HAPTIC¶
HAPTIC (HAProxy Template Ingress Controller) is a template-driven HAProxy Ingress Controller for Kubernetes that generates HAProxy configurations using Scriggo templates and deploys them via the HAProxy Dataplane API.
where possible
Community Project
This is an independent community project and is not affiliated with or endorsed by HAProxy Technologies.
What is HAPTIC?¶
HAPTIC is an event-driven Kubernetes controller that:
- Watches any Kubernetes resource - Ingresses, Services, Secrets, Gateway API resources, or any custom resource type you configure
- Renders Scriggo templates - A fast, Go-native template engine
- Validates before deployment - Deploy with confidence knowing configurations are validated before they reach your load balancers
- Deploys configurations to HAProxy pods via the Dataplane API
Unlike traditional ingress controllers with hardcoded configuration logic, HAPTIC uses a template-driven approach that gives you full control over the generated HAProxy configuration. This means you can:
- Define custom annotations that your platform users can use, implemented with just a few lines of template code
- Support new standards like Gateway API without waiting for controller updates
- Watch domain-specific CRDs and generate HAProxy configuration from any Kubernetes resource type
Key Features¶
Template-Driven Flexibility¶
Traditional ingress controllers embed configuration logic in code. HAPTIC inverts this:
- Full HAProxy access - Write any HAProxy directive using the configuration language
- Add features without code changes - New directives are template updates, not controller releases
- Iterate rapidly - Deploy configuration changes in minutes, not release cycles
- Rich template context - Access any Kubernetes resource, fetch external data via HTTP, and use controller state in your templates
- Everything is templatable - Generate not just
haproxy.cfgbut also map files, SSL certificates, CRT-lists, and custom auxiliary files
Production Ready¶
- High availability - Leader election with automatic failover
- Comprehensive validation - Validating webhook, template validation, and CI/CD-runnable tests
- Observability - Prometheus metrics, structured logging, and debug endpoints
- HTTP resource access - Fetch external data for use in templates
Ready to use out of the box
The Helm chart ships with Template Libraries enabled by default. They cover Kubernetes Ingress and Gateway API resources with annotation support comparable to existing HAProxy ingress controllers — no template authoring required. Customizing or extending the templates is entirely optional.
Architecture¶
The controller follows an event-driven architecture where changes to Kubernetes resources trigger a pipeline that renders templates, validates the output, and syncs configurations to HAProxy pods.
flowchart TB
subgraph API["Kubernetes API"]
R["Any Resource<br/>(Ingress, Gateway, CRDs, ...)"]
end
subgraph Controller["Controller Pod"]
W["Watcher"]
T["Template Engine"]
V["Validator"]
end
subgraph HAProxy["HAProxy Pod"]
D["Dataplane API"]
H["HAProxy"]
end
R --> W
W --> T
T --> V
V --> D
D --> H
Key components:
- Watcher - Subscribes to Kubernetes API for configured resource types
- Template Engine - Renders Scriggo templates with resource data as context
- Validator - Validates generated HAProxy configuration before deployment
- Dataplane Syncer - Applies configuration changes to HAProxy pods via the Dataplane API
Quick Start¶
helm install haptic oci://registry.gitlab.com/haproxy-haptic/haptic/charts/haptic --version 0.2.0-alpha.1 --namespace haptic --create-namespace
This installs both the controller and a 2-replica HAProxy Deployment, plus the default template libraries that cover Ingress and Gateway API out of the box. For the full walkthrough — including a sample app and end-to-end verification — see Getting Started.
Inspect the Deployed Configuration¶
Once you have an Ingress (or Gateway, HTTPRoute, …) the controller writes the rendered HAProxy config to a read-only HAProxyCfg CRD on every reconciliation:
CRD short names
HAProxyCfg (singular haproxycfg, short name hpcfg) is the output. The input — templates, watched resources, dataplane settings — lives in HAProxyTemplateConfig (short names htplcfg, haptpl). Edit that one, not HAProxyCfg. Use kubectl describe rather than kubectl get -o yaml, since the latter renders multiline configs as literal \n.
What Makes HAPTIC Different¶
The killer feature is templates. Suppose your platform users want a custom annotation that injects an X-Request-ID header for tracing. Add a snippet to your Helm values — no controller fork, no waiting for a release:
controller:
config:
templateSnippets:
# The frontend-filters-* glob picks this up automatically; the 300 prefix
# places it alongside the built-in header-manipulation snippets.
frontend-filters-300-request-id:
template: |
{%- for _, ingress := range resources.ingresses.List() %}
{%- var header = fallback(ingress.metadata.annotations["example.com/request-id-header"], "") %}
{%- if header != "" %}
http-request set-header {{ header }} %[uuid()]
{%- end %}
{%- end %}
Users opt in per-Ingress with example.com/request-id-header: "X-Request-ID". The same pattern works for rate limiting, header rewrites, custom ACLs — anything HAProxy can express. Override any snippet, replace the main template, or disable all libraries and start from scratch. See the Templating Guide.
Where to Go Next¶
- Essential: Getting Started → Templating → CRD Reference
- Custom resources beyond Ingress: Watching Resources
- Template tests for CI/CD: Validation Tests
- Reference: Supported Configuration, Troubleshooting
- Helm chart configuration: Deploying with Helm