| Age | Commit message (Collapse) | Author |
|---|
|
When cloud-init ran in the init stage (after networking had come up).
A bug could occur where cloud-init would attempt and fail to rename
network devices that had "inherited" mac addresses.
The intent of apply_network_config_names was always to rename only
the devices that were "physical" per the network config. (This would
include veth devices in a container). The bug was in creating
the dictionary of interfaces by mac address. If there were multiple
interfaces with the same mac address then renames could fail.
This situation was guaranteed to occur with bonds or vlans or other
devices that inherit their mac.
The solution is to change get_interfaces_by_mac to skip interfaces
that have an inherited mac.
Also drop the 'devs' argument to get_interfaces_by_mac. It was
non-obvious what the result should be if a device in the input
list was filtered out. ie should the following have an entry for
bond0 or not. get_interfaces_by_mac(devs=['bond0'])
LP: #1669860
|
|
Mock the use use of get_interfaces_by_mac in Digital Ocean and OpenNebula.
Its best to mock this for the tests as the results aren't expecting
it to fail.
Note, as it stands, OpenNebula relies on devices named 'eth0'.
The metadata (context) does not provide mac addresses.
|
|
When rendering network configuration to netplan, remove known
"builtin" configurations. The specific example here is Ubuntu Core
that has netplan configuration in etc/netplan/00-snapd-config.yaml.
We also delete the derived files since netplan will have created
these derived files in its generator that runs well before cloud-init.
LP: #1675576
|
|
While documentation indicates that the smbios product name should
contain 'Google Compute Engine', experimentation and bug reports
indicate that is not always the case. The change here is to change
the check for GCE to also consider a serial number that starts with
'GoogleCompute-'.
Also, ds-identify was not currently searching for GCE if no config of
datasource_list was found. Most images have a datasource_list defined.
So update the list to include GCE.
LP: #1674861
|
|
This change will add support for hashed passwords in cc_set_passwords.
It checks if a password is a hash with by checking that it matches
in fairly safe way, and also that the password does not have a ":" in it.
chpasswd needs to know if the password is hashed or not, so two lists
is created so chpasswd is feed with the correct one.
LP: #1570325
|
|
Accordingly to the documentation:
The ``partition`` option may also be set to ``auto``, in which this
module will search for the existance of a filesystem matching the
``label``, ``type`` and ``device`` of the ``fs_setup`` entry and
will skip creating the filesystem if one is found.
However, using this "auto" flag always recreates the partition no matter
if it has been done before or not.
This commit fixes a bug in which the "partition" attribute was always
set to None although in some cases it should not.
LP: #1634678
|
|
ef18b8ac4c added support for handing network config v2 (aka netplan
format). This just adds that feature to the list of supported features.
|
|
test_netconfig.py:test_apply_network_config_eni_ub would attempt to
remove any .link files that cloud-init had written. This was just
a failure to mock out all of its interaction with the host.
|
|
Network configuration version 2 format is implemented in a package
called netplan (nplan)[1] which allows consolidated network config
for multiple network controllers.
- Add a new netplan renderer
- Update default policy, placing eni and sysconfig first
This requires explicit policy to enable netplan over eni
on systems which have both (Yakkety, Zesty, UC16)
- Allow any network state (parsed from any format cloud-init supports) to
render to v2 if system supports netplan.
- Move eni's _subnet_is_ipv6 to common code for use by other renderers
- Make sysconfig renderer always emit /etc/syconfig/network configuration
- Update cloud-init.service systemd unit to also wait on
systemd-networkd-wait-online.service
1. https://lists.ubuntu.com/archives/ubuntu-devel/2016-July/039464.html
|
|
Previously, the distro had hard coded which network renderer it would
use. This adds support for just picking the right renderer based
on what is available.
Now, that can be set via a priority in system_info, but should
generally work. That config looks like:
system_info:
network:
renderers: ["eni", "sysconfig"]
When no renderers are found, a specific RendererNotFoundError is raised.
stages.py is modified to catch that and log it at error level. This
path should not really be exercised, but could occur if for example an
Ubuntu system did not have ifupdown, or a rhel system did not have
sysconfig. In such a system previously we would have quietly rendered
ENI configuration but that would have been ignored. This is one step
better in that we at least log the error.
|
|
This adds an integration test for setting passwords when given
as a list rather than a string. This also updates the docs and
tests so that Random is now RANDOM as is correct.
|
|
render_network_state should default to rendering on /.
The changes here just make it so render_network_state does not
require a target, but defaults to None, and uses target_path
to handle that.
|
|
As reported in bug 1671927, sysconfig had an issue with rendering
a loopback device. The problem was that some as yet unknown issue was
causing the openstack config drive to parse the provided ENI file rather
than reading the network_data.json. Parsing an ENI file would add a
a 'lo' device of type 'physical', and sysconfig was failing to render
that.
The change here is:
a.) add a 'loopback' type rather than 'physical' for network config.
{'name': 'lo', 'type': 'loopback', 'subnets': ['type': 'loopback']}
b.) support skipping that type in the eni and sysconfig renderers.
c.) make network_state just piggy back on 'physical' renderer for
loopback (this was what was happening before).
Tests are added for eni and sysconfig renderer.
|
|
Use inspect.getmro(mod) rather than mod.__bases__ to test if a potential
testcase class inherits from CloudTestCase. This allows testcases to be
based on CloudTestCase indirectly, adding greater flexibility to the
structure of test classes.
|
|
When testing for timezone we were testing that 'date' output would
contain 'HDT' for the current time. But after a 'spring forward', the
current time started to have 'HST'.
Instead of asking 'date' for the timezone that applies now, ask it
for a static date.
|
|
Previously, cloud-init would throw an exception if an interface had
both ipv4 and ipv6 addresses and a default gateway for each address
family. This change allows cloud-init to correctly configure
interfaces in this situation.
LP: #1669504
|
|
log a warning rather than raising ValueError if we see more than three
nameserver addresses.
LP: #1670052
|
|
|
|
There were some logic errors in sysconfig.py that appear to be the
result of accidentally typing "iface" where it should have been
"iface_cfg". This patch corrects those problems so that the module
can run successfully.
LP: #1665441
Resolves: rhbz#1389530
|
|
When the ds-identify code landed, it started writing /run/cloud.cfg
but at the moment, nothing was reading that. The result is that
ds-identify only worked to disable cloud-init entirely.
|
|
The iface:alias syntax for eni rendering is brittle with ipv6.
Replace it with using multiple iface stanzas with the same iface
name which is supported. Side-effect is that one can no longer
do 'ifup $iface:$alias' but requires instead use of ip address
{add|delete} instead.
LP: #1657940
|
|
The test_user_data_normalize and test_net files had gotten
the executable bit set, and thus are skipped by nose by default.
We could set run with the --exe flag, but they should not have
gotten this way.
Other changes here:
* replace TempDirTestCase with CiTestCase, which has some nice
tmp_dir() and tmp_path() functions. Going forward the intent
is to have CiTestCase be the base test case for tests.
* test_net: switch to CiTestCase and fix usage that was
silently broken, because of exe bit.
* populate_dir: return the list of files that it writes rather
than having no return value.
* CiTestCase:
* support tmp_path("foo") that returns a full path to 'foo'
under a tmpdir.
* add tmp_dir() to get a temp dir and clean up.
|
|
No shipping ubuntu has a python 3.4 that is less than 3.4.2.
Remove this workaround to avoid unnecessary complexity.
This reverts 04a60cf949.
|
|
cloud-init adds ssh_authorized_keys to the default user and to
root but for root it disables the keys with a prefix command.
However, if the public_key key is of type ecdsa-sha2-nistp521,
it is not parsed correctly, and the prefix command is not prepended.
Resolves: rhbz#1151824
LP: #1658174
|
|
this ensures that we call httpretty.reset() before calling
httppretty.register_uri(...), which ensures that we get a fresh
callback with the expected version of the metadata.
LP: #1658200
|
|
On EC2, instance metadata can include credentials that remain valid for as
much as 6 hours. Reading these and allowing them to be pickled represents
a potential vulnerability if a snapshot of the disk is taken and shared as
part of an AMI.
This skips security-credentials when walking the meta-data tree.
LP: #1638312
Reviewed-by: Ian Weller <iweller@amazon.com>
Reviewed-by: Ben Cressey <bcressey@amazon.com>
Reported-by: Kyle Barnes <barnesky@amazon.com>
|
|
This modifies get_data in DataSourceOpenStack.py to get the timeout
and retries values from the data source configuration, rather than
from keyword arguments. This permits get_data to use the same timeout
as other methods, and allows an operator to increase the timeout in
environments where the metadata service takes longer than five seconds
to respond.
LP: #1657130
Resolves: rhbz#1408589
|
|
Failures to load the kernel command line's url (cloud-config-url=)
would previously get swallowed. This should make it much more
obvious when that happens. With logging going to expected places
at sane levels (WARN will go to stderr by default).
|
|
pycodestyle has better checking for 2 lines blank lines.
This failed to build on zesty as a result.
Patching this here, and filed bug 1652329 to fix it more permenantly.
|
|
The adds in end-to-end testing of cloud-init. The framework utilizes
LXD and cloud images as a backend to test user-data passed in.
Arbitrary data is then captured from predefined commands specified
by the user. After collection, data verification is completed by
running a series of Python unit tests against the collected data.
Currently only the Ubuntu Trusty, Xenial, Yakkety, and Zesty
releases are supported. Test cases for 50% of the modules is
complete and available.
Additionally a Read the Docs file was created to guide test
writing and execution.
|
|
This has been a recurring ask and we had initially just made the change to
the cloud-init 2.0 codebase. As the current thinking is we'll just
continue to enhance the current codebase, its desirable to relicense to
match what we'd intended as part of the 2.0 plan here.
- put a brief description of license in LICENSE file
- put full license versions in LICENSE-GPLv3 and LICENSE-Apache2.0
- simplify the per-file header to reference LICENSE
- tox: ignore H102 (Apache License Header check)
Add license header to files that ship.
Reformat headers, make sure everything has vi: at end of file.
Non-shipping files do not need the copyright header,
but at the moment tests/ have it.
|
|
The correct order of precedence when reading the base config:
builtin config
system config
kernel command line provided config.
This reverts commit 63501f44, which actually broke the behavior it
reported to fix. It also adds some unit tests to ensure this behavior
is not broken again.
LP: #1582323
|
|
This just adds a unit test for a case found to be failing in curtin.
The issue was reported under bug 1649652.
|
|
'nobootwait' is an upstart specific extension to the mount syntax that is
not supported by other mount systems. As Ubuntu 16.04 moved from upstart
to systemd, support for 'nobootwait' was lost.
All examples using 'nobootwait' are updated to use the standard 'nofail',
which gives the expected behaviour of not failing to boot in case a volume
is missing. There are subtle differences in semantics between
'nobootwait' and 'nofail', but it is the best substitute that gives
behaviour similar to the upstart specific option.
|
|
Cloud-config provided like:
users:
- default
- name: foobar
groups: sudo, adm
Would result in adduser being called as:
useradd foobar --groups 'sudo, adm' -m
Which would cause error:
useradd: group ' adm' does not exist
The fix here is just to always normalize groups and remove whitespace.
Additionally a fix and unit tests to explicitly set system=False
or no_create_home=True. Previously those paths did not test the value
of the entry, only the presense of the entry.
LP: #1354694
|
|
CloudSigma would not get any datasources loaded during cloud-init local.
Thus, when the network datasource was removed, *no* CloudSigma
datasources would be loaded.
LP: #1648380
|
|
I've seen cases of unable to read from files as
well as the existing os errors so catch io error
and skip by using the smarter read_sys_net instead.
LP: #1625766
|
|
These are just simple syntax fixes to work correctly on python2.6.
Found when testing in a centos 6 container.
|
|
The previous commit added tests that would fail on any system that had
a nic named eth0 or eno1. The changes here supply the expected macs to
the function being tested so it does not query the system.
LP: #1644043
|
|
The previous behavior would miss ip6= on the command line and
would not pay attention to the written net-* or net6-* files if
only ip6= was found.
The fix here enables parsing the files if either ip= or ip6= is found,
and adds some tests as well.
LP: #1639930
|
|
This replaces long single lines in a log or console output
with multiple lines that are much easier to read.
It indents the stdout and stderr so logs are more easily
read also.
|
|
|
|
A user has pointed out that upon set up of a machine, users typically
remove repos (sometimes all of them) and then add repos in. This does
make sense for a typical user.
|
|
Cloud-init has for some time relied on walinuxagent to do some bits
of work necessary for instance initialization. That reliance has
not been needed for a while, but we have still defaulted to it.
This change uses the "builtin" path that Daniel Watkins added
some time ago by default. Also, Adjust tests that assumed the
non-__builtin__ Azure agent_command.
LP: #1538522
|
|
The version of sfdisk in wily (and onwards) only accepts sectors as a
valid disk size. As such, this refactors the MBR code path in
cc_disk_setup to use sectors.
- use --unit=S: while newer versions of sfdisk assume --unit=S, older
versions do not so we specifically pass it in. Versions of sfdisk
found in supported OSes such as centos6 wont assume --unit=S.
- add --force: this exists back to centos 6 (2.17.2), so it should
be fine, and is what we ultimately want.
"do what I say, even if it is stupid"
- keep --Linux. Even though this has been deprecated for quite some
time, we keep it until versions that want it are unsupported.
If necessary at some point we could check for util linux version
and if it had --Linux and use it in those cases.
Additionally, improve usefulness of some log messages.
LP: #1460715
|
|
This adds a call to 'activate_datasource'. That will be called
during init stage (or init-local in the event of a 'local' dsmode).
It is present so that the datasource can do platform specific operations
that may be necessary. It is passed the fully rendered cloud-config
and whether or not the instance is a new instance.
The Azure datasource uses this to address formatting of the ephemeral
devices. It does so by
a.) waiting for the device to come online
b.) removing the marker files for the disk_setup and mounts modules
if it finds that the ephemeral device has been reset.
LP: #1611074
|
|
An obvious fix for an issue raised by pyflakes 1.3.
|
|
aarch64 systems have functional dmidecode, so allow that to be used.
- aarch64 has support for dmidecode as well
|
|
Support AliYun(Ali-Cloud ECS). This datasource inherits from EC2,
the main difference is the meta-server address is changed to
100.100.100.200.
The datasource behaves similarly to EC2 and relies on network polling.
As such, it is not enabled by default.
|
|
Many of the unit tests in test_data would inadvertantly read the
system's /etc/cloud/cloud.cfg and /etc/cloud/cloud.cfg.d.
This was first noticed on a system deployed by MAAS, where
files in /etc/cloud/cloud.cfg.d/ are root read-only.
This changes those tests to actually make use of
FilesystemMockingTestCase functionality and adds 'reRoot()' to that
class which is easier to use for at least this use case.
LP: #1635350
|
