Make events etcd cluster optional

[codablock]

Related to #5004

The events cluster is not strictly needed anymore as etcd v3 is able to handle events and main in the same cluster. This PR only removes the strict requirement, but will still create the cluster via "kops create cluster". Future versions should consider not creating the cluster by default.

[k8s-ci-robot]

Hi @codablock. Thanks for your PR.

I'm waiting for a kubernetes member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[hakman]

/ok-to-test

[codablock]

Any chance to get this into 1.21? Would make things easier for us as I'm already actively using this

[hakman]

Hi @codablock. We will discuss about this today at our weekly meeting and get back to you.

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: hakman

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 [hakman]

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

[codablock]

@hakman It feels strange to solve it this way tbh. Replacing some IP (127.0.0.1) with a DNS name is not immediately clear why it is happening (it took me some time to understand it). Can't we do all this (127.0.0.1 vs DNS) in pkg/model/components/apiserver.go instead of later replacing it?

I'm also trying to understand the b.IsMaster, is this for apiservers running in dedicated instance groups?

[hakman]

@codablock I think we only get the info about the node type in NodeUp, so we can decide if we use 127.0.0.1 or hostname based on Master or APIServer type. In model we have a warning about adding a dependency on InstanceGroups.

kops/pkg/model/components/apiserver.go

Lines 186 to 191 in 1761076

	// But if we do, we end up with a weird dependency on InstanceGroups. We actually could tolerate
	// that in kops, but we don't really want to.
	// So instead, we assume that the etcd cluster size is the API Server Count.
	// We can re-examine this when we allow separate etcd clusters - at which time hopefully
	// the flag won't exist

@rifelpet @johngmyers @olemarkus any thoughts on which approach would be better?

[codablock]

@hakman I've update the PR to use your suggested solution (a few changes were required to make tests pass)

[johngmyers]

According to the apiserver command line reference, this flag is only in use when --endpoint-reconciler-type=master-count is enabled. The default is --endpoint-reconciler-type=lease and kOps does not support changing it from the default.

I believe we can remove the code for setting the --apiserver-count flag.

[hakman]

/lgtm