| Age | Commit message (Collapse) | Author |
|---|
|
This would cause problems in the event that we actually had a bad
clock. We add a retry in the main (for test) also, to ensure that
the oauth timestamp fix gets in place.
LP: #1499869
|
|
the previous version was broken. The vital fixes here are:
* adding parsedate and oauth1 imports to url_helper
* fix skew_data usage intending to use self.skew_data
Additionally:
* reorder imports in url_helper
* fixes to python3 -m cloudinit.sources.DataSourceMaas
LP: #1488507
|
|
|
|
|
|
oddly enough, the timestamp you pass into oauthlib must be a None
or a string. If not, raises ValueError:
Only unicode objects are escapable. Got 1426021488 of type <class 'int'>
|
|
|
|
clock skew
This functionality has been introduced to fix LP: #978127, but was lost
while migrating cloud-init to python3.
|
|
In both python2 and python3,
This throws "'module' object has no attribute 'oauth1'"
$ python3 -c 'import oauthlib; oauthlib.oauth1.Client("x")'
While this works fine:
$ python3 -c 'import oauthlib.oauth1 as oauth1; oauth1.Client("x")'
|
|
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.
|
|
|
|
|
|
|
|
to be behind trunk.
`tox -e py27` passes full test suite. Now to work on replacing mocker.
|
|
Couple of things here:
* do not re-try on user-data (404 means 'not here')
* re-generate headers on retry requests
LP: #1172742
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Use only util methods for reading/loading/appending/peeking
at files since it is likely soon that we will add a new
way of adjusting the root of files read, also it is useful
for debugging to track what is being read/written in a central
fashion.
|
|
This fixes (tested) bug 978127. The server was actually returning a 401
not a 403. As such, the fix here was insufficient. This will now take
either of those 2 error codes. I've also tested it by changing the clock
in the cloud-init upstart job with a stanza like below, and verifying
that we do see the problem and then it resolve itself:
pre-start script
offset="10 minutes ago"
past=$(date -R --date "$offset")
date --set "$past" &&
echo ===== "set date to $past [$offset]" ===== ||
echo ===== "failed to set date to $past [$offset]" ====
end script
LP: #978127
|
|
|
|
|
|
In the event of a 403 (Unauthorized) in oauth, try set a 'oauth_clockskew'
variable. In future headers, use a time created by 'time.time() +
self.oauth_clockskew'. The idea here is that if the local time is bad (or even
if the server time is bad) we will essentially use something that should be
similar to the remote clock.
This fixes LP: #978127.
LP: #978127
|
|
at files since it is likely soon that we will add a new
way of adjusting the root of files read, also it is useful
for debugging to track what is being read/written in a central
fashion.
|
|
the main function, which was usable for debugging maas was dropped during the
rework branch. I'm adding it back here as it is very useful. It is possibly
better implemented some other way than this, but this is good enough.
|
|
|
|
2. Adjust comment on sources list from depends
3. For the /etc/timezone 'writing', add a header that says created by cloud-init
|
|
|
|
a. This allows for more properties to be added as needed in the future, instead of being very restrictive.
2. Fix up all uses of the url reading to now use this new response object.
3. Also fixup user data including, such that if no response actual occurs the url content is not further processed.
|
|
|
|
|
|
Some of the cleanups were the following
1. Using standard (logged) utility functions for sub process work, writing, reading files, and other file system/operating system options
2. Having distrobutions impelement there own subclasses to handle system specifics (if applicable)
3. Having a cloud wrapper that provides just the functionality we want to expose (cloud.py)
4. Using a path class instead of globals for all cloud init paths (it is configured via config)
5. Removal of as much shared global state as possible (there should be none, minus a set of constants)
6. Other various cleanups that remove transforms/handlers/modules from reading/writing/chmoding there own files.
a. They should be using util functions to take advantage of the logging that is now enabled in those util functions (very useful for debugging)
7. Urls being read and checked from a single module that serves this and only this purpose (+1 for code organization)
8. Updates to log whenever a transform decides not to run
9. Ensure whenever a exception is thrown (and possibly captured) that the util.logexc function is called
a. For debugging, tracing this is important to not just drop them on the floor.
10. Code shuffling into utils.py where it makes sense (and where it could serve a benefit for other code now or in the future)
|
|
|
