feat: add metrics support

[Patrick Derks (TrayserCassa)]

No description provided.

[Copilot]

Pull request overview

This PR enables Prometheus metrics for the Shopware operator by adding a metrics collection package, wiring it into the reconciliation flow, and exposing the controller-runtime metrics endpoint via configuration and Helm.

Changes:

• Add internal/metrics with Prometheus gauges for Store state, deployment state/replicas, HPA settings, and scheduled task (CronJob) status.
• Update controllers to publish metrics on status updates and clean up metrics on Store deletion.
• Enable controller-runtime metrics server binding via config (METRICS_BIND_ADDRESS) and add a Helm Service intended to expose the metrics endpoint.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 8 comments.

Show a summary per file

File	Description
internal/metrics/metrics.go	New Prometheus metrics definitions + update/remove helpers for Store and CronJob status.
internal/controller/store_status.go	Emit metrics during status reconciliation; fetch CronJob for scheduled-task metrics.
internal/controller/store_controller.go	Remove metrics when a Store is being deleted.
internal/config/config.go	Default metrics bind address changed to :8080.
cmd/main.go	Enable controller-runtime metrics server using cfg.MetricsAddr.
helm/templates/metrics-service.yaml	New Service intended to expose the metrics endpoint.
helm/values.yaml	Add storeAPI.port value (currently tied to metrics Service).
go.mod	Promote prometheus/client_golang to a direct dependency.

---

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

[Copilot]

Pull request overview

Copilot reviewed 8 out of 8 changed files in this pull request and generated 5 comments.

---

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

[Copilot]

Pull request overview

Copilot reviewed 10 out of 10 changed files in this pull request and generated 7 comments.

---

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

[Copilot]

Pull request overview

Copilot reviewed 11 out of 11 changed files in this pull request and generated no new comments.

Suppressed comments (9)

internal/config/config.go:67

• Changing the default metrics bind address to :8080 enables the metrics listener even when the Helm chart has metrics.enabled: false, which contradicts the chart's opt-in semantics and can unintentionally expose metrics on the pod network. Consider keeping the default disabled ("0") and enabling it via Helm/env when needed.

	MetricsAddr string `env:"METRICS_BIND_ADDRESS, default=:8080"`

helm/values.yaml:118

• Same env-var naming mismatch here: the chart docs mention K8S_OPERATOR_URL, but the operator injects SHOPWARE_OPERATOR_URL.

  # shopwareOperatorUrl overrides the K8S_OPERATOR_URL injected into store
  # containers (admin, storefront, worker). When left empty the URL is

internal/metrics/metrics.go:59

• UpdateScheduledTaskMetrics sets shopware_store_scheduled_task_suspended to 0 when the CronJob is nil/not found, but the help string describes 0 as "active". This makes the metric description inaccurate for the "missing CronJob" case.

	storeScheduledTaskSuspended = prometheus.NewGaugeVec(prometheus.GaugeOpts{
		Name: "shopware_store_scheduled_task_suspended",
		Help: "Whether the scheduled task CronJob is suspended (1) or active (0)",
	}, []string{"store", "namespace"})

internal/metrics/metrics.go:165

• The code parses both available and desired from cond.Ready, but only available is used (desired comes from cond.StoreReplicas). Passing a throwaway pointer makes the intent clearer and avoids the misleading local desired variable.

	// Parse available/desired from the Ready field (format: "available/desired")
	var available, desired int
	if cond.Ready != "" {
		fmtScan(cond.Ready, &available, &desired)
	}

internal/metrics/metrics.go:205

• This new metrics package adds non-trivial behavior (state gauges, deployment replica parsing, and CronJob last-run status logic) but there are no accompanying unit tests. The repo already runs go test ./... (Makefile:89), so adding coverage here would help prevent regressions.

// UpdateScheduledTaskMetrics sets metrics for the scheduled task CronJob.
func UpdateScheduledTaskMetrics(store *v1.Store, cronJob *batchv1.CronJob) {
	name := store.Name
	ns := store.Namespace

helm/templates/metrics-service.yaml:27

• The Service uses a configurable .Values.metrics.port, but targetPort is hardcoded to 8080. If the port is changed, the Service will route to the wrong container port and metrics scraping will fail.

  ports:
    - name: http-metrics
      port: {{ .Values.metrics.port | default 8080 }}
      targetPort: 8080
      protocol: TCP

helm/templates/deployment.yaml:90

• .Values.metrics.port controls the exposed container port and the constructed operator URL, but the deployment does not set METRICS_BIND_ADDRESS. The manager will still bind to the default from code/config, which breaks custom ports and can desync from the Service.

        {{- if .Values.metrics.enabled }}
        - name: OPERATOR_SERVICE_URL
          value: "{{ .Values.metrics.shopwareOperatorUrl | default (printf "http://shopware-operator.%s.svc.cluster.local:%d" .Release.Namespace (.Values.metrics.port | default 8080 | int)) }}"
        {{- end }}

internal/config/config.go:72

• The comment says this URL is exposed to store containers as K8S_OPERATOR_URL, but the code injects SHOPWARE_OPERATOR_URL (api/v1/store_env.go). This is misleading and should be aligned with the actual env var name.

	// OperatorServiceURL is exposed to store containers as K8S_OPERATOR_URL so
	// the Shopware consumer knows how to reach the operator service.
	OperatorServiceURL string `env:"OPERATOR_SERVICE_URL"`

helm/values.yaml:114

• This comment says K8S_OPERATOR_URL is injected into store containers, but the operator currently injects SHOPWARE_OPERATOR_URL (api/v1/store_env.go). Please align the docs/config naming to avoid confusion for chart users.

This issue also appears on line 117 of the same file.

  # When enabled, a Service named 'shopware-operator' is created, the metrics
  # endpoint is exposed, and K8S_OPERATOR_URL is injected into every store
  # container so the Shopware consumer can reach the operator.