victoriametrics

Catalog date: 2026-06-24.

VictoriaMetrics — Prometheus-compatible time-series database used as the long-term storage layer. Queryable directly or through Grafana.

Note: this page's last_reviewed date predates a 2026-07-15 alphaswarm_platform change (feat(platform): publish VictoriaMetrics observability plane) that made kube-prometheus-stack a legacy, unselected rollback asset. Per the current observability/README.md, "VictoriaMetrics is the only metrics authority: vmagent owns discovery, scraping, relabeling, buffering, and remote write... and the raw OpenTelemetry gateway sends normalized OTLP metrics through vmagent" — Prometheus / kube-prometheus-stack is no longer in the selected path.

Identity

Field	Value
Service id	victoriametrics
Role	observability
Image	victoriametrics/victoria-metrics:v1.108.0 (compose) / victoriametrics/victoria-metrics:v1.147.0 (Kustomize Deployment, pinned by digest) — the two surfaces are on different versions
Port	8428 (HTTP — write + query)
Health	/health

Deployment surfaces

Surface	Where
Compose	service victoriametrics in alphaswarm_platform/compose/docker-compose.platform.yml — single-node, self-scrapes via a mounted promscrape.config (no remote-write in this surface)
Kustomize	observability/victoriametrics/ — single-node Deployment fed by vmagent's remote-write; no vmstorage/vmselect/vminsert cluster-mode split exists in this repo today

Dependencies

Upstream: vmagent (remote-write) in the Kustomize surface; the compose surface scrapes directly via its own promscrape.config instead of receiving remote-write. Neither surface is fed by Prometheus / kube-prometheus-stack, which is now an unselected legacy asset.

Downstream: Grafana (datasource).

Operations

• Retention: --retentionPeriod=90d in the Kustomize Deployment; --retentionPeriod=14d in the compose service. No audit_evidence-based 13/24-month retention override was found in alphaswarm_platform.
• Cardinality control: vmagent's metric_relabel_configs apply labeldrop rules per scrape job to bound cardinality.
• PromQL compatibility: queries that work in Prometheus work here; some MetricsQL extensions (histogram_quantiles, etc.) are used in dashboards.

See also

• prometheus.md — legacy/unselected scrape path; see the note above.
• grafana.md — primary query path.