[v2.11] add prime mode logic to the provisioning-tests script

[jferrazbr]

Issue #51881

Prepare KDM CI and test scripts to work when KDM marks Prime-only Kubernetes versions. Same idea as PR #1822 for v2.9, adapted per branch.

Problem

KDM CI and test scripts needs to:

• Detect when a selected Kubernetes version requires Prime.
• Choose staging or production Prime registry.
• Fail early with a clear message if Prime is required but no registry endpoint is available in the environment.
• Keep non-Prime behavior unchanged.

Solution

• Prime detection in scripts:

  • Read channels, compare the release item’s "minChannelServerVersion" with "LAST_COMMUNITY_RANCHER" to decide Prime.
  • Pick "staging" or "production" Prime based on the version tag pattern.
  • Set "PRIME_MODE" and "PRIME_REG_HOST" for downstream provisioning test.

• Early guard:

  • If "PRIME_MODE" is 1 but no registry endpoint is present, stop with a user-friendly message that explains why.

• Defaults and safety:

  • Respect an already set "CATTLE_AGENT_IMAGE" from the environment. Only set a default if it is empty.
  • Keep normal flow when Prime is not required.

• Workflow integration:

  • Variables "LAST_COMMUNITY_RANCHER" and "PRIME_AGENT_IMAGE" are passed from the workflow env. Scripts keep safe defaults for local runs.

[jferrazbr]

Green run with prime test from this commit ✔️

[Copilot]

Pull request overview

This PR adds Prime mode detection and registry routing logic to KDM provisioning tests for the v2.11 branch. When a selected Kubernetes version requires Rancher Prime (determined by comparing the version's minChannelServerVersion against a configured community cutoff), the scripts automatically select the appropriate Prime registry (staging or production) and configure the agent image accordingly.

Key changes:

• Adds version comparison utilities (AWK scripts) to detect Prime-only Kubernetes versions based on channel metadata
• Implements Prime registry selection logic with staging/production routing based on version tags
• Integrates early failure checks when Prime is required but registry credentials are unavailable (common in fork PRs)

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 12 comments.

Show a summary per file

File	Description
scripts/semver_g.awk	New AWK utility for semantic version comparison (A>B logic)
scripts/channels-pick-minchan-for-version.awk	New AWK utility to extract minChannelServerVersion from channel YAML
scripts/prime-route	Core Prime detection logic that sets PRIME_MODE, selects registry environment, and configures agent image
scripts/provisioning-tests	Integration point that sources prime-route and implements early failure for missing registry credentials
.github/workflows/provisioning-tests.yaml	Adds Vault secrets retrieval for Prime registries and sets LAST_COMMUNITY_RANCHER/PRIME_AGENT_IMAGE env vars
Dockerfile.dapper	Passes through new environment variables (LAST_COMMUNITY_RANCHER, PRIME_AGENT_IMAGE, STAGE_REGISTRY_ENDPOINT)
docs/release.md	Documents Prime configuration variables, their purpose, and maintenance instructions for release managers

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.