This article is going to walk you through the steps required to make Harness detect Deployed Instances after a Helm Native Deployment. We will help Harness with a label, but we need a Feature Flag to make it work.
Assuming k8s cluster version > 1.16, there are two requirements:
- Enable FF HELM_STEADY_STATE_CHECK_1_16;
- Update the deployment manifest to have a “release” label.
First, let’s make sure the target Account has “HELM_STEADY_STATE_CHECK_1_16” enabled:
Now, we need to follow this doc here.
You must add a release label on all Manifests related to Deployable Objects!
I used Bitnami Nginx because their Helm Chart has a few shortcuts (with Go Templates) to add labels everywhere. I just added the release label, and submitted the hacked Chart on a GCS Lab:
Important: The only critical label to achieve what we need is the “release” one. You may skip that “harness.io/release-name”, ok?
I’ll run a Basic Deployment, using a Helm Native Service:
And you can see that it takes the Pod correctly!
With that done, now you are able to keep using Helm Native, and explore the features that need Deployed Instances as a requirement.
<cloud: aws, azure, tanzu, gcp, on-prem, kubernetes, helm, charts>
<category: triggers, gitops, templates>