✨ support NodeRegistrationOptions.IgnorePreflightErrors

[sbueringer]

Signed-off-by: Stefan Büringer buringerst@vmware.com

What this PR does / why we need it:
For details see linked issue

Which issue(s) this PR fixes (optional, in fixes #<issue number>(, fixes #<issue_number>, ...) format, will close the issue(s) when PR gets merged):
Fixes #4581

[sbueringer]

/cc @vincepri @fabriziopandini
If I'm not mistaken, the IgnorePreflightErrors field was already added with kubeadm v1beta2 (I only had to add it to our structs not to our copies of the v1beta2 structs).

I assume as we stopped supporting kubeadm v1beta1 with v1alpha4 (i.e. Kubernetes versions < 1.15), it doesn't make sense to add validation for the v1beta1 case?

[sbueringer]

/retest

[fabriziopandini]

If I'm not mistaken, the IgnorePreflightErrors field was already added with kubeadm v1beta2

You are right; given that it already existed in v1beta2 and now also in v1alpha4, what about cleaning up following custom conversion func

cluster-api/bootstrap/kubeadm/types/v1beta2/conversion.go

Lines 73 to 96 in db47368

	func Convert_v1beta2_InitConfiguration_To_v1alpha4_InitConfiguration(in *InitConfiguration, out *bootstrapv1.InitConfiguration, s apimachineryconversion.Scope) error {
	// InitConfiguration.CertificateKey exists in v1beta2 types but not in bootstrapv1.InitConfiguration (Cluster API does not uses automatic copy certs). Ignoring when converting.
	return autoConvert_v1beta2_InitConfiguration_To_v1alpha4_InitConfiguration(in, out, s)
	}
	func Convert_v1beta2_NodeRegistrationOptions_To_v1alpha4_NodeRegistrationOptions(in *NodeRegistrationOptions, out *bootstrapv1.NodeRegistrationOptions, s apimachineryconversion.Scope) error {
	// NodeRegistrationOptions.IgnorePreflightErrors exists in v1beta2 types but not in bootstrapv1.NodeRegistrationOptions (Cluster API does not support it for now). Ignoring when converting.
	return autoConvert_v1beta2_NodeRegistrationOptions_To_v1alpha4_NodeRegistrationOptions(in, out, s)
	}
	func Convert_v1beta2_JoinControlPlane_To_v1alpha4_JoinControlPlane(in *JoinControlPlane, out *bootstrapv1.JoinControlPlane, s apimachineryconversion.Scope) error {
	// JoinControlPlane.CertificateKey exists in v1beta2 types but not in bootstrapv1.JoinControlPlane (Cluster API does not uses automatic copy certs). Ignoring when converting.
	return autoConvert_v1beta2_JoinControlPlane_To_v1alpha4_JoinControlPlane(in, out, s)
	}
	func Convert_v1beta2_DNS_To_v1alpha4_DNS(in *DNS, out *bootstrapv1.DNS, s apimachineryconversion.Scope) error {
	// DNS.Type was removed in v1alpha4 because only CoreDNS is supported, dropping this info.
	return autoConvert_v1beta2_DNS_To_v1alpha4_DNS(in, out, s)
	}
	func Convert_v1beta2_ClusterConfiguration_To_v1alpha4_ClusterConfiguration(in *ClusterConfiguration, out *bootstrapv1.ClusterConfiguration, s apimachineryconversion.Scope) error {
	// ClusterConfiguration.UseHyperKubeImage was removed in kubeadm v1alpha4 API
	return autoConvert_v1beta2_ClusterConfiguration_To_v1alpha4_ClusterConfiguration(in, out, s)
	}

Eventually also v1beta2 <-> v1alpha4 conversion tests can now be cleaned up (

cluster-api/bootstrap/kubeadm/types/v1beta2/conversion_test.go

Lines 71 to 76 in 29333cd

	func nodeRegistrationOptionsFuzzer(obj *NodeRegistrationOptions, c fuzz.Continue) {
	c.FuzzNoCustom(obj)
	// NodeRegistrationOptions.IgnorePreflightErrors does not exists in v1alpha4, so setting it to nil in order to avoid v1beta2 --> v1alpha4 --> v1beta2 round trip errors.
	obj.IgnorePreflightErrors = nil
	}

)

[sbueringer]

@fabriziopandini dropped the conversion and fuzzer func for IgnorePreflight, ptal

[sbueringer]

@fabriziopandini ping In case you have some time. Not sure who else wants to review / to cc.

[vincepri]

Shouldn't we restore the field during conversion instead of dropping it?

[sbueringer]

@vincepri
Very nice finding. I implemented it for the top-level types: KubeadmConfig, KubeadmConfigTemplate and KubeadmControlPlane.

Given that in the current case we convert between InitConfigurations and JoinConfigurations and they don't have ObjectMeta I'm not sure if it's possible in this case. (and if it makes sense to implement it)

[sbueringer]

@vincepri PTAL :) (push --force because of rebase onto current main)

[sbueringer]

/test pull-cluster-api-test-main
flakes

[fabriziopandini]

/lgtm

[vincepri]

/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: vincepri

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [vincepri]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[sbueringer]

@vincepri @fabriziopandini squashed the commits, so I need a lgtm again

[sbueringer]

/retest

[fabriziopandini]

/lgtm