@creamy-potato-29402 @white-balloon-205 Guys, there are lot of open issue injunction with HELM/K8s Charts install. Namespace issue, now i had to handle capability issue....These come from the separated backend. "helm template" uses its own engine and "helm install" uses the tiller. We would have an idea with my college @cold-coat-35200. Can you guys put an option into the chart resource for example: templateBackend: "tiller|template" => if we use tiller, helm install --dry-run --debug can generate the resource files, I think...maybe it's not easy to parse, but should not be impossible...it could be a good option to solve a lots of issue in that case the pulumi client can reach the particular tiller to ask for a proper output. K8s does the same with its API (it's quite new: https://kubernetes.io/blog/2019/01/14/apiserver-dry-run-and-kubectl-diff/ ) it should be the same story with tiller. What do you think? Bullshit? 🙂

This is definitely something we should consider. @creamy-potato-29402 and I discussed this a couple months ago, and I do think having the option would help ensure that Helm charts that really deeply rely on Tiller can still be installed via Pulumi with 100% fidelity. I opened https://github.com/pulumi/pulumi-kubernetes/issues/511 to track - feel free to add any additional notes there.

@cuddly-eye-68174 We’ve been waiting for the community to tell us they really want tiller support — we still don’t really hear much about this, but it’s porobably nice to have, so I’m guessing this would come after the namespacing issues.

@white-balloon-205 @creamy-potato-29402 based on the linked issue we may misunderstood each other, we're not saying to use tiller for creating the resources, just saying to use

helm install --dry-run --debug

, get the manifest part of the output, which is basically the same as the

helm template

output, except it has the release name and the correct capabilities fetched through tiller, and use that template to create the resources with pulumi

@white-balloon-205 actually I was meaning to suggest the same process that you have but use helm install --dry-run --debug to get the resource output... We like the Pulumi approach here 🙂 so i would not suggest to drop the current resource handling

Ahh - interesting. Sorry I misunderstood. Want to add a note on that to the issue?

I was suggesting something like this

that is a very interesting story

🙂

I added a note to the issue with a sample diff between the 2 command

thanks!