Lxd instance not found fix false positives (#7521)
* lxd: Add lxc command execution debug statement.
* lxd: avoid false positives in "instance not found" detection
Due to changes over time in the error message which lxd printed when an
instance wasn't found, the detection logic in the lxd connection plugin
matched any "not found" string. Unfortunately this also false triggered
on other errors e.g. "/usr/bin/python3: not found" from the payload,
giving a confusing error message "UNREACHABLE! ... instance not found"
to the ansible user.
* Update changelog fragment.
---------
Co-authored-by: Felix Fontein <felix@fontein.de>
(cherry picked from commit b4a2e9da50)
Co-authored-by: Tim Small <tim@seoss.co.uk>
* Fix more typos in plugins/.
* Fix typos in tests/unit/.
* Fix typos in tests/integration/.
* Fix more typos.
Co-authored-by: Sebastian Gumprich <rndmh3ro@users.noreply.github.com>
---------
Co-authored-by: Sebastian Gumprich <rndmh3ro@users.noreply.github.com>
* plugins/connection/lxd: convert FQDN to instance name
This allows to use FQDNs in the inventory and have the
connection driver do the translation when talking to
LXD that uses hostnames (no ".") for instance names.
Those are either globally unique or unique per network/
project in LXD.
```
all:
# Groups and hosts
children:
lxd_dmz:
vars:
ansible_lxd_project: dmz
hosts:
www01.dmz.example.com:
www02.dmz.example.com:
```
```
$ lxc list --project dmz
+---------+---------+----------------+------+-----------+-----------+----------+
| NAME | STATE | IPV4 | IPV6 | TYPE | SNAPSHOTS | LOCATION |
+-------+---------+------------------+------+-----------+-----------+----------+
| www01 | RUNNING | 192.0.2.1 (eth0) | | CONTAINER | 0 | t1 |
+-------+---------+------------------+------+-----------+-----------+----------+
| www02 | RUNNING | 192.0.2.2 (eth0) | | CONTAINER | 0 | t3 |
+-------+---------+------------------+------+-----------+-----------+----------+
```
Signed-off-by: Simon Deziel <simon.deziel@canonical.com>
* plugins/connection/lxd: VMs/containers are called instances
Update error string parsing to support the new format:
$ lxc stop c1 -- true
$ lxc exec c1 -- true
Error: Instance is not running
$ lxc exec does-not-exist -- true
Error: Instance not found
Signed-off-by: Simon Deziel <simon.deziel@canonical.com>
* plugins/connection/lxd: add changelog fragment
Signed-off-by: Simon Deziel <simon.deziel@canonical.com>
---------
Signed-off-by: Simon Deziel <simon.deziel@canonical.com>
* Stopped container error test
* Handle remote_addr change
Detect if the remote_addr option changed, and properly "reconnect" aka
update the internal state of the plugin instance.
* Add changelog fragment
* Apply suggestions from code review
Co-authored-by: Felix Fontein <felix@fontein.de>
---------
Co-authored-by: Felix Fontein <felix@fontein.de>
* Fixture for liblxc
Add a fixture to allow testing the lxc connection plugin both with and
without liblxc being present.
Also change the test from unittest to pytest.
* Update liblxc error message
The error is not specific to python2, so remove the version. Also add
a test for it.
* Migrate to options
Because the lxc plugin was only using PlayContext properties, using host
vars like `ansible_lxc_host` didn't work. This is fixed by instead using
the `get_option` method inherited from `AnsiblePlugin`.
The options are not yet available in the `__init__` function, so the
determination of the container name is moved to the `_connect` method,
which is the first time it is actually needed.
The default for the `remote_addr` option is removed, because the string
`inventory_hostname` is not very useful. At all. This seams to have been
spread with copy&paste and a bit of cargo culting. The variable priority
already takes care of setting the value.
* Add changelog fragment
* Fix for Py2.7
`TypeError: super() takes at least 1 argument (0 given)`
* Add plugin type to changelog fragment.
* Restore untemplated default
This partially reverts commit 429d8c8cfb.
---------
Co-authored-by: Felix Fontein <felix@fontein.de>
* chroot: added an example
Added a simple example of chroot connection plugin
Fixes: #6365
Signed-off-by: Abhijeet Kasurde <akasurde@redhat.com>
* Apply suggestions from code review
Co-authored-by: Felix Fontein <felix@fontein.de>
---------
Signed-off-by: Abhijeet Kasurde <akasurde@redhat.com>
Co-authored-by: Felix Fontein <felix@fontein.de>
This fixes the following warning on FreeBSD:
[WARNING]: The "jail" connection plugin has an improperly configured
remote target value, forcing "inventory_hostname" templated value
instead of the string
* Move licenses to LICENSES/, run add-license.py, add LICENSES/MIT.txt.
* Replace 'Copyright:' with 'Copyright'
sed -i 's|Copyright:\(.*\)|Copyright\1|' $(rg -l 'Copyright:')
Co-authored-by: Maxwell G <gotmax@e.email>
* fixes lxd connection plugin issue #4886
remote_addr value was set to literal string 'inventory_hostname' instead
of the value for inventory_hostname variable. solution found in PR
ansible/ansible#77894
* changelog fragment - bugfix - lxd connection plugin
* correct changelog fragment
* Update changelogs/fragments/4886-fix-lxd-inventory-hostname.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
* replace _host instance variable with calls to get 'remote_addr' option
suggested by felixfontein
Co-authored-by: Felix Fontein <felix@fontein.de>
* Update lxd connection to use documented vars
* Add a changelog fragment
* Add clarification to changelog description
* Shorten changelog fragment description
* added utf-8 markers to all .py files in plugins/{action,cache,callback}
* added utf-8 markers to all .py files in plugins/connection
* added utf-8 markers to all .py files in plugins/doc_fragments
* minor refactors
* minor refactors in plugins/connection/saltstack.py
* minor refactors in plugins/connection/qubes.py
* minor refactor in plugins/connection/lxc.py
* minor refactors in plugins/connection/chroot.py
* minor refactors in plugins/connection/funcd.py
* minor refactors in plugins/connection/iocage.py
* minor refactors in plugins/connection/jail.py
* added changelog fragment
* funcd connection plugin is now usable/loadable
* Update changelogs/fragments/allow_funcd_to_load.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
* fixed validation-modules for plugins/callback/hipchat.py
* fixed validation-modules for plugins/connection/lxc.py
* fixed validation-modules for plugins/modules/cloud/lxc/lxc_container.py
* fixed validation-modules for plugins/modules/monitoring/statusio_maintenance.py
* fixed validation-modules for plugins/modules/system/alternatives.py
* fixed validation-modules for plugins/modules/system/beadm.py
* fixed validation-modules for plugins/modules/system/cronvar.py
* fixed validation-modules for plugins/modules/system/dconf.py
* fixed validation-modules for plugins/modules/system/interfaces_file.py
* fixed validation-modules for plugins/modules/system/java_cert.py
* fixed validation-modules for plugins/modules/system/lvg.py
* fixed validation-modules for plugins/modules/system/lvol.py
* fixed validation-modules for plugins/modules/system/parted.py
* fixed validation-modules for plugins/modules/system/timezone.py
* fixed validation-modules for plugins/modules/web_infrastructure/rundeck_acl_policy.py
* Tidy up all pylint:blacklisted-name sanity checks ignore lines
* Missed one in statusio_maintenace.py
* fixed validation-modules for plugins/modules/system/filesystem.py
* Missed one in gconftool2.py
* Missed one in alternatives.py
* Using dummies now
* fixed indentation
* Made all the changes about replacing _ with dummy
* Rollback bug fixed
* Rollback bug fixed, part II
* added changelog fragment
* Improved changelog frag message per PR
* saltstack: fix put_file to preserve checksum
Use hashutil.base64_decodefile to ensure that the file checksum
is preserved, since file.write only supports text files.
Signed-off-by: Zac Medico <zmedico@gmail.com>
* Update changelogs/fragments/1472-saltstack-fix-put_file-to-preserve-checksum.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
* Remove and redirect oc module.
* Add changelog fragment for move of oc connection plugin to OKD collection.
* Update BOTMETA as well.
* Newer changelog fragment format.
* Update changelogs/fragments/oc-migration-removal.yml
Co-authored-by: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
* callback_type -> type.
* Mark authors as unknown.
* Add author field forgotten in #627.
* Fix author entries.
* Add author field forgotten in #127.
* Fix some types.
This was using k8S_AUTH_CONTEXT with a lowercase k. That doesn't match
with any of the other K8s stuff so users would probably be confused why
their envvar wasn't working.
* Update docker.py
#67832 fixes the need to run powershell modules on windows containers via docker connection. However, while running win_copy on windows containers, destination path is prefixed, which doesn't work in the case of windows. The changes in this PR take care of bypassing that while running the ansible role on windows containers.
* Update plugins/connection/docker.py
Co-Authored-By: Felix Fontein <felix@fontein.de>
* Create 80-update_docker_connection_plugin.yml
Add changelog fragment
* Update changelogs/fragments/80-update_docker_connection_plugin.yml
Co-Authored-By: Felix Fontein <felix@fontein.de>
Co-authored-by: Felix Fontein <felix@fontein.de>
