search

cncf / cluster

🖥🖥🖥🖥CNCF Community Cluster
https://cncf.io/cluster
152 stars 42 forks source link

Migrate Longhorn infrastructure to CNCF #139

Closed yasker closed 4 years ago

yasker commented 4 years ago

First and Last Name

Sheng Yang

Email

sheng@yasker.org

Company/Organization

Rancher Labs

Job Title

Software Architect and Engineering Manager

Project Title (i.e., summary of what do you want to do, not what is the name of the open source project you're working with)

Migrate Longhorn infrastructure to CNCF

Briefly describe the project (i.e., what is the detail of what you're planning to do with these servers?)

We want to migrate the infrastructure currently supports Longhorn from Rancher Labs to CNCF.

There are a few components:

  1. Drone for build/testing.
  2. Jenkins for nightly tests.
  3. Upgrade responder server for the upgrade checker feature.
  4. Potentially GitHub runner in the future.

Drone and Jenkins are also essential to the Longhorn development process, especially Jenkins nightly integration test result since it's currently not accessible by the public.

We can start with small items like the upgrade responder which is currently hosted by Rancher Labs. It notifies the end-user about a new version of Longhorn when it is available. Also, based on how many upgrade queries we received from the responder, we can also have an estimate on how many nodes are running Longhorn worldwide. We can provide that in a Grafana dashboard for the user to see the growth of Longhorn.

Is the code that you’re going to run 100% open source? If so, what is the URL or URLs where it is located? What is your association with that project?

Yes. The codes are located at https://github.com/longhorn/ . Longhorn is a CNCF sandbox project. I am a maintainer of Longhorn.

What kind of machines and how many do you expect to use (see: https://www.packet.com/bare-metal/)?

c3.small.x86 is more than enough. We want to go even smaller/cheaper if possible.

We want to start with one node dedicated to Rancher server and three nodes for Kubernetes cluster to run drone and other workloads. That's 4 nodes in total at the beginning. Those nodes should be good for any existing workloads except for the nightly test.

For nightly test, we will need two sets of four nodes to set up the test environment for the Kubernetes cluster, but the test jobs will only run about 5 hours per day. The test nodes will be released after the test completed (not sure if dynamically provisioning/destroy instance is supported though). We still need to migrate our infrastructure scripts from DO to Packet, so we can deal with this part later.

What OS and networking are you planning to use (see: https://support.packet.com/kb/articles/supported-operating-systems)?

Ubuntu 18.04

Any other relevant details we should know about?

No.

dankohn commented 4 years ago

Approved, thanks. Taylor will follow up with you.

yasker commented 4 years ago

Thank you @dankohn !

taylorwaggoner commented 4 years ago

@yasker I've created the Longhorn project in Packet and invited you to it. Please let me know if there are any other people you'd like invited. Thanks!

yasker commented 4 years ago

Hi @dankohn

It turns out c3.small.x86 is out of stock in Packet. The remaining low end c1.small.x86 is really weak. Can we use c3.medium.x86 instead? We're planning to provision a three-node Kubernetes cluster for now (instead of four we planned originally since we can use the cluster itself to host Rancher to save the cost of one machine). We also put the second part of the migration (from DO to Packet) on hold for now.

dankohn commented 4 years ago

Yes.

jacobsmith928 commented 4 years ago

Hi Sheng!

Feel free to hit me up if you need anything but for now great to use the medium. Once we have everything in Equinix facilities our capacity won’t be an issue. Thanks!

On Wed, Jun 10, 2020 at 8:36 PM Dan Kohn notifications@github.com wrote:

Yes.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/cncf/cluster/issues/139#issuecomment-642337591, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB46TKIHLLF6KW77QKNEUEDRWARJTANCNFSM4NZZSJLQ .

-- Jacob Smith VP, Bare Metal Strategy & Marketing Packet, an Equinix Company 267-226-8258 http://www.packet.com

yasker commented 4 years ago

Thanks @dankohn @jacobsmith928 , will do!

jacobsmith928 commented 4 years ago

How did everything go @yasker ?

yasker commented 4 years ago

@jacobsmith928 Working on it, so far so good.

Since this issue is a request and it has been approved, we should be OK to close this one and move the communication to offline e.g. slack?

jacobsmith928 commented 4 years ago

Yup, all good to close. @taylorwaggoner all you. :)

taylorwaggoner commented 4 years ago

Thanks!