Example command arg `boot_next` missing the underscore
(cherry picked from commit 4fe5d54b9e)
Co-authored-by: bluikko <14869000+bluikko@users.noreply.github.com>
* Fix exception in xattr module when existing extended attribute's value contains non-printable characters and the base64-encoded string contains a '=' sign
* Added changelog fragment for #3675
* Apply suggestions from code review
Co-authored-by: Felix Fontein <felix@fontein.de>
(cherry picked from commit 2f0ae0408d)
Co-authored-by: sc-anssi <sc-anssi@users.noreply.github.com>
* Add test which allows to query whether an action is available.
* Add documentation.
* Fix FreeBSD / macOS compatibility.
* Rename an_action -> a_module.
* Forgot one place.
* Fixed - TypeError: unexpected keyword argument
- File proxmox_group_info.py creates the error "TypeError:
get_group() got an unexpected keyword argument \'group\'\r\n'" if a
group parameter is used.
Issue is an argument naming conflict. After changing the argument
name to 'groupid', as used in method ProxmoxGroupInfoAnsible::get_group,
testing a Proxmox group name is working now.
* Changelog fragment added for #3649
changelog fragment for TypeError: unexpected keyword argument #3649
* Update changelogs/fragments/3649-proxmox_group_info_TypeError.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
While porting this module to make use of `pyone` I have overlooked one
attribute. Luckily the error only occurs when trying to rename an image
to a name that has already been taken.
Instead of telling the user which image ID already uses that name, the
module failed with the following error (along with a huge backtrace):
AttributeError: 'IMAGESub' object has no attribute 'id'
With this commit the error message is much more obvous again.
* add the possibility to restart all supervisorctl programs and groups
Signed-off-by: Thomas Sjögren <konstruktoid@users.noreply.github.com>
* Update changelogs/fragments/3551-supervisor-all.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
* add info about "all" in the name parameter description
Signed-off-by: Thomas Sjögren <konstruktoid@users.noreply.github.com>
* Update plugins/modules/web_infrastructure/supervisorctl.py
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
* provide more fitting description for runner timeout
* Update plugins/modules/source_control/gitlab/gitlab_runner.py
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Tim Herren <tim.herren@gmx.ch>
Co-authored-by: Felix Fontein <felix@fontein.de>
* KeycloakClientDiffBugs - Introduce test that passes.
* KeycloakClientDiffBugs - Add test to show that checking of redirect_uri's fails.
* KeycloakClientDiffBugs - (Fix1) Update so that checking of `redirectUris` no longer shows a change.
* KeycloakClientDiffBugs - Add test to show that checking of attributes's fails (sorting issue)
* KeycloakClientDiffBugs - (Fix2) Update so that checking of `attributes` no longer shows a change.
* KeycloakClientDiffBugs - Add test to show that checking of protocol_mappers's fail
* KeycloakClientDiffBugs - (Fix3) Update so that checking of `protocol_mappers` no longer shows a change when there is none.
* Introduce code fragment.
* Update the changelog to be based on the PR instead of the issue.
* Fix the readme
* Fix yaml indentation.
* Fix pep8
* Update changelogs/fragments/3610-fix-keycloak-client-diff-bugs-when-sorting.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update changelogs/fragments/3610-fix-keycloak-client-diff-bugs-when-sorting.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update plugins/modules/identity/keycloak/keycloak_client.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Remove need for .copy() after making normalise_cr not mutate the dict.
Co-authored-by: Pierre Dumuid <pierre@knowyourdata.com.au>
Co-authored-by: Felix Fontein <felix@fontein.de>
* Allow for the pacman repository check to be skipped when it's not needed
* Adding a changelog fragment
* Undo the variable rename because the lint doesn't like it
* Update changelogs/fragments/3606-pacman-speed-up-check-if-package-is-installed.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Elie Moreau <emoreau@atlassian.com>
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update gitlab_project_members.py
The actual search method doesn't accept path with namespace for project_name. If you have many project with same name, this module gitlab_project_members can't work.
* Update gitlab_project_members.py
* Update gitlab_project_members.py
* Update gitlab_project_members.py
* Create 3602-fix-gitlab_project_members-improve-search-method
* Rename 3602-fix-gitlab_project_members-improve-search-method to 3602-fix-gitlab_project_members-improve-search-method.yml
* Add initialize_with_readme option to gitlab_project. Update integration/unit tests.
* Fix pep8
* Revert unit tests
* Update plugins/modules/source_control/gitlab/gitlab_project.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update plugins/modules/source_control/gitlab/gitlab_project.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update plugins/modules/source_control/gitlab/gitlab_project.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Move initialize_with_readme to proper spot to only add if project does not exist.
* Update plugins/modules/source_control/gitlab/gitlab_project.py
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
* pipx - fixed bug in state=inject
* added changelog fragment
* copy/paste error in the integration test
* replaced injected package with simpler one
* testing force_lang = None
* disable UTF-8 emojis in pipx output
* better way to achieve the same outcome
* Adjsuted the changelog fragment
* added module mssql_script
* Add changelog fragment
* fixed fragment file (yamllint)
* added me as maintainer
* minor cleanup. (PR comments)
* adjusted for PR comments
- changed key for query results depending on output
- removed "empty" string defaults
* convert unserializable values to string
* python 2 support
* Apply suggestions from code review
Co-authored-by: Felix Fontein <felix@fontein.de>
* updates to docs
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
* dnf_versionlock: new module
* dnf_versionlock: fix style in doc
* dnf_versionlock: use check_rc in run_command
* dnf_versionlock: fix style and typos in doc
* Consistent Modules - Rename updated_?? to desired_?? in all the keycloak modules.
* Consistent Modules - Rename the comments, and add whitespace so that all the modules are a lot more consistent between each other.
* Consistent Modules - Remove final elif where a final else doesn't exist.
This is to address the inconsistency between the other modules.
Whilst I can see it being more descriptive, there should be a final "else:" to cater if the values is neither 'absent' or 'present'.
* Consistent Modules - Use dict() instead of {} like most of the other keycloak modules.
* Consistent Modules - Update keycloak authentication so that the if ordering is consistent for no-item.
* Consistent Modules - Move the 'Filter and map' process to always occur before getting an existing item.
* Consistent Modules - Be consistent with how to initialse before_?? and set it to dict() if it is None.
* Consistent Modules - Add module.exit_?? in the locations as per the other modules.
* Consistent Modules - Represent result['diff'] using dict(before=.., after=...) as per all the other modules.
* Consistent Modules - Add / Move location of when result['end_state'] is getting defined.
* Consistent modules - Add result['changed'] = False where we do nothing and exit because item exists.
* Consistent Modules - Set the value result['changed'] to True earlier so it shows up when in checking mode only.
* Consistent Modules - test for equality with a dict to assert there was no realm in the first place as per the other modules.
* Consistent Modules - Address the spelling.
* Consistent Modules - keycloak_group - Remove result['group'] as result['end_state'] is the consistent value used in the other modules.
* Consistent Modules - Order the lines in the section, Do nothing and exit consistently.
* Consistent Modules - Add result['end_state'] and still add deprecated `flow` return value.
* Consistent Modules - Add missing return documentation for `msg`.
* Consistent Modules - Tweak whitespace in the RETURN variable.
* Consistent Modules - Add result['group'] in addition to deprecated result['group'] response.
* Consistent Modules - Add return property, 'contains' to address test errors.
* Consistent Modules - Rename updated_?? to desired_?? in new modules since initial PR.
* Consistent Modules - Rename the comments, and add whitespace so that all the (recently added) modules are a lot more consistent between each other.
* Consistent Modules - Make indentation consistent within the response document.
* Consistent Modules - Use B(DEPRECATED) in a seperate line in the description.
* Consistent Modules - Add a lot of full stops to sentences.
* Consistent Modules - Use C(...) and I(...) formatting methods.
* Consistent Modules - Use "on success" everywhere for end_state response documentation.
* Consistent Modules - Update the documents for RETURN.proposed, RETURN.existing, RETURN.end_state to be the same.
* Consistent Modules - Add fragment.
* Remove period after short_description.
* Update changelog fragment.
* Consistent Modules - PRFeedback - Remove `module.exit_json(**result)` within the `Delete` section of the if statement.
There's a exit_json(..) immediately after.
* Consistent Modules - PRFeedback - Use `if not x_repr` instead of `if x_repr == dict()`.
* keycloak_authentication - Add a sample of the output.
* Replace `dict()` with `{}` for all the keycloak modules.
* Add the requested deprecated notices
* Update changelogs/fragments/3280-keycloak-module-cleanup-and-consistency.yml
Co-authored-by: Pierre Dumuid <pierre@knowyourdata.com.au>
Co-authored-by: Felix Fontein <felix@fontein.de>
* Added redis_incr module
Added redis_incr module and unit tests for the module.
Added suggested changes to increment_float, increment_int stuff
* rebased to main, version fix and unit test fix
* Renamed module,added check_mode and fixed retun value
* Update .github/BOTMETA.yml
* check_mode fix and version bump
* fix for lower Redis versions
* Update plugins/modules/database/misc/redis_data_incr.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Fix exception in pkgin module when all packages are already installed.
* Update plugins/modules/packaging/os/pkgin.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Add changelog fragment for #3583
Co-authored-by: Felix Fontein <felix@fontein.de>
* pkgng: package reinstallations count as changed
`upgrade_packages()` only looked for the string
"Number of packages to be upgraded", but the
`pkg upgrade` command also reports "Number of packages to be
reinstalled". Reinstallation occurs when package metadata other
than version changes (e.g. build options, single architecture to `*`
architecture). In any other respect, though, a required
reinstallation is the same as an upgrade.
* pkgng: check_mode should count queued actions
Writing tests caught a bug in PR #3393, which enabled
installing more than one package per `pkg` execution.
In converting the module's install/upgrade code to a
queue structure, check_mode got broken because the count
of actions performed was only updated in the `if not check_mode`
block that invokes `pkg`. This two-line change counts
the number of actions in the queue when check mode is
enabled.
* pkgng: add basic integration tests
Test installing a package
Test installing a package into a jail
* pkgng: test with zsh not bash package
Had been using bash package, because it's not
likely to disappear from the package repository
any time soon. Turns out that bash is already
installed on the Ansible community.general
FreeBSD CI VM image, which makes the test fail.
Zsh probably isn't going away any time soon,
either, and likely won't be installed on the
CI image.
* pkgng: remove redundant test for docker container
These tests should be skip/docker, but the test
playbook also redundantly checked whether it was
running in a docker container.
Checking whether `ansible_facts.distribution` is
`FreeBSD` is really sufficient to be sure whether
the test is running in an environment that supports
the `pkgng` module.
* pkgng: add state=absent test
Also renumber `jail=...` test to make diffs less
noisy when adding more non-jailed tests.
* pkgng: add state=latest idemptency test
Add test between state=present and
state=absent that ensures state=latest runs
successfully on an already up-to-date package
with changed=false.
* pkgng: add state=latest test
Including small filesize out-of-date package from
FreeBSD 11 package repository, because FreeBSD 11 is
currently EOL. This test might fail at some point in the
future if the pkg utility in a FreeBSD version > 14
makes breaking changes that prevents it from installing
older package formats.
If that occurs, the fix is to replace
`files/freebsd-release-manifests-20210413__FreeBSD_11.pkg`
with the version of freebsd-release-manifests from the oldest
non-breaking release of FreeBSD, and update the
references to it in tasks/main.yml accordingly.
* pkgng: use hand-generated test package
Instead of relying on a broken-ish installation
of an older package from FreeBSD 11, hand-generate
a package named `zsh` with no contents and a version
number of `0`. It can be installed on any architecture
and any revision of FreeBSD sucessfully, and it will
always be eligible for upgrade.
* pkgng: CI seems to dislike plus-signs in filenames
* pkgng: refactor renaming test package manifests
refactored to make intent of adding '+' sign clearer
* pkgng: refactor upgraded pattern matching
Implement russoz's suggestion to put all
variants in the pattern.
* pkgng: add cached=yes test
Changed idempotency test, because this covers both.
* pkgng: test pkg works on any FreeBSD version
Removing `ansible_distribution_version`, because the
test out-of-date zsh package should install on any
FreeBSD version.
* pkgng: move FreeBSD tasks to imported file
Refactoring tests for more reuse and easier
readability.
* pkgng: refactor tests for task reuse
Several tests need to install an out-of-date package
to test whether upgrades occur successfully, so this
commit refactors out the generation of the test package.
Also, factor out things like the name and path of the
test package so that is more maintainable should the
target test package (currently `zsh`) ever need to
change.
* pkgng: test install multiple packages
Multiple packages should install with a single
invocation of the `pkg` utility.
* pkgng: handle space- and comma-separated lists
The module expects a list of strings in the `name` parameter,
but long-standing documentation showed space- and comma-delimited
lists as a valid way of telling the module to act on multiple
packages. Passing these lists through to the `pkg` command can
have unexpected side-effects of upgrading packages when
`state=present` rather than `state=latest` and could result
in the module reporting the wrong number of packages for each
action performed.
* pkgng: test state=latest for mixed install/upgrade
Test that a list of packages given, one not installed
and one installed but out-of-date are handled correctly.
* pkgng: use YAML lists for multiple packages
I had been following the documentation examples, and
specifiying multiple packages on one line. The right way
to do it is with YAML list syntax.
* pkgng: add test for autoremove=yes
Install package with known dependencies, then
remove it and ask autoremove to run.
* pkgng: test autoremove=yes only on FreeBSD > 12
The CI test runner on FreeBSD 12.0 times out when
downloading large packages.
* pkgng: test jail creation logging less verbose
`ezjail-admin` had been spewing every filename copied
into the jail onto stderr, making the `ansible-test -v`
logs hard to scroll through. Changed it so that ezjail
output only shows up in the ansible task logs if
`ansible_verbosity > 1` (`-vv...`). Full `ezjail-admin`
output is always logged to `/tmp/ezjail.log`.
* pkgng: pass tests when package dependencies change
Tests that install packages with dependencies were failing
when those dependencies were not already installed, because
the count of installed packages was greater than the count
of requested to be installed packages. This change checks
for a count of installed packages that is greater than or
equal to the count of requested to be installed packages.
Example:
- Test installs packages `zsh` and `fish`
- `fish` has a dependency on `pcre2`
- `pkg` reports `Number of packages to be installed: 3`
* pkgng: test annotation for single package
Add/modify/remove annotation for a single package
* pkgng: fix annotation operations
Annotation has been broken at least since the migration to
collections. There are some breaking typos and function argument
omissions in the code with nothing in `git blame` but "Initial commit".
New integration tests uncovered this breakage. Works now.
* pkgng: test multiple annotations
Test multiple annotations specified both in new (YAML list)
syntax and old (comma-delimited) syntax.
Rename some annotation test tags from the single-annotation
tests to make sure that single and multiple annotation tests
don't collide.
* pkgng: test invalid annotate strings
Ensure that some likely to occur, invalid strings
given to the annotate module parameter 1. fail,
2. do not cause an exception, and 3. do not create
any actual annotations on packages.
* pkgng: fix check_mode for annotate
Actions specified in the `annotate` parameter would
always be performed, even if `check_mode=yes`.
This commit fixes `check_mode` for the annotation
functions and adds integration tests to ensure that
check mode is honored in the future.
* pkgng: call module.run_command with list
Instead of calling `run_command` with a formatted
string, call it with a list to maintain safer argument
separation.
Also, introduce a wrapper for `run_command`, `run_pkgng`,
which manages the process environment and some common
command line switches, based upon the module parameters.
Introduced in this commit, also pass annotation values
to `pkg annotate` via stdin, which is safer with long
values than putting them in argv.
* pkgng: update documentation to match annotate type
Missed updating the documentation to match the change
of the annotate parameter from comma-separated string
to list of strings.
* pkgng: fix syntax for Python 2
Seems *args, **kwargs have to be the last
two arguments in Python 2 syntax. CI sanity
tests were failing.
* pkgng: oops, pkgsite should be from closure
I put `pkgsite` as a keyword argument to `run_pkgng()',
which wasn't correct at all and resulted it in capturing
the second positional argument given by any caller.
`pkgsite` should have been `p["pkgsite"]`, coming from the
closure environment of `main()`.
* pkgng: Fix changelog fragment quote formatting
Co-authored-by: Felix Fontein <felix@fontein.de>
* pkgng: add test prepare task
Make sure test environment does not contain the
test package.
* pkgng: make integration test package more flexible
Make integration test package building template-based,
so more flexible if in the future the name of the test package,
currently `zsh`, must change.
* pkgng: convert generator to list for 2.9
Ansible 2.9 can't count items from a generator,
so convert `select` filter output via `list`
before passing to `count`.
* pkgng: Ansible 2.9 has no "false" Jinja2 test
Apparently Ansible 2.9 doesn't have the "false" test
in Jinja2 contexts. Switching to use `rejectattr(...)`
instead of `selectattr(..., "false")`.
* pkgng: test pkgsite parameter
* pkgng: add tests for chroot and rootdir params
Refactor install tasks into an include file so they
can be reused and tested with various parameters.
* pkgng: clean up test package creation
Make pkg json manifest more readable. Create package using
FreeBSD's `pkg create` instead of manually using tar. This
change also simplifies the manifest to a single file for
the integration test role.
* pkgng: fix#3428; install from local file
Fixes a bug in which the module reported failure
when installing from a local (to the target host)
file path.
Fixes#3428
* pkgng: changelog fragment formatting issue
* pkgng: check for all test package extensions
pkg recently changed file extensions, so for the
tests to work on multiple recent versions of FreeBSD,
we must check for `pkg create` output with varying
filename.
* Revert "pkgng: fix#3428; install from local file"
As the module has invoked pkg with the `-g` flag for
at least 7 years, I'm not sure when it was possible
to install packages from files on the target host
filesystem, because pkg rejects file paths when
the `--glob` flag is enabled.
I considered doing a rework of the pkg invocation
to enable good support for installing from local
files, but it looks like more of a job than for this PR.
This reverts commit 5f94eac41f.
* pkgng: Add minor_changes fragment for #3526
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
* Redfish: Do not set the boot source override mode if not provided by the user
Signed-off-by: Mike Raineri <michael.raineri@dell.com>
* Corrected changelog file extension
Signed-off-by: Mike Raineri <michael.raineri@dell.com>
* Update changelogs/fragments/3509-redfish_utils-SetOneTimeBoot-mode-fix.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
* nmcli: Fix ipv6.dns not being recongnized as list
There was a missing comma on the previous line.
* nmcli: Add changelog fragment for #3563
* nmcli: Update changelogs/fragments/3563-nmcli-ipv6_dns.yaml
Make the fix description more descriptive
Co-authored-by: Ajpantuso <ajpantuso@gmail.com>
* nmcli: Remove ipv4.route-metric from list-typed properties
Co-authored-by: Ajpantuso <ajpantuso@gmail.com>
* nmcli: Update fragment 3563 with ipv4.route-metric bug
Co-authored-by: Ajpantuso <ajpantuso@gmail.com>
* Redfish: perform manager network interface configuration even if property is missing
Signed-off-by: Mike Raineri <michael.raineri@dell.com>
* Update changelogs/fragments/3404-redfish_utils-skip-manager-network-check.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
