VictoriaMetrics¶

Connect HolmesGPT to VictoriaMetrics, a Prometheus-compatible time-series database, for metric queries during investigations.

When to Use This¶

✅ You run VictoriaMetrics as a Prometheus replacement or long-term store
✅ You want lighter resource usage than a full kube-prometheus-stack
✅ You have vmsingle or vmcluster running in your cluster

Prerequisites¶

A running VictoriaMetrics instance (vmsingle, vmselect, or vmauth) reachable from where Holmes runs
The HTTP API endpoint (typically port 8428 for vmsingle, 8481 for vmselect)

Configuration¶

HolmesGPT uses its built-in Prometheus toolset to query VictoriaMetrics — VM's HTTP API implements the Prometheus query API, so no separate toolset is required. Set subtype: victoriametrics so the UI shows this as a VictoriaMetrics integration (not a generic Prometheus one):

Holmes CLI Holmes Helm Chart Robusta Helm Chart

Add the following to ~/.holmes/config.yaml. Create the file if it doesn't exist:

toolsets:
  prometheus/metrics:
    enabled: true
    subtype: victoriametrics
    config:
      prometheus_url: "http://vmsingle-vmsingle.monitoring.svc.cluster.local:8428"

When using the standalone Holmes Helm Chart, update your values.yaml:

toolsets:
  prometheus/metrics:
    enabled: true
    subtype: victoriametrics
    config:
      prometheus_url: "http://vmsingle-vmsingle.monitoring.svc.cluster.local:8428"

Apply the configuration:

helm upgrade holmes holmes/holmes --values=values.yaml

When using the Robusta Helm Chart (which includes HolmesGPT), update your generated_values.yaml:

holmes:
  toolsets:
    prometheus/metrics:
      enabled: true
      subtype: victoriametrics
      config:
        prometheus_url: "http://vmsingle-vmsingle.monitoring.svc.cluster.local:8428"

Apply the configuration:

helm upgrade robusta robusta/robusta --values=generated_values.yaml --set clusterName=<YOUR_CLUSTER_NAME>

The subtype field is optional — omitting it still works, but the frontend catalog will display the integration under the generic "Prometheus" card rather than the dedicated "VictoriaMetrics" one.

For the complete list of supported configuration fields (authentication headers, timeouts, label filters, SSL verification, etc.), see the Prometheus toolset configuration reference.

Compatibility Notes¶

VictoriaMetrics implements the Prometheus query API, but a few less-common endpoints that Holmes uses have partial or no support:

Holmes feature	VM support
PromQL instant and range queries	Full
Label discovery (`/api/v1/labels`, `/api/v1/label/<name>/values`)	Full
Series queries (`/api/v1/series`)	Full
Metric metadata (`/api/v1/metadata`)	Partial — fewer descriptions than Prometheus
Alert / recording rules (`/api/v1/rules`)	Requires `vmalert`; absent in `vmsingle`

For day-to-day metric investigations, Holmes works the same as with Prometheus. Advanced metric discovery (via the metadata API) and alert-rule listing (via vmalert) may be reduced or unavailable.

Capabilities¶

Inherits every tool from the Prometheus toolset. See the Prometheus capabilities reference for the full list of tools Holmes can use against a VictoriaMetrics endpoint.