Manifest a Management Cluster : Equinix based similar to pair management cluster

[hh]

This will be written as a manifest or similar creating a management cluster similar to pair:

• [x] Firstly it should deploy clusters similarly to our pair go code
• [ ] It may pull in manifests and parts of the pair helm chart
• [x] Lilkely a cluster-api-packet deployment of the above
• [ ] Should then include the sharingio/coder/.sharingio/init manifest etc we currently use the run coder on pair
• [x] DNS should work similarly to they way it does on the pair management cluster

[BobyMCbobs]

Currently setting up a cluster as is defined in https://github.com/cncf-infra/coder-infra The cloud-provider-equinix-metal controller isn't starting although it's priority class is system-role-critical

[hh]

Still not starting?

[BobyMCbobs]

Still not starting?

This issue is still occurring. I'm working on it

[BobyMCbobs]

There's an auto-import for the kubernetes template, when using the Helm chart https://github.com/coder/coder/pull/3550 It appears to only support the kubernetes template.

Unsure how it pulls it in

[BobyMCbobs]

What domain will the Coder server run on?

[hh]

I think it's fine to have it be a terraform variable or similar input.

It should configure the NS entries over at dnsimple using an api key.

For now you can use coder.sharing.io

[BobyMCbobs]

Thank you for your response @hh

Terraform over-complicates it at this point, so just doing plain or envsubst'd yaml.

Gonna bring up with PowerDNS, since this has been previously discussed. Supported DNS01 providers https://cert-manager.io/docs/configuration/acme/dns01/#supported-dns01-providers We'd wanna use RFC2136 with PowerDNS (or Route53)

I'll be bringing it up on https://coder.sharing.io then.

[hh]

I like it!

[BobyMCbobs]

progress https://github.com/cncf-infra/coder-infra/issues/1

[BobyMCbobs]

Managed to get a cluster up using cluster-api-provider-packet with

• cilium
• cluster-api
• cert-manager
• external-dns
• ingress-nginx
• kubevirt
• longhorn
• metallb
• powerdns
• stackgres
• Coder

Currently there's issue for Coder to talk to the Postgres cluster. Trying to figure that out. Have also reached out to the Coder folks

[github-actions[bot]]

This issue is becoming stale. In order to keep the tracker readable and actionable, I'm going close to this issue in 7 days if there isn't more activity.