When the `vmid` parameter is not supplied and the module can only rely on
name look-up an early failure can happen if the targeted VM doesn't exist.
In this case a task execution with the parameter `state` set to `absent`
will actually fail instead of being considered ok.
This patch introduces a deferred error-checking for non-existent VMs
by assigning the value -1 to the `vmid` parameter, allowing the actual
verification to be performed in the right code paths.
Is also help to differentiate between a non-existent `vmid` or non-existent
VM `name`.
Previously:
TASK [ansible-role-proxmox-instance : Remove instance-test]
changed: [localhost]
...
TASK [ansible-role-proxmox-instance : Remove instance-test]
fatal: [localhost]: FAILED! => changed=false
msg: VM instance-test does not exist in cluster.
Now:
TASK [ansible-role-proxmox-instance : Remove instance-test]
ok: [localhost]
...
TASK [ansible-role-proxmox-instance : Remove instance-test]
ok: [localhost]
Update changelogs/fragments/811-proxmox-kvm-state-absent.yml
With suggestions from Felix Fontein <felix@fontein.de>.
* Specify device for Pushover notification
New parameter: device
Example:
- community.general.pushover:
msg: '{{ inventory_hostname }} has been lost somewhere'
app_token: wxfdksl
user_key: baa5fe97f2c5ab3ca8f0bb59
device: admins-iPhone
delegate_to: localhost
Using the Pushover API, you can specify a device where the message should be delivered to. Instead of notifying all devices (the default), the message is sent only to the specified device. Multiple devices can be given separated by a comma.
This change is downwards compatible: omitting the device key sends the message to all devices (as before).
* Added changelog fragments file for pushover
File format as specified in https://docs.ansible.com/ansible/devel/community/development_process.html#changelogs-how-to.
* Added version_added information
As suggested by Felix (thanks!).
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
* Fix get_installed_versions: correctly parse "default" version.
gem query output of
bundler (default: 2.1.4, 1.17.2)
Gets parsed as:
['default:', '1.17.2']
Fix this by skipping "default: " if present in the list of versions - by adding
it as an optional part of the regex, grouped as a non-capturing group to keep
the index of existing group.
This now correctly parses the above input as
['2.1.4:', '1.17.2']
Fixes#782
* Fix gem get_installed_versions (cont): add changelog fragment
* Update changelogs/fragments/783-fix-gem-installed-versions.yaml as per suggestion
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
* Fix for casks with @ in the name
* Add changelog fragment
* Update changelogs/fragments/homebrew-cask-at-symbol-fix.yaml
Period required at the end of changelog entry
Co-authored-by: Andrew Klychkov <aaklychkov@mail.ru>
* Use double backticks
Co-authored-by: Andrew Klychkov <aaklychkov@mail.ru>
* Fix terraform changed status detection test (#561)
* Add changelog fragment
* Update changelogs/fragments/563-update-terraform-status-test.yaml
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
* This commit adds proxmox inventory module and proxmox_snap for snapshot management
* Fixed pylint errors
* Missed this one..
* This should fix the doc errors
* Remove proxmox_snap to allow for single module per PR
* Changes as suggested by felixfontein in #535
* Reverted back to AnsibleError as module.fail_json broke it. Need to investigate further
* Made importerror behave similar to docker_swarm and gitlab_runner
* FALSE != False
* Added myself as author
* Added a requested feature from a colleague to also sort VMs based on their running state
* Prevent VM templates from being added to the inventory
* Processed feedback
* Updated my email and included version
* Processed doc feedback
* More feedback processed
* Shortened this line of documentation, it is a duplicate and it was causing a sanity error (> 160 characters)
* Added test from PR #736 to check what needs to be changed to make it work
* Changed some tests around
* Remove some tests, first get these working
* Disabled all tests, except the one I am hacking together now
* Added mocker, still trying to figure this out
* Am I looking in the right direction?
* Processed docs feedback
* Fixed bot feedback
* Removed all other tests, started with basic ones (borrowed from cobbler)
* Removed all other tests, started with basic ones (borrowed from cobbler)
* Removed all other tests, started with basic ones (borrowed from cobbler)
* Removed init_cache test as it is implemented on a different way in the original foreman/satellite inventory (and thus also this one)
* This actually passes! Need to check if I need to add asserts as well
* Made bot happy again?
* Added some assertions
* Added note about PVE API version
* Mocked only get_json, the rest functions as-is
* Fixed sanity errors
* Fixed version bump (again...) ;-)
* Processed feedback
* Adding types to sysupgrade documentation
* Apply suggestions from code review
* Adding installurl flag. Changing wording in example.
* Use None for installurl by default
* Changing word case in description
* sysupgrade: use module structure recommended by Ansible unit test docs
* Adding unit test for sysupgrade
Signed-off-by: Andrew Klaus <andrew@aklaus.ca>
Co-authored-by: Abhijeet Kasurde <akasurde@redhat.com>
Co-authored-by: Andrew Klychkov <aaklychkov@mail.ru>
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Abhijeet Kasurde <akasurde@redhat.com>
Co-authored-by: Andrew Klychkov <aaklychkov@mail.ru>
Co-authored-by: Felix Fontein <felix@fontein.de>
* restart from last state
* test (sanity) doc fragment placeholder
* test (sanity) remove doc fragment placeholder
* remove internal params from DOCUMENTATION
* update ignore-2.10.txt
* doc: add changelog fragment
* shorten changelog fragment
* Revert "shorten changelog fragment"
This reverts commit f9aea0d1eaefda139fd5b79bd0eb127c09a433fb.
* test with posix/group1
* test with posix/group3
* test with posix/group5
* test with posix/group4
* test with posix/group3
* New modules/action plugins automatically get a changelog entry
* fix: styles
* Revert "remove internal params from DOCUMENTATION"
This reverts commit 7d5fcf4b17e4cd5b0afc08fd1bd3fcef5fcaee26.
* drop neutral/informative/stateless behaviour
* update tasks after changes in module
* use FQCN in EXAMPLES
* add tests to validate error handling about required params
* doc: remove outdated sentence
* do not document internal parameters
* display timeout value in failure message
* remove inapropriate comment
* merge results and clean them up only once
* conditionally remove tmp path
* at least one iteration is required
* remove deprecated code
* move variables declaration to conditional block
* dissociate async and connection timeout
* improve warnings (conditions + values)
* remove ANSIBLE_METADATA (no more needed); fix typo
* update DOCUMENTATION
* Drop field 'version_added' (no more needed).
* Add a note about check_mode support.
* catch early errors before resetting connection and processing the loop
* fix typo
* change posix group (due to xtables locks); add 'version_added' in doc
* update deprecation (replace Ansible 2.12 by community.general 2.0.0)
* bump version_added to 1.0.0
* update ignore-2.11.txt
* ignore errors for 2.9 as for 2.10 & 2.11
* move action plugin to system/ and replace it by a symlink
* remove action-plugin-docs override in tests/sanity/ignore*.txt
* update action plugin docstrings
* bump version_added to 1.1.0
* use lowercase booleans
* extend usage of namespaces to ansible builtin modules
* postgresql_set: allow to pass an empty string as a value
* add check_mode to CI for the case
* add changelog fragment
* add pause
* fix
* fix ci
* fix
* fix
* add suggested
* New module: shutdown
* Add symlink to plugin
* Fix
Signed-off-by: Amin Vakil <info@aminvakil.com>
* Fix
* Fix
* Add seealso
* Fix seealso
* Add future-import, metaclass boilerplate
* Change pre_shutdown_delay to delay
* Cleanup before executing shutdown
* Fix
* Remove unused connect_timeout paramater
* Improve documentation
* Remove deprecated function and calling it
* Remove double calling delay function
* Remove unneeded call in check delay function
* Make check mode more realistic
* Remove extra blank line
* Remove unnecessary imports and fix copyright year
* Add shutdown_command and integration test
* Fix integration test
* Don't fail on local AND enabled check_mode
* Add copyright
* Skip ubuntu1804 as systemd-sysv is not installed on container
* Ignore ubuntu 18 on task
* Readd integration tests
* Do not run integration test on ubuntu 18
* Improve integration test and add delay, msg testing
* Fix ubuntu 18 integration test
* Remove unnecessary condition
* Update xfconf.py
- ensure correct behaviour, even in desktop environments which don't use English as the default language
- add double as content type
* set environ_update for entire module
* set envvar LANGUAGE instead of LANG because of priority order in evaluating them
* initial commit
* add contrib info
* remove nonfunctional \n in error messate
* add documentation and optional quiet and raw arguments
* add changelog fragment
* add aliases
* add initial test
* change folder name
* add pip install jc for tests
* simplify changelog since tests were failing
* add newline to end of file
* fix trailing whitespace in comment causing test failure
* use pip3 since tests are failing using python2.6 pip
* skip python 2.7 tests since jc only supports python 3
* use pip instead of pip3
* add python version requirement
* Type: Wrong package names
In Red Hat systems, python packages are preceeded by `python3-`
* Use Python 2 packages on CentOS 7 and Fedora <= 28.
Co-authored-by: Frank Brütting <fbruetting@users.noreply.github.com>
* Enable/disable health and agent checks
Health and agent checks can cause a disabled service to re-enable
itself. This adds "health" and "agent" options that will also
enable or disable those checks, matching if the service is to be
enabled/disabled.
* Update plugins/modules/net_tools/haproxy.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update plugins/modules/net_tools/haproxy.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update plugins/modules/net_tools/haproxy.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update plugins/modules/net_tools/haproxy.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Changes to documentation and changelog.
Changes for the haproxy documentation to resolve issues with
the CI/CD, and adding a changelog fragment.
* Update changelogs/fragments/689-haproxy_agent_and_health.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update changelogs/fragments/689-haproxy_agent_and_health.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update plugins/modules/net_tools/haproxy.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Add an example of health/agent disable.
* Update plugins/modules/net_tools/haproxy.py
Co-authored-by: Andrew Klychkov <aaklychkov@mail.ru>
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Andrew Klychkov <aaklychkov@mail.ru>
Sometimes Jira returns dicts as "errors" instead of simple strings.
For example, when a user specifies a field that cannot be set, Jira
returns a dict with the field name as a key and the error message as the
value.
In the rare case that we have both a "errorMessages" list and an
"errors" dict, when we combine those values later with join(), Python
raises a TypeError.
Transform each individual error message into a string, and then join()
the list of strings.
