syself / cluster-api-provider-hetzner

Cluster API Provider Hetzner 🚀 Kubernetes Infrastructure as Software 🔧 Terraform/Kubespray/kOps alternative for running Kubernetes on Hetzner
https://caph.syself.com
Apache License 2.0
539 stars 51 forks source link

:sparkles: Use LoadBalancer IPv6 address - take 2 #1334

Open rbjorklin opened 3 weeks ago

rbjorklin commented 3 weeks ago

What this PR does / why we need it: I'm trying to take over syself/cluster-api-provider-hetzner#1227 to push it to completion. Please see the original PR for a full description. If @JochemTSR returns I would be happy to move my commits back to the original PR.

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 #1218

Special notes for your reviewer: Even after making changes to the .envrc file as suggested in the "Running local e2e test" section I am unable to run the already existing e2e tests. Before I put any more time into troubleshooting on my side I would appreciate it if someone at Syself could review the instructions to ensure they are still up-to-date.

EDIT: Is a robot user required for e2e tests even if I'm not running the robot tests?

TODOs:

janiskemper commented 3 weeks ago

Thanks a lot @rbjorklin! @guettli I think you can have a look here

guettli commented 2 weeks ago

@rbjorklin @JochemTSR What about having a meeting where we talk about the goals and strategies to achieve them? Unfortunately, I will be on vacation from Friday afternoon. I will be back on July 8th. We could meet today, tomorrow, or in two weeks. Maybe you two could sync up first? It would be great if you could create an overview, maybe not in GitHub, but in a Google Doc.

Here are some questions that come to mind:

I know some of these questions are not related to this PR, but I would like to have a better understanding of how community members use our CAPI provider (aside from our commercial customers).

JochemTSR commented 2 weeks ago

@rbjorklin @JochemTSR What about having a meeting where we talk about the goals and strategies to achieve them? Unfortunately, I will be on vacation from Friday afternoon. I will be back on July 8th. We could meet today, tomorrow, or in two weeks. Maybe you two could sync up first? It would be great if you could create an overview, maybe not in GitHub, but in a Google Doc.

Here are some questions that come to mind:

* Do you plan to use DNS entries for the LoadBalancer?

* Will you use an IPv6-only LB or dual-stack?

* What about the Kubernetes network: IPv6-only, dual-stack, or IPv4-only?

* Why do you want that: Do you want to save costs? Do you want to provide IPv6 services?

* Which locations do you use?

* What about Hetzner bare-metal?

* Do you use Kubernetes with IPv6 in other providers?

* Are you using AMD only, or do you plan to use ARM as well?

I know some of these questions are not related to this PR, but I would like to have a better understanding of how community members use our CAPI provider (aside from our commercial customers).

@guettli @rbjorklin I made a quick write-up here: (https://cloud.jochemram.net/s/gkcfLMbkpExcCfr). I am available for a meeting tomorrow anytime after 5PM UTC. Please reach out at dev@jochemram.net for contact details or for editing access to the document.

Regards, Jochem

guettli commented 1 week ago

@guettli @rbjorklin I made a quick write-up here: (https://cloud.jochemram.net/s/gkcfLMbkpExcCfr). I am available for a meeting tomorrow anytime after 5PM UTC. Please reach out at dev@jochemram.net for contact details or for editing access to the document.

Regards, Jochem

Hi Jochen, I don't email a lot, so that I propose another way: we could use the Kubernetes Slack. I am "Thomas Güttler" there. I searched for your name, but have not found it.

Friday 5PM UTC is too late for me. I can answer via Slack, but I can't dive into the source code. I will be back 8th Juli. But please contact me today via Kube Slack, and send me the password for above document. Thank you.