Before "5ca23e3bf (Changed to use first_kube_control_plane to parse
kubeadm_certificate_key (#11875), 2025-01-14)", kubespray would have
problem adding new control planes when the order of the nodes in kubectl
output and the ansible inventory were not the same.
But the underlying problem is that the operation is fundamentally
something that should be done only once, and recorded for all host in
play.
Since `register` and `sef_fact` when used with `run_once` set the
variable for all the hosts, use it. Also allows to use the variable
directly instead of relying on hostvars to make the task more readable.
Most variables should have a default instead of relying on the default
filter.
(Note that the variable is misnomed, this should be certs and not keys,
but it's not worth breaking compat).
This should make 'no space left on device' problems easier to handle
Use /tmp/releases as local_release_dir CI created machine, while keeping
the same folder on the runner (needed for gitlab-ci runner pods)
* CI: Try a full ssh connection on hosts instead of only checking the port
If we only try the port, we can try to connect in the playbook which is
executed next even though the managed node has not yet completed it's
boot-up sequence ("System is booting up. Unprivileged users are not
permitted to log in yet. Please come back later. For technical details,
see pam_nologin(8).")
This does not account for python-less hosts, but we don't use those in
CI anyway (for now, at least).
* CI: Remove connection method override when creating VMs
This prevented wait_for_connection to work correctly by hijacking the
connection to localhost, thus bypassing the connection check.
* Remove etcd member by peerURLs
The way to obtain the IP of a particular member is convoluted and depend
on multiple variables. The match is also textual and it's not clear
against what we're matching
It's also broken for etcd member which are not also Kubernetes nodes,
because the "Lookup node IP in kubernetes" task will fail and abort the
play.
Instead, match against 'peerURLs', which does not need new variable, and
use json output.
* Add testcase for etcd removal on external etcd
* do not merge
* fixup! Remove etcd member by peerURLs
* fixup! Remove etcd member by peerURLs
The 'old' playbook and the collection use '-' and '_' as separator,
which breaks the logic in scripts/testcases_run.sh.
Add aliases using the old schemes to make the test work and avoid
breaking anything.
Both '-' and '_' variants will be deleted once we switch to supporting
collection only.
This allows to use kubespray_defaults (once) instead of redefining
defaults in the tests.
Test test files becomes imported tasks rather thand standalone
playbooks.
Not really a reason not to, and this actually breaks daily-ci because
some jobs depends on this one so the whole pipeline is invalid if it's
not created.
This uses the same logic than the other versions, with simplications for
crictl and crio whose versionning scheme is tied to upstream kubernetes.
Also move some version variables in vars/ rather than defaults/, because
they are not used elsewhere and don't really make sense as modifiable by
the user.
* Remove tag master
Following it's deprecation in 4b324cb0f (Rename master to control plane
- non-breaking changes only (#11394), 2024-09-06)
* Add fail fast path when using removed tags
- Used for the master tag, but this could be used for other things in
the future
The checksums are not a defaults and are not meant to be changed from
the inventories.
Furthermore, role defaults have a lower priority that hosts facts, which
technically means a rogue hosts could hijack the hashes for its
variables.
This is now handled directly at the failfast-ci level (== integration
Github <-> Gitlab).
The whole pipeline will not be triggered unless:
- The author is a maintainer
- The PR has the /ok-to-test label
The switch to not use system packages for containerd packages happened
multiples releases ago ; there should not be any up-to-date installation
of kubespray needing that cleanup.
Remove those steps and variables only used by them.
* Delete unused scripts
- gen_tags.sh: not the right file, produce garbage even if path is fixed
- premoderator.sh: not used since ef6d24a49 (CI require a 'lgtm' or
'ok-to-test' labels to pass (#11251), 2024-05-31)
- gitlab-branch-cleanup: unused AFAICT
* CI: inline molecule logs
Single use site -> less indirection makes it easier to read.
With the current github-workflow setup, workflows are triggered on every
forked repository (which is quite wasteful).
Add a condition to only run on the main repository.
kubespray-defaults currently does two things:
- records a number of default variable values (in particular values used
in several places)
- gather and compose some complex network facts (in particular,
`fallback_ip` and `no_proxy`
There is no actual reason to couple those two things, and it makes using
defaults more difficult (because computing the network facts is somewhat
expensive, we don't want to do it willy-nilly)
Split the two and adjust import paths as needed.
The recommended usage of kubespray is to use the default versions.
So putting them in inventory/sample is not really very helpful, and
causes:
- churn (keeping the inventory/sample up to date)
- support issues (mismatch between defaults and sample inventory)
Remove all concrete versions from the inventory sample.
bootstrap-os does not do anything in sudoers since e2ad6aad5 (bootstrap:
rework role (#4045), 2019-02-11).
So SSH pipelining working is effectively a pre-requisite anyway.
Max Gautier
