v2.0.6

What's Changed

• dind template spec by Uchinda Padmaperuma (@uchinda-sph) in #43

Full Changelog: v2.0.5...v2.0.6

v2.0.4

What's Changed

• fix: duplicate values by Uchinda Padmaperuma (@uchinda-sph) in #42

Full Changelog: v2.0.3...v2.0.4

v2.0.3

What's Changed

• Render full template_spec when containerMode is dind by Uchinda Padmaperuma (@uchinda-sph) in #40

Full Changelog: v2.0.2...v2.0.3

v2.0.2

What's Changed

This release introduces support for custom pod annotations on the Actions Runner Controller (ARC) runner scale set pods and resolves a bug where custom resource limits/requests were ignored when running in Docker-in-Docker (DinD) mode.

---

🚀 Key Improvements & Fixes

1. Karpenter Disruption Prevention

• Added a new input variable runner_template_spec_metadata_annotations allowing custom annotations to be attached to the runner pod templates.
• Defaulted the annotations to include:

  {
    "karpenter.sh/do-not-disrupt": "true"
  }

  This prevents Karpenter from disrupting and terminating active runner nodes while GitHub Actions jobs are in progress.

2. Resolved DinD Resource Limit Bug

• Fixed a template validation bug in modules/gha-runner-scale-set/templates/values.yaml where the template.spec.containers block was completely omitted when containerMode.type = "dind".
• Custom resource constraints (defined via runner_resources / action_runner_resource) are now correctly generated, rendered, and enforced by the operator. Manual intervention (kubectl edit) is no longer required to provision custom CPU/memory sizes.

---

📂 Changed Files

• variables.tf — Added the new annotations variables.
• main.tf — Forwarded runner annotations to the scale set submodule.
• modules/gha-runner-scale-set/variables.tf — Declared annotations map.
• modules/gha-runner-scale-set/main.tf — Integrated annotation variables into local value computations.
• modules/gha-runner-scale-set/templates/values.yaml — Updated Helm payload formatting for DinD container configurations.

---

🛠️ Verification & Validation

• All static analysis checks via Checkov have succeeded.
• Terraform validations and lint rules passed successfully in CI/CD pipeline runs.

Full Changelog: v2.0.1...v2.0.2

v2.0.1

What's Changed

Feature: Configurable Runner Resource Requests and Limits

• Added runner_resources variable
  • Users can now configure resource requests and limits (CPU and Memory) for self-hosted GitHub Actions runners.
  • The new variable supports setting both requests and limits for cpu and memory.
  • Example default:

    runner_resources = {
      requests = {
        cpu    = "1"
        memory = "2Gi"
      }
      limits = {
        cpu    = "1"
        memory = "2Gi"
      }
    }

• Propagated configuration
  • The new variable is available at both the root and gha-runner-scale-set module levels.
  • Module logic updated to pass the resource settings into the generated pod spec.

Impact:
Existing deployments remain unaffected due to sensible defaults, but users wanting to fine-tune runner resource allocation can now do so easily.

Upgrade Guide:
No changes are needed unless you want to customize runner resources—just set the runner_resources variable in your configuration.

Full Changelog: v1.7.1...v2.0.1

v2.0.0

Breaking Change Release Note

Upgrade Minimum Helm Provider Version to v3.0+

This release introduces a breaking change:
The required version of the hashicorp/helm Terraform provider has been updated from >= 2.6, < 3.0 to >= 3.0 across all modules (gha-runner-scale-set-controller, gha-runner-scale-set, and root).
Impact:

• You must upgrade your Helm provider to version 3.0 or newer for all Terraform runs.
• Existing configurations using Helm provider v2.x will no longer be compatible and may fail during initialization or apply.

Example Required Update in Configuration

terraform {
  required_providers {
    helm = {
      source  = "hashicorp/helm"
      version = ">= 3.0"
    }
  }
}

Other Change

• Fixed the syntax for the kubernetes provider attribute in providers.tf by explicitly assigning as kubernetes = { ... }.

Migration Instructions

• Update your local and CI/CD Terraform Helm provider to >= 3.0.
• Review upstream dependency compatibility, ensuring you are not relying on deprecated provider features removed in Helm provider 3.x.

Action Required:
If you use this module, upgrade your Terraform Helm provider immediately to avoid deployment errors.
Refer to the Terraform Helm Provider v3.x documentation for guidance on migration and updated usage.

---

Full Changelog: v1.7.1...v2.0.0

v1.7.2

Full Changelog: v1.7.1...v1.7.2

v1.7.1

Release Notes

Changed

• DOCKER_GROUP_GID Environment Variable:
  Updated the value of DOCKER_GROUP_GID from 123 to 1001 in the runner environment. This aligns the Docker group with standard GID practices and improves compatibility with Docker socket permissions.

• Liveness Probe Tuning:
  Adjusted periodSeconds for the liveness probe from 10 to 5 seconds. This enables faster health checks and potentially quicker recovery from failures.

• Docker Socket Mount Path:
  Changed the Docker socket mountPath from /run/docker to /var/run to follow standard Docker conventions and ensure correct mounting of the Docker socket inside the runner container.

Full Changelog: v1.7.0...v1.7.1

v1.7.0

Release Notes

Features & Enhancements

• Chart Version Upgrades

  • Upgraded gha-runner-scale-set-controller and gha-runner-scale-set Helm chart versions from 0.9.3 to 0.12.1.
  • Updated Terraform Helm provider version constraint to >= 2.6, < 3.0.

• Runner and Controller Pod Customization

  • Introduced new variables for advanced scheduling and configuration:
    • controller_node_selector, controller_tolerations, controller_affinity, controller_topology_spread_constraints
    • runner_node_selector, runner_tolerations, runner_affinity, runner_topology_spread_constraints
    • runner_template_spec_metadata_labels, runner_template_spec_config_type, runner_container_mode_type
  • These enable fine-grained control over pod placement, affinity, tolerations, and topology spread for both controller and runner pods.

• PodSpec Refactor

  • Refactored how runner PodSpecs are constructed for better flexibility and maintainability.
  • Deprecated listener_podspec_map and custom_podspec_map in favor of new variable-driven configuration.

• Controller Enhancements

  • Added support for:
    • runner_max_concurrent_reconciles - Controls the maximum number of concurrent reconciles for the EphemeralRunner controller.
    • namespace_override - Allows overriding the release namespace for all resources in the controller chart.

Documentation

• README Updates
  • Documentation updated to include:
    • New variables and configuration options.
    • Revised tables and descriptions reflecting the latest capabilities and version constraints.

---

Migration Notes:

• If you rely on custom pod scheduling or topology, set runner_container_mode_type == "" and update your configuration as needed.
• The Helm provider version is now pinned and not compatible with Helm 3.x and above.

Full Changelog: v1.6.2...v1.7.0

v1.6.0

What's Changed

• Upgrade github actions runner helm code and chart to v2 by Poh Peng Wang (@thepoppingone) in #27
• Decouple arc-system and arc-runner Uchinda Padmaperuma (@uchinda-sph) in #30

Full Changelog: v1.4.2...v1.6.0