haproxy/deploy/test/README.md

Helm Chart Testing

Three test scripts are provided:

1. local-test.sh — Offline lint + template validation (no cluster needed)
2. integration-test.sh — Deploy to a real Kind cluster and verify resources
3. ct-test.sh — Wrapper around chart-testing (ct), matching what CircleCI runs

1. Offline Tests (local-test.sh)

Runs helm lint and helm template — no cluster required.

Prerequisites

• Helm v3.x

Usage

# Test all charts
./test/local-test.sh

# Test a specific chart
./test/local-test.sh haproxy-unified-gateway
./test/local-test.sh kubernetes-ingress

What It Tests

Test	Description
Chart.yaml metadata	Verifies required fields (name, version, appVersion, description)
helm lint (defaults)	Runs helm lint with default values
helm template (defaults)	Renders templates and checks for errors
Deployment vs DaemonSet	Confirms charts that support controller.kind render the correct resource type
HugConf cleanup hook	Verifies cleanup hook renders/skips based on hugconf.create and targets correct name
Metrics port	Verifies metrics container port (31060) renders, --metrics-auth flag is correct for both kube-rbac (default) and none, metrics Service has both stat and metrics ports
ServiceMonitor/PodMonitor	Renders with monitoring.coreos.com/v1 API, skipped without it, metrics Service only with ServiceMonitor
CI values (lint + template)	Lint and template for every ci/*.yaml file

---

2. Integration Tests (integration-test.sh)

Installs charts into a real Kubernetes cluster, waits for pods to become Ready, verifies key resources exist, then cleans up.

By default the script creates a Kind cluster named dev-helm-charts, runs all tests, and deletes the cluster when done.

Prerequisites

• Kind
• Helm v3.x
• kubectl

Usage

# Test all charts (creates + destroys Kind cluster automatically)
./test/integration-test.sh

# Test a specific chart
./test/integration-test.sh haproxy-unified-gateway
./test/integration-test.sh kubernetes-ingress

Environment Variables

Variable	Default	Description
TIMEOUT	120	Max seconds to wait for pods per install
KEEP_NS	0	Set to 1 to keep test namespaces after run (for debugging)
CI_FILTER	(empty)	Glob pattern to filter ci/ values files (e.g. deployment-*)
TEST_FILTER	(empty)	Run only a specific test: defaults, daemonset, hpa, pdb, metrics-port, monitoring, hugconf-cleanup, ci
KIND_KEEP_CLUSTER	0	Set to 1 to keep the Kind cluster after tests finish
KIND_CLUSTER_NAME	dev-helm-charts	Custom Kind cluster name
SKIP_KIND	0	Set to 1 to skip Kind management and use your existing kubeconfig

Examples

# Run tests and keep the Kind cluster for manual inspection afterwards
KIND_KEEP_CLUSTER=1 ./test/integration-test.sh haproxy-unified-gateway

# Re-run tests against the cluster you kept (reuses existing cluster)
./test/integration-test.sh haproxy-unified-gateway

# Delete the cluster manually when done
kind delete cluster --name dev-helm-charts

# Use your own cluster instead of Kind
SKIP_KIND=1 ./test/integration-test.sh haproxy-unified-gateway

# Longer timeout for slow clusters
TIMEOUT=300 ./test/integration-test.sh haproxy-unified-gateway

# Only run deployment-related ci/ test cases
CI_FILTER="deployment-*" ./test/integration-test.sh haproxy-unified-gateway

# Run only the monitoring test (ServiceMonitor/PodMonitor)
TEST_FILTER=monitoring ./test/integration-test.sh haproxy-unified-gateway

# Run only the metrics port test
TEST_FILTER=metrics-port ./test/integration-test.sh haproxy-unified-gateway

# Run only the HugConf cleanup test
TEST_FILTER=hugconf-cleanup ./test/integration-test.sh haproxy-unified-gateway

# Run only the defaults test
TEST_FILTER=defaults ./test/integration-test.sh haproxy-unified-gateway

# Keep namespaces after the run for manual inspection
KEEP_NS=1 ./test/integration-test.sh haproxy-unified-gateway

# Then clean up manually when done
kubectl get ns | grep '^it-' | awk '{print $1}' | xargs kubectl delete ns

What It Tests

For each chart, the script runs these phases:

Test	Description
Install (defaults)	helm install with default values, wait for pods, verify resources
Install (DaemonSet)	Same but with controller.kind=DaemonSet (if chart supports it)
Install (HPA)	Enables HPA, verifies HPA resource is created
Install (PDB)	Enables PDB, verifies PDB resource is created
Metrics port	Verifies controller pod has metrics container port 31060, --metrics-auth=kube-rbac arg, and main Service has stat port
Install (monitoring)	Installs Prometheus Operator CRDs, verifies ServiceMonitor/PodMonitor/metrics Service with both stat + metrics ports
HugConf cleanup	Verifies HugConf CR is deleted after helm uninstall
Install ci/ values	Installs with each ci/*.yaml file, waits for pods

Each test:

1. Creates an isolated namespace (it-<chart>-<suffix>-<timestamp>)
2. Runs helm install --wait
3. Verifies pods reach Running/Completed state
4. On failure: dumps pod status, events, and container logs
5. Runs helm uninstall and deletes the namespace

On Failure

When a test fails, the script automatically prints:

• Pod status (kubectl get pods -o wide)
• Recent events (kubectl get events)
• Container logs from unhealthy pods (last 20 lines)

---

3. Chart Testing (ct-test.sh)

Wrapper around chart-testing (ct) — the same tool CircleCI uses. Runs ct lint and ct install locally so you can validate before pushing.

Prerequisites

• ct v3.x
• yamale (pipx install yamale)
• yamllint (pipx install yamllint)
• Helm v3.x
• Kind — only for install mode
• kubectl — only for install mode

ct Usage

# Lint all charts
./test/ct-test.sh lint

# Lint a specific chart
./test/ct-test.sh lint haproxy-unified-gateway

# Install all charts on a Kind cluster (creates + destroys automatically)
./test/ct-test.sh install

# Install a specific chart
./test/ct-test.sh install haproxy-unified-gateway

# Lint + install in one go
./test/ct-test.sh all haproxy-unified-gateway

ct Environment Variables

Variable	Default	Description
KIND_CLUSTER_NAME	ct-dev	Kind cluster name
KIND_KEEP_CLUSTER	0	Set to 1 to keep the Kind cluster after tests
SKIP_KIND	0	Set to 1 to skip Kind management and use existing kubeconfig
CT_ARGS	(empty)	Extra arguments passed to ct (e.g. --debug)

ct Examples

# Lint with debug output
CT_ARGS="--debug" ./test/ct-test.sh lint haproxy-unified-gateway

# Install and keep cluster for inspection
KIND_KEEP_CLUSTER=1 ./test/ct-test.sh install haproxy-unified-gateway

# Use an existing cluster
SKIP_KIND=1 ./test/ct-test.sh install haproxy-unified-gateway

# Delete the cluster manually when done
kind delete cluster --name ct-dev

CircleCI Compatibility

The CircleCI pipeline (.circleci/config.yml) runs:

1. ct lint --all — validates Chart.yaml schema, runs helm lint with each ci/*.yaml file
2. ct install --all — installs each chart with each ci/*.yaml file on a Kind cluster

ct-test.sh runs the exact same commands locally using the config in test/ct.yaml, so if it passes locally it will pass in CI.

---

CI Values Files

Both scripts use the ci/ values files in each chart directory. These are the same files used by chart-testing (ct) in CircleCI.

Naming convention:

<mode>-<feature>-values.yaml

Examples:

• deployment-default-values.yaml — Deployment with defaults
• daemonset-hostport-values.yaml — DaemonSet with host port mapping
• deployment-hpa-values.yaml — Deployment with HPA enabled

Adding New Test Cases

1. Create a new values file in <chart>/ci/ following the naming convention.

2. Validate offline:

   ./test/local-test.sh <chart-name>
   

3. Verify ct lint passes (matches CircleCI):

   ./test/ct-test.sh lint <chart-name>
   

4. Deploy to a cluster:

   CI_FILTER="<your-new-file>" ./test/integration-test.sh <chart-name>
   

5. The file is automatically picked up by all three scripts and CircleCI's ct lint.