Metrics Reference¶
The operator exposes Prometheus metrics on the manager binary's metrics endpoint (defaults to :8443 HTTPS or :8080 HTTP, see --metrics-bind-address in flags.md). All metric names are stable across patch releases; new labels may be added.
Histogram bucket boundaries are tuned to the operator's 30s HTTP timeout — anything in the 20-30s range indicates rate-limiter saturation or upstream slowness, both of which warrant paging.
Fleet API client¶
Latency, error rate, and rate-limiter saturation for outbound calls to the Fleet Management API. Use these to alert on Fleet API regressions.
| Metric | Kind | Labels | Buckets | Description | Source |
|---|---|---|---|---|---|
fleet_api_errors_total | CounterVec | operation, status | — | Total number of Fleet Management API errors. | pkg/fleetclient/metrics.go:49 |
fleet_api_rate_limiter_wait_duration_seconds | Histogram | — | 0.005, 0.025, 0.1, 0.5, 1, 2, 5, 10, 20, 30 | Time spent waiting for the Fleet Management API rate limiter. | pkg/fleetclient/metrics.go:60 |
fleet_api_request_duration_seconds | HistogramVec | operation, status | 0.005, 0.025, 0.1, 0.5, 1, 2, 5, 10, 20, 30 | Duration of Fleet Management API requests in seconds. | pkg/fleetclient/metrics.go:34 |
fleet_api_requests_total | CounterVec | operation | — | Total number of Fleet Management API requests. | pkg/fleetclient/metrics.go:42 |
Controller reconciliation¶
Per-CRD reconcile outcome counters and gauges that track owned attributes / discovered collectors. Use these to alert on reconciliation health.
| Metric | Kind | Labels | Buckets | Description | Source |
|---|---|---|---|---|---|
fleet_discovery_list_collectors_result_size | GaugeVec | namespace, name | — | Number of collectors returned by the last ListCollectors call. | internal/controller/metrics.go:82 |
fleet_external_sync_owned_keys | GaugeVec | namespace, name | — | Number of attribute keys currently owned by an ExternalAttributeSync CR. | internal/controller/metrics.go:72 |
fleet_pipeline_name_migrations_total | Counter | — | — | Total Fleet pipeline name-scope migrations (old pipeline deleted and recreated under the scoped name). | internal/controller/metrics.go:92 |
fleet_resource_sync_age_seconds | HistogramVec | kind | 0, 60, 300, 600, 1800, 3600, 7200 | Age in seconds since last successful sync, sampled at each reconcile. | internal/controller/metrics.go:61 |
fleet_resource_synced_total | CounterVec | kind, reason | — | Total reconciliation outcomes by resource kind and reason. | internal/controller/metrics.go:50 |