| Age | Commit message (Collapse) | Author |
|---|
|
Resize of btrfs fails if the mount point for the file system we are trying
to resize, i.e. the root of the filesystem is read only. With this change
we use a known (currently snapper specific) rw location to work around a
flaw that blocks resizing of the ro filesystem.
LP: #1734787
|
|
Network configuration in OpenNebula would only work if the host correctly
guessed the names of the devices in the guest. OpenNebula provided data
in its context.sh like 'ETH0_NETWORK', but if the guest named devices
differently then results were not predictable. This would occur with
Predictable Network Interface Names. To address this,
newer versions (of OpenNebula provide the mac address ETH0_MAC.
This function is present in 4.14 and documented officially in 5.0 docs.
This provides support for reading the mac addresses from the context.sh.
It also fixes cases where context.sh provided a field (ETH0_NETWORK
or ETH0_MASK) with a empty string. Previously the empty string would
be used rather than falling back to the default.
LP: #1719157, #1716397, #1736750
|
|
The EC2 test platform uses boto, and boto decodes console output
with decode('utf-8', 'replace). It is known that Ubuntu consoles
contain non-utf8 characters, making this call lossy.
The change here is to patch the boto session to include a OutputBytes
entry in the console_output response, and then to utilize that in
console_log.
More information on problem and solution at:
https://github.com/boto/botocore/issues/1351
|
|
The previous commit added a test that would attempt to create and use
/run/cloud-init/. This just modifies it to use a temp dir instead.
|
|
The behavior changes and improvements include:
- Only import keys into the default user that contain the name of the
default user ('ubuntu', or 'centos') or that contain 'cloudinit'.
- Use instance or project level keys based on GCE convention.
- Respect expiration time when keys are set.
Do not import expired keys.
- Support ssh-keys in project level metadata (the GCE default).
As part of this change, we also update the request header when talking
to the metadata server based on the documentation:
https://cloud.google.com/compute/docs/storing-retrieving-metadata#querying
LP: #1670456, #1707033, #1707037, #1707039
|
|
When subp raised a ProcessExecutionError, that exception's stderr and
stdout might end up being the string '-' rather than bytes.
This mean that:
try:
subp(mycommand, decode=False)
except ProcessExecutionError as e:
pass
Would have 'e.stdout' set to '-' while the caller would expect bytes.
Also reduce the try/except block in subp to a specifically the two lines
that may raise an OSError.
|
|
Tests run on EC2 would successfully resolve the ipv4 dns address
and that caused false positives on failure reports.
Basically, dns lookup of 172.16.15.14 would return
ip-172-16-15-14.us-east-2.compute.internal.
which then shows up in the ntpq output unless you provide -n.
|
|
New mkfs.vfat and fatlabel tools included in the dosfsutils package no
longer support creating vfat disks with lowercase labels. They silently
default to an all uppercase label eg CONFIG-2 instead of config-2. This
change makes cloud-init handle either upper or lower case.
LP: #1598783
|
|
Zesty goes EOL as of January 13, 2017. This removes it as a valid
OS for testing.
|
|
This issue was first identified when manual_cache_clean was set, as
ds-identify would write /run/cloud-init/cloud.cfg with
# manual_cache_clean
that would generate a warning as cloud-init expected to load a dict.
Any other "empty" config would also log such a warning.
Also fix reading of di_report to allow it to be None, as ds-identify
would write:
di_report:
# manual_cache_clean
which reads as 'di_report: None' rather than di_report: {}.
LP: #1742479
|
|
This stores a hash of the OAuth tokens as an 'id' for the maas
datasource. Since new instances get new tokens created and those tokens
are written by curtin into datasource system config this will provide
a way to identify a new "instance" (install).
LP: #1712680
|
|
Due to additional platforms getting added this test was not taking into
account platform specific mirrors nor was it checking that no additional
entries were added.
|
|
This fixes the incorrectly named 'family' value for images as 'os'.
Families are already defined in util.py:OS_FAMILY_MAPPING and a family
is a collection of OSes.
This makes the properties function part of the super class of image as
it is only overrided by the lxd backend.
|
|
While the generated ssh keys are throw away keys, generating SSH keys
with the default name (e.g. id_rsa) can trigger security scanners or
draw unnecessary attention.
The change here simply renames 'id_rsa' to 'cloud_init_rsa' to avoid
a false positive reported by a scanning tool.
|
|
This enables integration tests to utilize AWS EC2 as a testing platform by
utilizing the boto3 Python library.
Usage will create and delete a custom VPC for every run. All resources
will be tagged with the ec2 tag, 'cii', and the date (e.g.
cii-20171220-102452). The VPC is setup with both IPv4 and IPv6
capabilities, but will only hand out IPv4 addresses by default. Instances
will have complete Internet access and have full ingress and egress access
(i.e. no firewall).
SSH keys are generated with each run of the integration tests with the key
getting uploaded to AWS at the start of tests and deleted on exit. To
enable creation when the platform is setup the SSH generation code is
moved to be completed by the platform setup and not during image setup.
The nocloud-kvm platform was updated with this change.
Creating a custom image will utilize the same clean script,
boot_clean_script, that the LXD platform uses as well. The custom AMI is
generated, used, and de-registered after a test run.
The default instance type is set to t2.micro. This is one of the smallest
instance types and is free tier eligible.
The default timeout for ec2 was increased to 300 from 120 as many tests
hit up against the 2 minute timeout and depending on region load can
go over.
Documentation for the AWS platform was added with the expected
configuration files for the platform to be used. There are some
additional whitespace changes included as well.
pylint exception was added for paramiko and simplestreams. In the past
these were not already flagged due to no __init__.py in the subdirectories
of files that used these. boto3 was added to the list of dependencies in
the tox ci-test runner.
In order to grab console logs on EC2 the harness will now shut down an
instance before terminating and before collecting the console log. This
is to address a behavior of EC2 where the console log is refreshed very
infrequently, but one point when it is refreshed is after shutdown.
|
|
This simply increases test coverage in rendering of network config
for SUSE and SLES.
|
|
This fixes a traceback when attempting to bounce the network after
hostname resets.
In artful and bionic ifupdown package is no longer installed in default
cloud images. As such, Azure can't use those tools to bounce the network
informing DDNS about hostname changes. This doesn't affect DDNS updates
though because systemd-networkd is now watching hostname deltas and with
default behavior to SendHostname=True over dhcp for all hostname updates
which publishes DDNS for us.
LP: #1722668
|
|
The cli help docs and argument parser allow the 'init' mode value
which caused a traceback.
Fix the cli to support 'init', 'config' and 'final' modes for the
cloud-init modules subcommand.
Add a check in the cli to raise a ValueError if a new
subcommand ends up allowing an unsupported/unimplemented modes.
Drive by unit test additions for a bit better coverage of error
handling.
LP: #1736600
|
|
The previous OVF datasource change added a debug message that referenced
an un-used variable. The failure path would be triggered if an image was
booted with a iso9660 filesystem attached to a device that was not a
cdrom.
A unit test is added for the specific failure found.
Additional safety to avoid 'cidata' labels is also added to the OVF
checker.
LP: #1737704
|
|
The exception was incorrectly creating a string and not a bytes object.
|
|
The instance identity document is a better source for region information,
partly because region isn't actually in meta-data at all, only
availability-zone, which happens to be named similarly.
Reviewed-by: Ethan Faust <efaust@amazon.com>
Reviewed-by: Cyle Riggs <cyler@amazon.com>
Reviewed-by: Tom Kirchner <tjk@amazon.com>
Reviewed-by: Matt Nierzwicki <nierzwic@amazon.com>
[ajorgens@amazon.com: rebase onto 0.7.9]
[ajorgens@amazon.com: changes per merge proposal discussions]
|
|
Make sure that some temporary files used by the config drive tests get
cleaned up properly.
|
|
Previously the OVF transport would not be identified except for when
config files set 'ovf_vmware_guest_customization'. It would also
return DS_MAYBE almost always.
The change here is to add support to ds-identify for storing the
iso9660 filesystems that it finds (ISO9660_DEVS). Then the OVF check
will check that the iso9660 filesystem has ovf-env.xml on it. The least
wonderful part of this is that the check is done by 'grep' for case
insensitive ovf-env.xml.
Future improvement would be to identify VMware's OVF by label or UUID
so we could avoid the grep.
LP: #1731868
|
|
In the VMware customization workflow, we have some options for the user
to upload scripts for additional customization. Based on user request,
those custom scripts can be either run before regular customization or
after. For post customization scripts, we decide whether to run the scripts
just after customization or post system reboot.
|
|
The NoCloudKVM platform was inserting ssh keys via user-data
rather than through meta-data like it is done on other platforms.
This way we are not forced to change the user-data provided.
Also, provide meta-data including a uuid as the instance-id.
|
|
The motivation for this is that
a.) 1.7.1 runs with python 3.6 (bionic)
b.) we want to run pylint on tests/ and tools for the same reasons
that we want to run it on cloudinit/
The changes are described below.
- Update tox.ini to invoke pylint v1.7.1.
- Modify .pylintrc generated-members ignore mocked object members (m_.*)
- Replace "dangerous" params defaulting to {}
- Fix up cloud_tests use of platforms
- Cast some instance objects to with dict()
- Handle python2.7 vs 3+ ConfigParser use of readfp (deprecated)
- Update use of assertEqual(<boolean>, value) to assert<Boolean>(value)
- replace depricated assertRegexp -> assertRegex
- Remove useless test-class calls to super class
- Assign class property accessors a result and use it
- Fix missing class member in CepkoResultTests
- Fix Cheetah test import
|
|
Each DataSource subclass must define its own get_data method. This branch
formalizes our DataSource class to require that subclasses define an
explicit dsname for sourcing cloud-config datasource configuration.
Subclasses must also override the _get_data method or a
NotImplementedError is raised.
The branch also writes /run/cloud-init/instance-data.json. This file
contains all meta-data, user-data and vendor-data and a standardized set
of metadata keys in a json blob which other utilities with root-access
could make use of. Because some meta-data or user-data is potentially
sensitive the file is only readable by root.
Generally most metadata content types should be json serializable. If
specific keys or values are not serializable, those specific values will
be base64encoded and the key path will be listed under the top-level key
'base64-encoded-keys' in instance-data.json. If json writing fails due to
other TypeErrors or UnicodeDecodeErrors, a warning log will be emitted to
/var/log/cloud-init.log and no instance-data.json will be created.
|
|
The 'cloud-init clean' command allows a user or script to clear cloud-init
artifacts from the system so that cloud-init sees the system as
unconfigured upon reboot. Optional parameters can be provided to remove
cloud-init logs and reboot after clean.
The 'cloud-init status' command allows the user or script to check whether
cloud-init has finished all configuration stages and whether errors
occurred. An optional --wait argument will poll on a 0.25 second interval
until cloud-init configuration is complete. The benefit here is scripts
can block on cloud-init completion before performing post-config tasks.
|
|
This groups up each test platform into its own directory rather
than having files spread between four different directories for
one platform. Platforms tend to be worked on one at a time and
so having the platforms together makes more sense than apart.
|
|
There is a race condition where our sandboxed dhclient properly writes a
lease file but has not yet written a pid file. If the sandbox temporary
directory is torn down before the dhclient subprocess writes a pidfile
DataSourceEc2Local gets a traceback and the instance will fallback to
DataSourceEc2 in the init-network stage. This wastes boot cycles we'd
rather not spend.
Fix handling of sandboxed dhclient to wait for both pidfile and leasefile
before proceding. If either file doesn't show in 5 seconds, log a warning
and return empty lease results {}.
LP: #1735331
|
|
The NoCloudKVMImage.execute() would modify the image in /srv/citest
that meant that after the first time you ran a test, the image was
dirty.
The change here is to make the image operate on a qcow backed image.
Also modify Snapshot to then copy the qcow rather
than creating another chained qcow. The reason being that the image
might go away or change after the snapshot has been returned.
Also
* drop use of 'override_templates' which was only relevant to LXD.
* NoCloudKVM.create_image() returned an instance before
now it has create_instance which creates an instance.
* NoCloudKVMInstance has a 'disk' attribute separate from 'name'
|
|
|
|
Instead of using 'dpkg -i' to install a package and then running
apt-get -f install, to hope that it would install needed dependencies
we can just use 'apt-get' directly to do the install.
The 'dpkg/apt-get -f' path was a problem if the installed deb was
older than the available deb. In that case it would get replaced.
|
|
Currently when dns and dns search info is provided, it is not rendered
when outputting to sysconfig format.
This patch causes the DNS and DOMAIN lines to be written out rendering
sysconfig.
LP: #1705804
|
|
Cloud-init integration tests should not depend on a curtin test ppa.
We already had a cloud-init test ppa for explicitly this purpose.
Just use it instead.
|
|
If user upgraded to new cloud-init and attempted to run 'cloud-init init'
without rebooting, cloud-init restores the datasource object from pickle.
The older version pickled datasource object had no value for
_network_config or fallback_nic. This caused the Ec2 datasource to attempt
to reconfigure networking with a None fallback_nic. The pickled object
also cached an older version of ec2 metadata which didn't contain network
information.
This branch does two things:
- Add a fallback_interface property to DatasourceEC2 to support reading the
old .fallback_nic attribute if it was set. New versions will
call net.find_fallback_nic() if there has not been one found.
- Re-crawl metadata if we are on Ec2 and don't have a 'network' key in
metadata
LP: #1732917
|
|
Add opensuse distro support to cc_ntp module.
LP: #1726572
|
|
Without this the entire stage can fail, which will leave an instance
unaccessible.
Reviewed-by: Tom Kirchner <tjk@amazon.com>
Reviewed-by: Matt Nierzwicki <nierzwic@amazon.com>
Reviewed-by: Ben Cressey <bcressey@amazon.com>
|
|
The etc/hosts file is was not properly setup for openSUSE or SLES
when manage_etc_hosts is set in the config file.
Improve the doc to address the fact that the 'localhost' ip is
distribution dependent (not always 127.0.0.1).
LP: #1731022
|
|
The rh_subscription module doesn't perform null checks when attempting to
iterate on the enabled and disable repos arrays. When only one is
specified, cloud-init fails to run.
|
|
We had used some dns records in i9n.brickies.net (my personal domain)
as a temporary solution until we got names registered in the cloud-init.io
namespace.
We now have CNAME records for:
ubuntu.i9n.cloud-init.io
cloudinit1.cloud-init.io
cloudinit2.cloud-init.io
|
|
3 things here:
a.) link to a bug that we opened to track what made us add
dns entries for hostname of our guests.
b.) spelling fix.
c.) raise an instance of a NotImplementedError not the class.
|
|
Integration test harness changes:
* Enable collection of console log in nocloud-kvm and lxd.
* Collect the console log to results for all test runs.
* change 'tmpfile' to pick name locally instead of using 'mktemp'.
* drop the 'instance' attribute from nocloud-kvm Image and
demote LXDImage.instance to a private attribute.
This is because Images do not actually have instances.
(LXDImage internally uses a booted system to modify the image).
* Add 'TargetBase' as a superclass of Image and Instance providing
implementations of execute, read_data, write_data, pull_file,
and push_file. These all depend on an implementation of _execute.
* Improve '_execute' implementations to support accepting stdin.
* execute supports 'rcs=False' meaning 'do not raise exception'.
* Drop support for pylxd < 2.2. older versions cannot determine
exit code of 'execute', which makes them unusable.
* make NoCloudKVMInstance._execute run as root via sudo. This required
some changes so that 'hostname' could be reverse-looked up in order
to avoid sudo taking a long time (~20 seconds).
* re-use existing ssh connection in nocloud-kvm.
Test changes here:
* do not use /tmp, but rather /var/tmp (LP: #1707222)
* make keys_to_console assertions more strict.
* change user test cases to always add default (ubuntu) user
so that nocloud-kvm's execute which operates over ssh can work.
|
|
VPC instances have the option to specific local only IPv4 addresses. Allow
Ec2Datasource to enable dhcp4 on instances even if local-ipv4s is
configured on an instance.
Also limit network_configuration to only the primary (fallback) nic.
LP: #1728152
|
|
A recent cleanup of the resizefs module broke resizing when a system was
booted with root=PARTUUID=<uuid> and the device /dev/root does not exist.
This path is exposed with the Ubuntu 16.04 but not with Ubuntu 17.10. A
recreate exists under bug 1684869.
LP: #1725067
|
|
There was fallout in a full integration test run from my adding of
test_no_warnings_in_log which asserted that there could not be a WARNING
found in the /var/log/cloud-init.log
This fixes 2 of the cases:
* TestCommandOutputSimple had a valid WARNING written, so adjust its
test case to allow for that.
* TestLxdDir had a valid config in the test but the module would
log a WARNING, so fix the module.
Also updates lxd unit tests to look for WARN themselves.
|
|
Tests are currently run by creating a temporary subclass of each class
and then executing it (in get_suites). When running the tests suite
the output would contain the temporary class name. That was less than
useful, and made batch runs almost impossible to identify which
test case had an error.
This change goes from output of:
FAIL: test_no_warnings_in_log \
(tests.cloud_tests.testcases.get_suite.<locals>.tmp)
To
FAIL: test_no_warnings_in_log \
(tests.cloud_tests.testcases.modules.ntp.TestNtp)
|
|
Fix three things related to the ntp module:
1. Fix invalid cloud-config schema in the integration test which
provided empty dicts instead of emptylists for pools and servers
2. Correct logic in the ntp module to allow support for the minimal
cloud-config 'ntp:' without raising a RuntimeError. Docs and schema
definitions already describe that cloud-config's ntp can be empty.
An ntp configuration with neither pools nor servers will be
configured with a default set of ntp pools. As such, the ntp module
now officially allows the following ntp cloud-configs:
- ntp:
- ntp: {}
- ntp:
servers: []
pools: []
3. Add a simple unit test which validates all cloud-config provided to
our integration tests to ensure it adheres to any defined module
schema so as more jsonschema definitions are added, we validate our
integration test configs.
LP: #1724951
|
|
When operating in expected path, cloud-init should avoid logging with
warning. That causes 'WARNING' messages in /var/log/cloud-init.log.
By default, warnings also go to the console.
Since jsonschema is a optional dependency, and not present on xenial
and zesty, cloud-init should not warn there.
Also here:
* Add a test to integration tests to assert that there are no
warnings in /var/log/cloud-init.log.
* Update one integration test that did show warning and the related
documentation and examples.
LP: #1724354
|
|
Update unit tests to pass a 0 instead of 'off' to validate that network
state is properly written.
|
