Log generic warning on non-systemd systems.

[aciba90]

If the SSH status is checked in a non-systemd system, do not
distinguish between service statuses and log a generic warning
as we cannot expect status exit codes fulfilling any convention.

Proposed Commit Message

Log generic warning on non-systemd systems.

If the SSH status is checked in a non-systemd system, do not
distinguish between service statuses and log a generic warning
as we cannot expect status exit codes fulfilling any convention.

Additional Context

#1422 (comment)

Checklist:

• My code follows the process laid out in the documentation
• I have updated or added any unit tests accordingly
• I have updated or added any documentation accordingly

[blackboxsw]

The PR in 1449, which only just landed while you were working on this PR, refactored some unittests to use pytest instead of subclassing from CiTestCase. #1449 provided a parameterized TestHandleSSHPwauth test case test_no_restart_when_service_is_not_running. Would these new tests make more sense if parameterized in either TestHandleSSHPwauth.test_no_restart_when_service_is_not_running or a new parameterized testcase instead of separate unittests?

Two other minor points:

• generally in unittests, I think we want to start avoiding calls to self.tmp_cloud and prefer tests.unittests.util.get_cloud instead as we try to retired tmp_cloud (and CiTestCase) where possible.
• I don't think we need to set cloud.distro.init_cmd = ["service"] on ubuntu distro because default is init_cmd = ["service"] and this only gets reset to "systemctl" if uses_systemd == True, which you've mocked in this test.

[aciba90]

Sorry for breaking the integration tests and thanks for pointing out the updates.

I have parametrized the tests.

[aciba90]

@TheRealFalcon this PR is ready now. Thanks.

[TheRealFalcon]

LGTM, thanks for your flexibility on this!

Since @blackboxsw requested changes, I'll wait until he gives his approval.

[blackboxsw]

LGTM

[dermotbradley]

@blackboxsw @TheRealFalcon. This change now means I get the following on the console for Alpine Linux systems (openrc/busybox init-based) where previously I did not:

2022-05-15 19:06:05,300 - cc_set_passwords.py[WARNING}: Ignoring config 'ssh_pwauth: False'. SSH service 'sshd' is not available. Error: unexpected error while running command.
Command: ['rc-service', 'sshd', 'status']
Exit code: 3
Reason: -
Stdout:  * status: stopped
Stderr: .

This is because the sshd service is started later in the boot sequence.

Also note that I have "ssh_pwauth: False" defined in /etc/cloud/cloud.cfg

[aciba90]

2022-05-15 19:06:05,300 - cc_set_passwords.py[WARNING}: Ignoring config 'ssh_pwauth: False'. SSH service 'sshd' is not available. Error: unexpected error while running command.
Command: ['rc-service', 'sshd', 'status']
Exit code: 3
Reason: -
Stdout: * status: stopped
Stderr: .

@dermotbradley I think we could fix this by:

1. if service != systemd and status != 0 then applying the config
2. reword the warning to be more generic, as it is not very accurate. We could say something like SSH service 'sshd' will not be restarted because it is not running or not available.

Let me prepare a PR.