Since the task and connection both have the same 'retries' keyword, the task default
would override the connection value.
Do not pass 'retries' from the task to the connection options.
* Set ssh_connection retries default value back to 0
It was 0 before the move to config and was changed to 3 by accident.
(cherry picked from commit a8de35e131)
Co-authored-by: Sam Doran <sdoran@redhat.com>
* Use the module redirect_list when getting defaults for action plugins (#73864)
* Fix module-specific defaults in the gather_facts, package, and service action plugins.
* Handle ansible.legacy actions better in get_action_args_with_defaults
* Add tests for each action plugin
* Changelog
Fixes#72918
(cherry picked from commit 5640093f1c)
* Fix tests for < 3.8
Change:
- Newer Solaris drops setfacl. Add a fallback for its chmod ACL syntax.
Test Plan:
- New units
Tickets:
- Fixes#74282
Signed-off-by: Rick Elrod <rick@elrod.me>
Change:
- Regression introduced in #70785
- When macOS chmod ACL syntax is used, Solaris-derived chmods return
with a status of 5. This is also used for our sshpass handling,
because sshpass will return 5 on auth failure. This means on Solaris,
we incorrectly assume auth failure when we reach this branch of logic
and try to run chmod with macOS syntax.
- We now wrap this specific use of chmod in an exception handler that
looks for AnsibleAuthenticationFailure and skips over it. This adds
another authentication attempt (something we normally avoid to prevent
account lockout), but seems better than the regression of not allowing
other fallbacks to be used.
- Without this patch, if setfacl fails on Solaris (and sshpass is used),
we do not try common_remote_group or world-readable tmpdir fallbacks.
Test Plan:
- New unit
Signed-off-by: Rick Elrod <rick@elrod.me>
On some systems, curses.tigetstr() returns None, which does not work as a control character.
* Add unit tests
* Sort imports
* Skip on older Python
This is an action plugin and only runs on the controller, so no need to test of Python 2. Making
the import hackery work on Python 2 would required some more work which I am not sure is
worth it since we are moving away from Python 2 support on the controller.
* Make the tests work on Python 2 and 3
(cherry picked from commit 55b401a3e7)
Co-authored-by: Sam Doran <sdoran@redhat.com>
* finish migrating ssh plugin to config system
fixes#72739fixes#57220
* fix connection detection in reset
* correct options for connection meta reset
Co-authored-by: David Shrewsbury <Shrews@users.noreply.github.com>
Use correct ssh executable and options in all cases on connection plugin
* Also nicer naming/comments
Co-authored-by: Sviatoslav Sydorenko <wk.cvs.github@sydorenko.org.ua>
Change:
- Use `chmod +a` in the fallback chain to allow MacOS to use ACLs to
allow an unprivileged user to become an unprivileged user.
Test Plan:
- CI, new tests
Tickets:
- Fixes#70648
Signed-off-by: Rick Elrod <rick@elrod.me>
Change:
- Previously CachePluginAdjudicator#flush only removed entries from the
cache backend that it knew about by using them earlier. Now it calls
the underlying plugin's flush() method.
Test Plan:
- New unit tests
Tickets:
- Fixes#68770
Signed-off-by: Rick Elrod <rick@elrod.me>
Change:
- Refactoring to make it harder to get wrong and easier to read.
- Generalize become_unprivileged tests and fix some that never worked
but also never failed.
Test Plan:
- CI, new units/integration tests
Signed-off-by: Rick Elrod <rick@elrod.me>
- ensure we preserve the typeerror part of the exception so loop defereed error handling
can postpone those caused by undefined variables until the when check is done.
- fix tests to comply with the 'new normal'
- human_to_bytes and others can issue TypeError not only on 'non string'
but also bad string that is not convertable.
Co-authored-by: Sloane Hertel <shertel@redhat.com>
Co-authored-by: Sloane Hertel <shertel@redhat.com>
Change:
- Remove _get_item() alias as it has been deprecated
- Update tests
- Remove relevant sanity curtailment
- Add changelog
Test Plan:
CI, grep
Signed-off-by: Rick Elrod <rick@elrod.me>
* Only allow groups which were hardcoded in module_defaults.yml
only load action groups from the collection if module_defaults contains a potential group for the action
* Fix tests using modules that override those whitelisted in lib/ansible/config/module_defaults.yml
Third party modules should not be using group/ - use the action name instead
* add externalized module_defaults tests
add the missing group and collections
ci_complete
Co-authored-by: Matt Davis <mrd@redhat.com>
* changelog
ci_complete
* Fix import in tests
ci_complete
* Update with requested changes
ci_complete
* don't traceback since we don't validate the contents of module_defaults
ci_complete
Co-authored-by: Matt Davis <mrd@redhat.com>
* various deprecation, display, warning, error fixes
* Update lib/ansible/utils/display.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update lib/ansible/utils/display.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update lib/ansible/utils/display.py
Co-authored-by: Felix Fontein <felix@fontein.de>
* cleanup, test fixes
* add collection name to deprecated() calls
* clean up redirect entries from uncommitted tests
* fix dep warning/error header text to match previous
Co-authored-by: Felix Fontein <felix@fontein.de>
* fix delegated interpeter
* allow returning fact if it is 'the right host'
* added note for future fix/efficiency
as it stands we rerun discovery for the delegated host
unless its saving facts to itself
* fixed test lacking delegate_to mock
* Fix filedescriptor out of range in select() when running commands
* Simplify the run_command() code
Now that we're using selectors in run_command(), we can simplify some of
the code.
* Use fileobj.read() instead of os.read()
* No longer use get_buffer_size() as we can just slurp all of the data
instead.
Also use a simpler conditional check of whether the selector map is
empty
Co-authored-by: Toshio Kuratomi <a.badger@gmail.com>
* Split test_play_context_make_become_cmd into files
For NWO migration. Split the become module assertions into distinct test
files and functions. For now, this is done naively - there is probably
room to abstract these tests out and remove some of the duplication
later on.
Signed-off-by: Rick Elrod <rick@elrod.me>
* use default_exe variable instead of hardcoding /bin/bash
Signed-off-by: Rick Elrod <rick@elrod.me>
* Move become plugin tests to their proper directory and rename them accordingly
Signed-off-by: Rick Elrod <rick@elrod.me>
* Fix up fixtures and imports.
* Remove stray file.
Co-authored-by: Matt Clay <matt@mystile.com>
* Split out cache plugin unit tests.
* Rename unit tests to match code under test.
* Relocate unit test code to match code under test.
* Another rename.
* Update sanity ignores.
* Fix UNC path joining in the powershell shell plugin, add test
* Remove testy bits and a redundant line
* Fix style nits
* Update to use os.ntpath
* Add changelog for #66604
As per:
https://github.com/pytest-dev/pytest-mock#note-about-usage-as-context-manager
pytest-mock is not meant to be used within a `with` context or as a
decorator. Instead, pytest-mock will automatically unpatch the mocked
methods when each test is complete.
In newer pytest-mock, this use actually throws an exception and causes
the tests to fail.
This hasn't been hit in Ansible's CI yet, because the docker image
that the tests run in uses an older version of pytest-mock. However,
there is no constraint on the upper bound of pytest-mock in
test/lib/ansible_test/_data/requirements/constraints.txt which means
that when running the tests locally, outside of that docker image, the
tests never pass.
This patch removes the `with` context in each such case.
Signed-off-by: Rick Elrod <rick@elrod.me>
Sam Doran