This PR includes:
- An important fix to charset encoding of from address
- Documentation and examples cleanup
- PEP8 fixes
- Warning on insecure access
- Strict parameter typing
- More modern interface (using lists rather than comma, space or pipe-delimited strings)
- Warn on failure to send mail to some recipients
```
[WARNING]: Failed to send mail to 'foobar': 550 5.1.1 <foobar>:
Recipient address rejected: User unknown in local recipient table
```
- Warn on failure to parse some headers
```
[WARNING]: Skipping header 'Foobar', unable to parse
```
- Return failed recipients as return value
- Changed default encoding to utf-8
* made callbacks backwards compatible
This fixes#30597 for those that were not inheriting from base.
Added deprecation notice so those callbacks get updated.
Callback must either inherit from base (directly or indirectly),
which already implements this or implement set_options themselves.
* added note about porting guide
This is to catch vault secrets from config and
cli. Previously vault_password_file in config was
missed since it was added by setup_vault_secrets,
so check after setup_vault_secrets.
* Restore correct coloring to selective callback
This fixes the bug raised in #30506
* Fix format issues for Python 2.6 & indent
Removed the zero length fields to support format under Python 2.6
Fixed E128 continuation line under-indented for visual indent issue
* Add Routing Engine Facts
- Map routing engine output information to routing_engines facts dict.
- Add fact 'has_2RE', which is a quick way to determine how many REs
the chassis has.
* Fix a typo
* Fix more typos
* Add slot number to routing_engine dict
* Add facts about the installed chassis modules
* Fix typo
* Fixed another typo
* Fix Path
* Change path again.
* More Typos
* Add some deubgging
* Add additional information for hardware components.
- Return information about the Routing Engines.
- Return a fact to easily determine if the device
has two routing engines.
- Return information about the hardware modules.
* Addressed pep8 stardard failures.
* Add unit test fixtures.
* Rename fixture.
* Fix unit test failures.
- Rename the fixture file to what the unit test expects.
- Strip out junos namespace attributes.
Rename file to match what the unit test expects.
* Scrubbed the routing engine serial numbers.
* Add unit test facts for new tests.
- Add unit test for ansible_net_routing_engines fact
- Add unit test for ansible_net_modules fact
- Add unit test for ansible_net_has_2RE
* Fixed spacing.
* win_scheduled_task: rewrite for additionality functionality and bug fixes
* fixes for docs and os version differences
* started with the testing
* doc fix
* added more tests
* added principals tests
* finished tests for win_scheduled_task rewrite
* feedback from PR
* change to fail when both new and deprecated args are set
* change diff variable to match new standard and update doc sentance
* Don't ask for password confirm on 'ansible-vault edit'
This is to match the 2.3 behavior on:
ansible-vault edit encrypted_file.yml
Previously, the above command would consider that a 'new password'
scenario and prompt accordingly, ie:
$ ansible-vault edit encrypted_file.yml
New Password:
Confirm New Password:
The bug was cause by 'create_new_password' being used for
'edit' action. This also causes the previous implicit 'auto prompt'
to get triggered and prompt the user.
Fix is to make auto prompt explicit in the calling code to handle
the 'edit' case where we want to auto prompt but we do not want
to request a password confirm.
Fixes#30491
* finalize lookup documentation
* minor fixes to ansible-doc
- actually show which file caused error on when listing plugins
- removed redundant display of type and name
* smart quote fixes from toshio
Currently, MIQ only supports an alert type of 'prometheus', so rather than have the caller of manageiq_provider pass this info, just set it as the default.
When calling manageiq_user to an already existing user (but leaving out the password so that it doesn't automatically 're-create' the user), the module fails with:
fatal: [127.0.0.1]: FAILED! => {"changed": false, "failed": true, "module_stderr": "Shared connection to 127.0.0.1 closed.\r\n", "module_stdout": "Traceback (most recent call last):\r\n File \"/tmp/ansible_Fr7Nt3/ansible_module_manageiq_user.py\", line 324, in <module>\r\n main()\r\n File \"/tmp/ansible_Fr7Nt3/ansible_module_manageiq_user.py\", line 315, in main\r\n res_args = manageiq_user.edit_user(user, name, group, password, email)\r\n File \"/tmp/ansible_Fr7Nt3/ansible_module_manageiq_user.py\", line 229, in edit_user\r\n if self.compare_user(user, name, group_id, password, email):\r\n File \"/tmp/ansible_Fr7Nt3/ansible_module_manageiq_user.py\", line 189, in compare_user\r\n (group_id and user['group']['id'] != group_id)\r\nKeyError: 'group'\r\n", "msg": "MODULE FAILURE", "rc": 0}
The 'group' field turns out to be 'current_group_id' (at least with ManageIQ 4.6). Update the comparison accordingly.
* add 'update_password' param to manageiq_user
Currently with the manageiq_user module, if you call it repeatedly while passing the 'password' parameter, it will always run the task and mark it as 'changed'.
Following the pattern of the AWS IAM module, add an 'update_password' parameter that takes 'always' (default) or 'on_create'. This will let you set an initial password when creating a user, but allow the user to modify their password and not stomp over their password changes if you re-run the playbook/task that created the user.
* don't stomp password when other fields change
Handle case where user fields change, but we don't want to stomp on a potentially user-changed password. Previously, if a non-password field changed, and the password param was passed in, it would ignore the 'update_password': 'on_create' setting (ie it would update/modify the password even if the user already exists).
Add trailing ',' to list of params.
* windows: fix list type in legacy module utils
* only change the return for the list type instead of affecting it all
* additional null check when using an array
* Fix tags in ec2_instance_facts
The method boto3_tag_list_to_ansible_dict in module_utils/ec2.py changed
and does no longer check whether the returned result of boto3 uses
"key" or "Key" as the tag key identifier.
This fixes ec2_instance_facts to make this check in its own, since boto3
may return "key" instead of "Key"
* Since the indices for the tags are already formatted to lowercase
by the snaking, we can assume, that the index for the tags are already
formatted
* timezone module: fixed platform decision rule for Linux
— For better handling of environments where timedatectl is unavailable
* timezone module: allow absence of configuration files if specific commands are available
* timezone module: remove duplicated line
* timezone module: fixed docs to clarify returned diff
* timezone module: fixed “undefined variable err”
* Revert "timezone module: fixed docs to clarify returned diff"
This reverts commit 4b783227f7.
* timezone module: revert platform decision rule; just warn instead of futher command checks
* timezone module: [NosystemdTimezone] enhanced error message
* changed RunCommand result from Tuple to CommandResult for easier future extensibility
* moved Win32 Dictionary->multi-null-string environment munging into C#
As-merged, had several issues that prevented idempotent usage. Some args were defined at the wrong UI level. Dual-state args didn't match up with typical Ansible UI.
* fixed issue with default callback inheritance
- callbacks need to document same options as callbacks they inherit from to get them configured
- since default is also used by many 3rd party callbacks for inheritance, making the code 'tolerate' the missing docs
and fallback to using the direct constant to configure it's options.
* Added nopackages option and Fix#24997
Adding a new option - nopackages.
This enables the option to add the --nopackages flag while registering a new node to RHN Satellite. We are not uploading the rpm data on our nodes and since we started utilizing ansible for nodes registration, I figures it would be useful for others as well.
Also-
Fixes#24997 (verified in my lab)
* Fixed documentation
* Documentation changes:
- typo fix in "default"
- Added "version_added" and set to 2.4
* Documentation changes:
- Removed trailing whitespaces in nopackages['version_added']
* This change is unrelated for this feature pull request and shouldn't be here (and also seems wrong, see #25079).
* Changed "version_added" to 2.5 in the module docs
It could be something like '10beta4', which StrictVersion() would
reject. When Postgres 10 is released, it will be '10', which
StrictVersion() would STILL reject.
Fortunately, psycopg2 has a 'server_version' connection attribute that
is guaranteed to be an integer like 90605 for version 9.6.5, or 100000
for version 10. We can safely use this for version-specific code.
* Replace pause in integration tests with until.
Use resource prefix instead of generating a random number
Only try to delete keys if they exist
* Add alias to tests
1) import_role was never resulting in a static inclusion of the role
tasks due to a logic error.
2) no error was raised when import_role tried to use a with loop, resulting
in a strange error down the execution path.
* Consistency and document treatment of default bool values
* Document that default bool values can be any Ansible recognized bool.
choose the one that reads better in context
* For fragments used by the copy module, make bool types use type=bool and not choices
* Edit for clarity
keyUsage and extendedKeyUsage are currently statically limited via a
static dict defined in modules_utils/crypto.py. If one specify a value
that isn't in there, idempotency won't work.
Instead of having static dict, we uses keyUsage and extendedKyeUsage
values OpenSSL NID and compare those rather than comparing strings.
Fixes: https://github.com/ansible/ansible/issues/30316
The dellos action plugins should add the remote address of the switch
provider to the play context. This was fixed in issue #23589 in an
almost identical manner for the eos, ios, iosxr, and vyos action
plugins.
Fixes: #30350
Current openssl_certificate is mistakenly taking its derivating its
version number from the csr version number.
Thos two fields are completly unrelated and hence the version number of
the certificate should be able to be directly specified (via
selfsigned_version parameter).
* Updating the nsupdate module to accept a list for 'value' instead
of a string. This is to allow manipulating 1:many DNS records.
A string can still be supplied so it should be backwards compatible.
Addresses issue #25554
* Update nsupdate.py
* Update nsupdate.py
* Update nsupdate.py
* openssl_certificate: Fix parameter assertion in Python3
Parameter assertion in Python3 is broken. pyOpenSSL get_X() functions
returns b'' type string and tries to compare it with '' string, leading
to failure.
The error mentionned above has been fixed by sanitizing the inputs from
a user to the assert only backend.
Also, this error was hidden by the fact that the improper check method
was called in the generate() functions.
* Add simple integration test for openssl_certificate
* remove subject == issuer assertion
* run integration tests only on supported hosts
* change min supported version to 0.15.x
* Add test for more CSR fields
* also convert dict members to bytes
* fix version_compare
* openssl_{csr, certificate}: Fail if pyOpenSSL <= 0.15
Previous 0.13 pyOpenSSL was a C-binding, and required the parameter
passed to add_extention to be in ASN.1. This has changed with the move
to 0.14 and it is now all pythong and string based.
Previous the 0.15 release, the `get_extensions()` method didn't exist,
since the modules rely heavily on it we ensure pyOpenSSL version is at
last 0.15.0.
* check pyopenssl version in openssl_csr integration test
* ec2_group: Handle name conflict with empty vpc_id.
If several groups exist with the same name (and vpc_id is None) then
treat the group outside the vpc as preferred (same as it would for a vpc
group with vpc_id specified). Also don't run the egress rules code in
that case.
* Handle lack of `IpPermissionsEgress` attribute on EC2 classic groups
In EC2 classic groups, the `while True` loop checking for egress
permissions will continue infinitely.
* Handle incompatible combinations of EC2 Classic + VPC groups
* Fix integration tests in accounts lacking EC2 classic
This change checks against the security group created, instead of the
module parameters, for VPC ID. This means that new accounts with a
default VPC will still wait properly for the first egress rule to
populate.
* Fix conditional for storing described groups with preference for matching VPC IDs
* Revert `vpc_id is None` on conditional to allow for default VPCs
Per the new style of execution, for dynamic tasks conditionals are expected
to only affect the include task itself and should not be inherited by child
tasks. This patch brings the behavior inline with this expectation.
Fixes#27845
* Clean up nxos_snmp_contact & nxos_snmp_location
* Bring nxos_snmp_community in line
* Bring nxos_snmp_host in line
* And I would have gotten away with it too,
if it weren't for those meddling sanity tests
* Bring nxos_snmp_traps & nxos_snmp_user in line
* Appease Shippable
* nxos_file_copy bug fix
Signed-off-by: Trishna Guha <trishnaguha17@gmail.com>
* provider gets set to None in module level when transport is cli
Signed-off-by: Trishna Guha <trishnaguha17@gmail.com>
* update ec2_vpc_subnet_facts module to use boto3 and support gathering updated fact items from AWS API
add version_added to new parameter
added return docs and other requested changes
removed errant extra blank line
updates per review
* update per review: fix AWSRetry backoff implementation and fix example that was not correct
* Cleanup password error handling for E-Series auth
The E-Series auth module was using some erronous behavior on handling
the status codes when updating the system password. This patch resolves
some of these issues.
* Fix validate_certs param in E-Series auth module
The auth module was ignoring the validate_certs parameter for making
HTTPS calls to the back-end API. This patch resolves the ignored
parameter.
Pull the get_poller_result inside the if block so that if the caller has
wait_for_deployment_completion=False, it doesnt block and wait for it to
finish.
Also, since the result contains information about the deployment, provide
None values for it in the output.(Not sure if this needs to be documented)
Fixes#26014
* fixes#26623
* Test-Path (and thus `-type path` in Get-AnsibleParam) fail on a nonexistent drive letter, since it can't be mapped to a PSProvider.
* added support and basic smoke tests for
* Refactor E-Series LunMapping module to use module_utils
Refactor the NetApp E-Series module to utlize the common module_utils
and doc_fragments.
* Remove the default LUN number
By providing a default LUN number, it interferes with the ability of the
API to determine the appropriate LUN value.
* Fix ignored validate_certs parameter
The validate_certs parameter was not being provided to the underlying
requests method. This patch resolves the issue by passing the value to
all relevant HTTP requests.
Fixes#29060
Allow delegate_to if transport is either nxapi or eapi.
Persistent connection uses `cli` transport and create
a local socket on control node. Hence delegate_to is not allowed
for `cli` transport.
However as `nxapi` and `eapi` transport does not use persistent connection
it is possible to use delegate_to in this case.
* Fix unwanted deprecation message in network module argspec
Fixes#25663Fixes#24537
* segregate provider spec and top level arg spec
* add deprecation key in top level arg spec
* remove action plugin code to load provider and add
that logic at a common place in network_common.py file
* Fix CI issue
* Minor change
* Using docstrings conflicts with the standard use of docstrings
* PYTHON_OPTIMIZE=2 will omit docstrings. Using docstrings makes future
changes to the plugin and module code subject to the requirement that we
ensure it won't be run with optimization.
* Create instance-store AMI instances with 'terminate' as the shutdown behavior since it is required.
* Match on the error code instead of searching for a string in the message.
* Narrow conditional to only fix shutdown behavior if fixing it would help
* Fix pep8.
* module and vault fixes
- fix module_path cli option and usage, which fixes#29653
- move --output to be in subset of vault cli, no need for all vault enabled cli to use it
- added debug to loader to see directories added
* refactor firewalld module with object abstraction
This change creates a FirewallTransaction object that each
individual transaction type is a sub-class of as they all follow the
same pattern to enable or disable something in the firewall.
Also, there's a few bugfixes here:
- Fix the "source" type to handle permanent operations
- Remove ambiguity of required parameters for only specific use
cases that can lead to transactions effectively being a no-op.
Instead, pick sane defaults and document them.
- Change how imports are done so globals are no longer needed
This is based on the original feedback by Toshio from the last
refactor attempt:
https://github.com/ansible/ansible-modules-extras/pull/3383
Signed-off-by: Adam Miller <maxamillion@fedoraproject.org>
* fix line too long for pep8 for shippable tests
Signed-off-by: Adam Miller <maxamillion@fedoraproject.org>
* remove firewalld from pep8/legacy-files
Signed-off-by: Adam Miller <maxamillion@fedoraproject.org>
* only complain about ini deprecation if value is set
* set plugin config for stdout and other types
* updated plugin docs, moved several plugins to new config
* finished ssh docs
* fixed some issues seen in plugins while modifying docs
* placeholder for 'required'
* callbacks must use _plugin_options as _options already in use
from __future__ unicode_literals leads to developer confusion as
developers no longer can tell whether a bare literal string is a byte
string or a unicode string. Explicit marking as u"" or b"" is the way
to solve the same problem in the Ansbile codebase.
* Change ansible-doc usage to show -a is for internal use
ansible-doc -a is for testing that documentation is sane. It should not
be used by normal users in production. The main reason for this is that
it is designed to fail if there are any undocumented modules or plugins.
This is good for testing that all plugins we ship are documented. It is
not good for end users who may have undocumented third-party plugins.
The config variables defined with eval, like INVENTORY_IGNORE_EXTS,
are not stored properly once the eval is processed.
This causes references to the constant to still have the eval in the
value.
since we want to make namespaced facts drop ansible_ prefix but don't have the
time before release to perfect this feature, we are going to postpone it for now
until we have the resources to fix this issue. That way we won't have people relying
on the 'incorrect' names for a release.
* Adding tls settings for Zabbix host
* Using the correct Ansible version
* Removing wildcard import
* Added module_utils package
* Set version_added for visible_name back to 2.3
* Added description for >= Zabbix 3.0; Added parameters for function
* Setting version_added --> 2.5; Removed trailing whitespace
* This commit includes a unit test to exercise the _is_role
function and make sure it doesn't break in any Python version.
* Import os.path and other minor fixups
* Remove 'required: false' statements from the argument docs
* Remove 'required=False' parameters from argument spec
* Remove 'default: null' statements from the argument docs
This adds the --syn option to filter SYN packets. Can be negated.
I added a generic append_match_flag function which can be used to add
match flags without parameters. It also allows negating the flag
if the added param allows this.
Not sure if I took the best approach here so all feedback welcome :)
* cloud: azure: fix typo introduced in commit 16d23e9
The commit "Add reference to VNET resource group (#26052)"
removed an used variable.
* network: aos: error hint never shown
- better variable precedence management
- universal plugin option handling
- also updated comments for future directions
- leverage fragments for plugins
- removed fact namespacing
- added 'firendly name' field
- updated missing descriptions
- removed some unused yaml entries, updated others to reflect possible future
- documented more plugins
- allow reading docs using alias
- short licenses
- corrected args for 'all plugins'
- fixed -a option for ansible-doc
- updated vars plugins to allow docs
- fixed 'gathering'
- only set options IF connection
- added path list and renamed pathspec mostly the diff is , vs : as separator
- readded removed config entries that were deprecated but had no message ... and deprecated again
- now deprecated entries give warning when set
The ec2_vpc_route_table module notifies about a change on the route table when the instance Id of the NAT instance has changed, but in fact, nothing changes. The module call the create_route function the AWS SDK to add a new route with the same cidr. The AWS SDK should return an error instead of nothing.
Call replace_route function instead of create_route when a route table with the same cidr but with different target destination is present.
