For historic reasons kubeadm kept track of a skew of 2+
etcd version in a map in the constants.go file.
This is really not required because kubeadm supports only
2 version of etcd mapped to two Kubernetes versions,
which are essentially the control plane versions.
Refactor the constants.go map to only include 2 versions.
Make sure that's reflected in a unit test.
Instead of pinning the versions as literal numbers,
start using the version we get on build time.
Adapt various unit tests and functions to allow this change
to work, since during unit tests we need actual values
and the build versions are not populated. This is achieved
by requiring the functions to accept a map[uint8]string
used for testing.
By using k8s.io/system-validators v1.12.1 a KubeletVersion
can be passed to the CgroupsValidator. The library
can then decide if to throw an error or a warning.
An error is thrown if the version is >= 1.35.
During upgrade pass a KubeletVersion that is with MINOR+1
to account for the target upgrade version. That is just
an assumption that the user will upgrade the kubelet too,
but it is better to throw an error for this
common use case instead of just a warning that is seen
after the upgrade command has finished.
When retrieving the cluster-info CM, ensure the cluster pointed
out by the current context in the kubeconfig is validated.
Add unit test for the above.
Make the function GetClusterFromKubeConfig() to return various
errors. Handle the errors on call sites. Add unit tests
for the update.
The above changes prevent panics when the users has manually
edited and malformed the kubeconfig in the cluster-info CM.
The newControlPlane flag has been historically problematic, since
it implies that the function FetchInitConfigurationFromCluster
cannot handle the cases where a node is worker node but
we still want to fetch its NodeRegistrationOptions conditionally,
in cases such as "upgrade node" for workers.
To fix this issue, replace the flag newControlPlaneNode with
two new flags getNodeRegistration and getAPIEndpoint.
If getNodeRegistration is true, we fetch the NRO, and if
getAPIEndpoint is true, we fetch the API endpoint for
that node.
Additionally, rename skipComponentConfigs to getComponentConfigs
for consistency and flip its value accordingly everywhere.
When waiting for the kube-apiserver to report 'ok'
in the 'init' and 'join' phase 'wait-control-plane', a client
constructed from the 'admin.conf' is used. In the case of the
kube-apiserver, the discovery client is used so that
anonymous-auth works. But if 'admin.conf' is used as is,
it would point to the CPE and not the LAE.
Implement a new method WaitControlPlaneClient() for both
init.go and join.go that patches the 'Server' field in the
loaded v1.Config to point to the LAE, before constructing
a client set and using it in the kube-apiserver waiter.
Replace context.TODO() and poll timeout context propagation with direct
context.Background() calls to prevent timeout errors from masking actual
API errors. Affects discovery/file.go, phases/upgrade/health.go, and
util/apiclient/wait.go for consistent error reporting.
- Make it clear that FeatureGates are kubeadm-specific
- Add note that custom etcd image results in etcd upgrade skip
- Explain how is ClusterName is used
