| Age | Commit message (Collapse) | Author |
|---|
|
Thanks to Ben Howard.
|
|
* removed the 'CLOUD-INIT-IGNORE' section, as we're just blindly writing
the file now. removed the now-unnecessary import of 're' and 'os'
* removed try/except block around write_apt_snippet. This will bubble
up and cloud-init will let it through even to the console. Catching
it and turning it into a debug would just hide it.
* removed 'default' as a synonym for 'whatever cloud-init thinks is best'
If people are going to change this, I'd rather they be specific.
* supported value of "0"
* fixed some complaints from
./tools/run-pylint cloudinit/CloudConfig/cc_apt_pipelining.py
|
|
- Changed values to be more simplistic and intuitive
- Only allow pipelining values up to 5
- Changed to per_instance over per_always to remove need
for tracking the values
- Fixed Python style
|
|
|
|
LP: #942061
|
|
|
|
|
|
|
|
main now is more useful for debugging. now it does:
* get: just dump contents of a url provided after oauth
* crawl: walk through using indexes, dumping content
* check-seed: validate the seed is good
uses MD_VERSION in the url, and appends that to the metadata url
in the config file. (previously it assumed the url in the config
was the full url)
does not require user-data in the http seed. if the user did not
specify user-data, it wont be there, so do not fail on that case.
|
|
- cloud-config option of "apt-pipelining"
- Address LP: 948461
|
|
|
|
This adds to the 'main' in cloudinit/DataSourceMaaS.py a method for
testing oauth_headers.
|
|
|
|
|
|
|
|
This commits a generally functional read_maas_seed_url, and re-works how
the content is checked, to share between read_maas_seed{url,dir}.
|
|
|
|
Also, add in the headers_cb which will be required for oauth.
|
|
Tests at this point seem to indicate that seed-dir would work.
|
|
Because Fedora's run-parts does not accept '--regex' and debian's
run-parts skips files with a '.' in the *without* '--regex=.*', we're
forced to include our own version of run-parts.
LP: #933553
|
|
if public-keys in metadata are a list (rather than a dict), then
return them as such.
This allows populating NoCloud keys with:
public-keys:
- ssh-rsa .... myname.mylabel
- ssh-dsa .... myname.mylabel2
Or:
cat <<EOF
public-keys:
$(ssh-add -L | sed 's,^\(.*\), - "\1",')
EOF
|
|
|
|
The purely local non-device (vfat/iso9660) sources were broken by the
last set of changes here. This restores them to functional.
If the seed is from a device, then the default behavior is to be 'net' mode.
For seed via cmdline, the user can specify 'ds=nocloud-net' and for
seed via filesystem seed dir, they can just populate the other directory.
To make it easier, when attaching a seed device, the user does not need
to specify 'dsmode' of 'net' in the metadata file. They still can, but that is
the default. It seems that that is more likely to be what is desired.
LP: #942695
|
|
Also improves the logic in is_container following ubuntu's
/etc/init/container-detect.conf .
LP: #941955
|
|
This branch also adds tests for part-handler registration and part-handler
handling.
|
|
|
|
|
|
|
|
This makes the user able to pass in multi-line input to the public-key flag,
and it will be handled correctly (just as if it came from the authorized_keys
file)
|
|
document usage of DataSourceNoCloud from vfat or iso disk.
|
|
Previously, the 'ifup --all' and update of /etc/network/interfaces
was done only if the specified 'dsmode' (which defaults to 'pass') was
either 'pass' or *this* dsmode.
That meant that it would be updated once on DataSourceConfigDrive
and on DataSourceConfigDriveNet.
This changes that to only happen on local.
|
|
Instead of a metadata entry named 'interfaces', use 'network-interfaces'
which is a somewhat less likely namespace collision.
|
|
|
|
This allows you to attach a disk in ISO9660 or vfat filesystem format
labeled 'cidata' with 'user-data' and 'meta-data' on it.
It provides a much easier way to interact with cloud-init in nocloud
than mounting the image or the OVF method.
|
|
|
|
|
|
I believe this will resolve LP: #898373 by making DataSourceOVF restrict
its mounting of filesystems to iso9660. By doing this, it will never
mount a ext3 (or btrfs or any fs that mountall would fsck) and thus will
avoid any races with that.
LP: #898373
|
|
|
|
|
|
LP: #923043
|
|
|
|
If you were trying to parse this output with something, an empty field would be
difficult to handle, as you'd have to know the expected lengths of each field.
The '.' means empty, but then all fields are non-whitespace delimited by one or
more whitespace.
|
|
This copyright change reflects previous changes that Juerg made for pylint and
pep8 cleanups.
From: Juerg Haefliger <juerg.haefliger@hp.com>
Date: Mon, 16 Jan 2012 10:45:12 +0100
|
|
LP: #915232
|
|
|
|
|
|
This is actually a pylint bug, but it considers use of string.letters
and string.whitespace deprecated.
|
|
|
|
|
|
From: Juerg Haefliger <juerg.haefliger@hp.com>
This pulls in the named patch for LP: #914739 with a few other changes.
|
