Prometheus
Buoy talks to any Prometheus-compatible HTTP API (Prometheus, Thanos, Mimir, Cortex, VictoriaMetrics) to power PromQL queries, charts, dashboards, and per-object Metrics tabs. Unlike the CRD-detected integrations, Prometheus needs one bit of setup: a source URL. Once a source is configured, every metrics surface lights up.
Setup
Open Settings → Prometheus → Add source and fill in:
| Field | Notes |
|---|---|
| Name | A label for the source (e.g. mimir-prod). |
| URL | Base URL of the Prometheus API (e.g. https://prom.example.com). Buoy appends /api/v1/.... |
| Headers | Optional. Bearer tokens (Authorization: Bearer …) or tenancy headers (X-Scope-OrgID: tenant-a). |
| Cluster Label | The label that distinguishes clusters in this Prometheus’ series (e.g. cluster, k8s_cluster). Leave blank for a single-cluster Prometheus. |
| Context → Cluster Value | Shown once a cluster label is set: override how each kubeconfig context name maps to that label’s value. Unmapped contexts use the context name verbatim. |
| TLS insecure | Skip certificate verification (self-signed endpoints). |
| Timeout | Per-request timeout in milliseconds (default 15000). |
Hit Test to verify reachability — it calls /api/v1/status/buildinfo and reports latency or the error. Mark one source as default with the star; the default is used unless a query targets another.
You can add multiple sources (e.g. per environment or per tenant) and switch between them.
Cluster-aware queries
When a source has a cluster label, Buoy automatically scopes queries to the cluster matching your active kube context, so the same dashboard follows you across clusters without editing the PromQL. The label value comes from the Context → Cluster mapping, falling back to the context name. Leave the cluster label blank for a single-cluster Prometheus and queries run unfiltered.
Where metrics show up
Once at least one source is configured:
/prom <expr>(alias/metrics <expr>) — opens the PromQL workspace, optionally seeded with an expression to run. Full editor with autocomplete (metric names, label names, label values, functions), a query builder, time-range picker, and charts. The command bar also offers inline PromQL autocomplete while you type/prom …./dashboards— open the dashboards manager to build and save multi-panel dashboards.- Metrics tab — appears on detail views for kinds that have one (e.g. Pods, Nodes, workloads), backed by queries scoped to that object.
- Bulk dashboard — select rows in a list and open a dashboard scoped to the selection from the Actions menu.
If no source is configured, these entrypoints tell you so (e.g. /prom reports “No Prometheus source configured. Add one in Settings → Prometheus.”) rather than failing silently.