search

pachyderm / helmchart

Helm Chart for Pachyderm
5 stars 9 forks source link

Enterprise server (Alpha 24) #121

Closed chainlink closed 3 years ago

chainlink commented 3 years ago

TODO: Add option to deploy pachd and enterprise server (or one or the other)

chainlink commented 3 years ago

@actgardner It looks like the enterprise server is still trying to read the main pachd configuration (getting STORAGE_BACKEND required still)

chainlink commented 3 years ago

Fixed in: https://github.com/pachyderm/pachyderm/pull/6393

chainlink commented 3 years ago

@actgardner this is the initial config for a separate enterprise server. I've tried it in minikube and it looks like it comes up successfully. There's currently no env vars included, but if there's some we'd like to plumb through that's fine too. Let me know if I'm missing anything (ingress? etc)

actgardner commented 3 years ago

The enterprise server uses some env vars. I think the biggest open question is around switching to kube dns in 2.0 (so we could switch ETCD_SERVICE_HOST to etcd:2379 for instance). Right now we use the following configs:

We might not care about plumbing through the ports. I'm not sure what the typical practice is around ingresses in helm charts - it seems like they're different enough in every cluster that it would be hard to make a generic template?

chainlink commented 3 years ago

This is what we have for ingress for the enterprise-server-in-pachd case https://github.com/pachyderm/helmchart/blob/v2.0.x/pachyderm/templates/dash/ingress.yaml#L29-L36

Would we also need this in the standalone enterprise server

(Right now the ingress annotations are tied to traefik for hub, but I'm going to make the ingress more generic)

chainlink commented 3 years ago

I'll add in those other Env vars and make em configurable

chainlink commented 3 years ago

Manifests now pass tests in main repo. Going to do remaining tasks as follow on.

chainlink commented 3 years ago

TODO: Change to deployTarget