🐛fix ClusterCache doesn't pick latest kubeconfig secret proactively

[mogliang]

What this PR does / why we need it:

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 #12399

/area clustercache

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: mogliang / name: Mog Liang (8ce65c7, 8e7dc16, 754d34a)

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
Once this PR has been reviewed and has the lgtm label, please assign enxebre for approval. For more information see the Code Review Process.

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

Needs approval from an approver in each of these files:

• OWNERS

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

[k8s-ci-robot]

Welcome @mogliang!

It looks like this is your first PR to kubernetes-sigs/cluster-api 🎉. Please refer to our pull request process documentation to help your PR have a smooth ride to approval.

You will be prompted by a bot to use commands during the review process. Do not be afraid to follow the prompts! It is okay to experiment. Here is the bot commands documentation.

You can also check if kubernetes-sigs/cluster-api has its own contribution guidelines.

You may want to refer to our testing guide if you run into trouble with your tests not passing.

If you are having difficulty getting your pull request seen, please follow the recommended escalation practices. Also, for tips and tricks in the contribution process you may want to read the Kubernetes contributor cheat sheet. We want to make sure your contribution gets all the attention it needs!

Thank you, and welcome to Kubernetes. 😃

[k8s-ci-robot]

Hi @mogliang. Thanks for your PR.

I'm waiting for a kubernetes-sigs 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-sigs/prow repository.

[mogliang]

added unit test

[sivchari]

/ok-to-test

[sbueringer]

You are hitting tooManyConsecutiveFailures in your scenario, right?

Would it also be enough for your use case to make HealthProbe.Timeout/Interval/FailureThreshold configurable?

[mogliang]

not really.
We have proxy between mgmt cluster and target clusters.
we see after kubeconfig updated (proxy address changed), the existing connection (clustercache probe) still works, so it doesn't refetch kubeconfig and still cache the old one, but new connections fails (etcd client)

[sbueringer]

So if the health check would open a new connection it would detect it, right?

[mogliang]

that's correct~ another idea is to disconnect for a given time (e.g. 5m) period to force refresh connection. will this be better?

[sbueringer]

I would just try to extend the health check that it does two requests. One over the existing connection one with a new connection

[mogliang]

let it check it

[mogliang]

@sbueringer
https://github.com/kubernetes-sigs/cluster-api/compare/release-1.9...mogliang:cluster-api:dev/qliang/v1.9.5-fixetcd2?expand=1

copied the implemenation from tlscache.get.

need mention that, establish tcp connection takes ~1sec in our case, and normal ping may take ~100ms. so, this do adds some reconcile time.

[sbueringer]

Oh wow. 1s is really not good :)
And yeah it's a lot of code to copy

[sbueringer]

Do we know what the etcd client is doing in the case that is failing for you? I would have hoped that that doesn't take 1s (at least in our scale tests the entire KCP reconcile was much faster than that)

Maybe we can do something more lightweight

[mogliang]

our env is a bit complex, the mgmt cluster is on cloud side, the target cluster are on users' local office. an relay connection is needed between them.

When mgmt trying to estalbish new tcp connection to target cluster, there are a few work need be done on relay proxy side, so the time taken is a bit long.

From what i've seen, normally kubeclient will reuse the underlying tcp connection. However, when kcp use etcd client, it tries to establish the new underlying tcp connections. that may take a few more time.

but what blocks us is that: the kubeconfig in clustercache may already outdated (don't have latest relay info), and etcd client cannot use it to create connection.

[mogliang]

/hold