Grafana Alloy

OpenTelemetry Collector distribution with programmable pipelines. This app runs the HassOS scenario from alloy-resources, vendored into the image so it runs self-contained. The Alloy UI is at http://<homeassistant_ip>:12345.

What it collects

The scenario is split into a base (the metric/log sinks + shared log pipeline) plus toggleable input pipelines. run.sh assembles the active config from the enabled inputs.

Toggle	Default	Input
enable_unix	true	node_exporter (unix) metrics.
enable_process	true	process_exporter metrics.
enable_docker	true	cAdvisor metrics + Docker container logs.
enable_journal	true	systemd journal logs.
scrape_exporter	false	Scrape the Home Assistant Exporter add-on at :9878.
scrape_core	false	Scrape HA core /api/prometheus (no auth, job homeassistant).

Options

Each option maps to the environment variable the scenario reads.

Key	Env var	Default	Notes
metrics_url	METRICS_PRIMARY_URL	—	Prometheus remote_write URL.
metrics_tenant	METRICS_PRIMARY_ORG	—	Sent as X-Scope-OrgID (multi-tenant Mimir).
logs_url	LOGS_PRIMARY_URL	—	Loki push URL.
logs_tenant	LOGS_PRIMARY_ORG	—	Loki tenant_id.
cluster_name	CLUSTER_NAME	HomeAssistant	cluster external label on every series.
env	ENV	—	env external label.
region	REGION	—	region external label.
instance	INSTANCE	—	Override the instance label on all metrics (host name).

Protection mode

Disable Protection mode so the Docker (cAdvisor + log) components can reach the Docker socket (docker_api: true). The rest of the pipeline runs regardless.

Examples

Local Mimir + Loki (multi-tenant)

metrics_url: "http://mimir.lan:9009/api/v1/push"
metrics_tenant: "home"
logs_url: "http://loki.lan:3100/loki/api/v1/push"
logs_tenant: "home"
cluster_name: "house"
env: "prod"
region: "home-lab"

Metrics only

metrics_url: "http://prometheus.lan:9090/api/v1/write"
logs_url: ""        # leave empty to skip the logs pipeline target
cluster_name: "house"