Fleet Management, Kubernetes Monitoring and the OpenTelemetry engine: Alloy at scale

1. Fleet Management and remotecfg #

Remote configuration.

The principle #

Updating 300 Alloys one by one doesn't scale. With Fleet Management (Grafana Cloud, or self-hosted for enterprise offers), you create pipelines matched to attributes; each Alloy polls the service with the remotecfg block and applies the pipelines matching its attributes. The inventory shows each collector's health, version and attributes.

config.alloy #

remotecfg {
  url            = sys.env("GCLOUD_FLEET_URL")
  id             = constants.hostname
  attributes     = {"env" = "production", "team" = "payments"}
  poll_frequency = "1m"

  basic_auth {
    username = sys.env("GCLOUD_FLEET_USER")
    password = sys.env("GCLOUD_RW_API_KEY")
  }
}

Configuration validated with alloy validate (Alloy v1.21.3, November 2026).

• id: collector ID; without it, Alloy generates a UUID kept in alloy_seed.json.
• attributes: used to select pipelines; the collector. prefix is reserved (collector.os and collector.version are added automatically).
• poll_frequency: 1 min by default, 10 s minimum.
• The alloy-remote-config protocol is open (Apache 2.0): you can implement your own server.

Keep a minimal local configuration (the remotecfg block and self-monitoring): if the service is unreachable, Alloy keeps the last configuration received.

2. Kubernetes Monitoring #

The k8s-monitoring Helm chart.

Several specialized Alloys #

The k8s-monitoring Helm chart (version 4 and later) deploys, via the Alloy Operator, several Alloys with distinct roles. In v4, you name the collectors yourself (collectors) and apply presets; each feature (clusterMetrics, podLogs, applicationObservability, autoInstrumentation…) is attached to a collector, and destinations are typed by protocol (prometheus, loki, otlp, pyroscope).

helm repo add grafana https://grafana.github.io/helm-charts
helm upgrade --install k8s-monitoring grafana/k8s-monitoring \
  --namespace monitoring --create-namespace -f values.yaml

The autoInstrumentation option also deploys Beyla as a DaemonSet. The values.yaml format changed between major versions: start from the Grafana Cloud wizard or the chart repository examples.

3. The OpenTelemetry engine #

A Collector configuration in Alloy.

alloy otel #

Already have OpenTelemetry Collector YAML configurations? Alloy's OTel engine (public preview) runs them as-is, without translating them to Alloy syntax. With alloy otel-supervisor, a built-in OpAMP supervisor lets Fleet Management drive that configuration remotely.

alloy otel --config otelcol.yaml
alloy convert --source-format=otelcol --output=config.alloy otelcol.yaml   # or convert to Alloy syntax

Sources #

• Grafana Alloy : remotecfg
• Grafana Fleet Management
• Kubernetes Monitoring Helm chart
• Alloy OTel Engine