad that input correctly. Note, that a .config file
on disk is not modified.
Also, we change 'configs' to just be 'config', to be possibly compatible
with the a future 'snappy config /' that dumped:
config:
pkg1: data1
pkg2: data2
|
|
|
|
|
|
|
|
|
|
|
|
We are treating JoyentMetadataClient as a unit which the data source
depends on, so we mock it out instead of providing a fake implementation
of it.
|
|
|
|
|
|
|
|
Cloud-init crashed when received multipart userdata object with
'application/octet-stream' part or some other 'application/*' part
except archived ones (x-gzip and friends). These parts are not
processed by cloud-init and result only in a message in the log.
We used some non-python3-friendly techniques while generating
this log message which was a reason for the crash.
|
|
HTTPretty can causes hangs on Python 3.4.2 (and maybe Python 3.4.1), due
to a Python bug (fixed in Python 3.4.3). This works around the problem
in the appropriate Python versions.
See https://github.com/gabrielfalcao/HTTPretty/pull/193 and
https://github.com/gabrielfalcao/HTTPretty/issues/221 for details.
|
|
This is to work-around the fact that email.message_from_string uses
cStringIO in Python 2.6, which can't handle Unicode.
|
|
HTTPretty can causes hangs on Python 3.4.2 (and maybe Python 3.4.1), due
to a Python bug (fixed in Python 3.4.3). This works around the problem
in the appropriate Python versions.
See https://github.com/gabrielfalcao/HTTPretty/pull/193 and
https://github.com/gabrielfalcao/HTTPretty/issues/221 for details.
|
|
dmidecode and /sys/class/dmi/id/* use different names for the same
information. This modified the logic in util.read_dmi_data to map from
dmidecode names to sysfs names before looking in sysfs.
|
|
get_cmdline_url was passing a string to response.contents.startswith()
where response.contents is now bytes.
this changes it to convert input to text, and also to default to text.
|
|
|
|
UrlResponse: biggest change... make readurl return bytes, making user
know what to do with it.
util: add load_tfile_or_url for loading text file or url
as read_file_or_url now returns bytes
ec2_utils: all meta-data is text, remove non-obvious string translations
DigitalOcean: adjust for ec2_utils
DataSourceGCE, DataSourceMAAS: user-data is binary other fields are text.
openstack.py: read paths without decoding to text. This is ok as paths
other than user-data are json, and load_json will handle
load_file still returns text, and that is what most things use.
|
|
explicitly test compressed user-data.
|
|
This was previously broken in python3 as the userdata would be bytes
rather than a string.
LP: #1423972
|
|
just seems to make more sense to decode here.
Add a test showing the previous failure (testBytesInPayload)
And one that should pass (testStringInPayload)
Also, add a test for unencoded content in the ovf xml (test_userdata_plain)
And explicitly set encoding on another test (test_userdata_found).
|
|
|
|
|
|
This should fix the Azure data source on Python 3, and is appropriate as
XML shouldn't really be read as a string.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
painful, and not perfect, but at this point the output builds
on a vivid system python2 (bddeb --python2) or python3.
* remove use of cheetah by bddeb in favor of builtin renderer
* add '--python2' flag to bddeb and knowledge of python 2 and python3
package names.
* read-dependencies can now read test-requirements also.
* differenciate from build-requirements and runtime requirements.
|
|
- Refactor both the base64 encoding and decoding into utility functions.
Also:
- Mechanically fix some other broken untested code.
|
|
|
|
|
|
|
|
|
