Use deployment results

This guide explains how to use deployment results in running application code. Deployment results are values that deployers produce while deploying a vector, such as service endpoints, generated URLs, identities, and allocated resources.

You do not author deployment results in vector configuration. Konfidence computes them as a side effect of deploying the vector and merges them into the same vector data bundle that applications read at runtime.

Prerequisites

A deployer produces deployment results for an artifact in the vector.
The producing artifact has been deployed; results are available from that point (including during the migration phase), not only after the vector is activated.
Your application can access vector data through the vector data service.

Let deployers produce deployment results

Each producing deployer writes deployment results to the corresponding ArtifactDeployment.

Konfidence aggregates those deployment results into the VectorData custom resource through the VectorDeployment status. The landscape orchestrator reads the VectorData resource and passes the data to the landscape as a ConfigMap.

Expose a Kubernetes Service

To let other components in the vector discover a Service, add the konfidence.cloud/deployment-result annotation to it. The annotation value is the stable name consumers use to look it up:

yaml

apiVersion: v1
kind: Service
metadata:
  name: candidates
  annotations:
    konfidence.cloud/deployment-result: candidates

Only Kubernetes Services can be exposed this way today. For how the deployer turns the annotation into a deployment result, see Expose a Service as a deployment result.

A deployment result is identified by the pair (name, type), which must be unique within a component. Giving two Services in the same component the same annotation value is rejected at deploy time.

Read deployment results at runtime

Deployment results are part of the vector configuration bundle. Request the whole bundle by passing the vector ID as the flag key (see Access vector data in your application).

In the returned object, deploymentResults is keyed by component name (its last path segment), and each value is an array of the results that component published. Each result has a stable name (the annotation value), a type, and a spec. Select the result you need by its name — do not rely on array position.

For an http-k8s-service result, spec carries the Service's Namespace, K8sName, and ServicePorts. Build the in-cluster address from those fields and a port taken from ServicePorts; do not hard-code the port:

json

{
  "deploymentResults": {
    "candidates": [
      {
        "name": "candidates",
        "type": "http-k8s-service",
        "spec": {
          "Namespace": "vector-42",
          "K8sName": "candidates-7f3a",
          "ServicePorts": [{ "name": "http", "port": 80 }]
        }
      }
    ]
  }
}

Here a consumer looks up the result named candidates and resolves it to candidates-7f3a.vector-42.svc.cluster.local:80.

The single-flag and bulk endpoints resolve feature flags only. Deployment results and authored config are available through the whole-bundle response.