Monitoring

The monitoring stack provides metrics collection, visualization, a service dashboard, and external uptime monitoring. It runs as Docker containers on a dedicated VM in each environment.

Deployment order

Monitoring deploys after Networking, Step-CA, and NTP. It is the last infrastructure service to deploy before applications.

Architecture

graph TD
    subgraph Monitoring VM
        Prometheus[Prometheus]
        Grafana[Grafana]
        Homepage[Homepage]
    end

    subgraph Targets
        NE1[Node Exporter\nmonitoring VM]
        NE2[Node Exporter\nmedia-stack]
        NE3[Node Exporter\nnetworking]
        NE4[Node Exporter\nwebsite]
    end

    subgraph Monitoring VM
        UptimeKuma[Uptime Kuma]
    end

    NE1 & NE2 & NE3 & NE4 -->|metrics| Prometheus
    Prometheus -->|datasource| Grafana
    UptimeKuma -->|health checks| Prometheus

Prometheus scrapes node exporters across all VMs for system metrics
Grafana visualizes metrics from Prometheus with dashboards
Homepage provides a service dashboard with status widgets and quick links
Uptime Kuma provides uptime monitoring for internal services

Components

Component	Image	Port	Purpose
Prometheus	`prom/prometheus`	9090	Metrics collection and storage
Grafana	`grafana/grafana`	3000	Metrics visualization
Homepage	`ghcr.io/gethomepage/homepage`	3002	Service dashboard
Uptime Kuma	`louislam/uptime-kuma`	3001	Uptime monitoring

Hosts

Environment	VM	IP
WIL	Monitoring	`10.2.20.30`
LDN	Monitoring	`10.3.20.30`
External	VPS	`178.156.190.134`

File Locations

Monitoring Stack

File	Purpose
`playbooks/infrastructure/monitoring/deploy.yml`	Main playbook
`playbooks/infrastructure/monitoring/tasks/monitoring-stack.yml`	Deployment task
`playbooks/infrastructure/monitoring/templates/compose.yaml.j2`	Docker Compose definition
`playbooks/infrastructure/monitoring/templates/prometheus.yml.j2`	Prometheus scrape config
`playbooks/infrastructure/monitoring/templates/grafana-datasources.yml.j2`	Grafana datasource provisioning
`playbooks/infrastructure/monitoring/templates/services.yaml.j2`	Homepage services dashboard
`playbooks/infrastructure/monitoring/templates/bookmarks.yaml.j2`	Homepage bookmarks
`playbooks/infrastructure/monitoring/templates/settings.yaml.j2`	Homepage theme and layout
`playbooks/infrastructure/monitoring/templates/widgets.yaml.j2`	Homepage widgets
`playbooks/infrastructure/monitoring/handlers/main.yml`	Container lifecycle handlers
`environments/<env>/group_vars/infra_monitoring/`	Per-environment variables

Deployment

task ansible:deploy-monitoring ENV=wil

Monitoring Stack Deployment

The task file:

Creates directory structure under /opt/monitoring/
Sets ownership to monitoring_uid:monitoring_gid
Deploys Prometheus configuration
Deploys Grafana datasource provisioning
Deploys Homepage configuration files (services, bookmarks, settings, widgets)
Deploys Docker Compose file
Starts all containers

Prometheus

Prometheus scrapes metrics from node exporters running on infrastructure and application VMs.

Scrape Configuration

The prometheus.yml.j2 template generates the scrape config:

global:
  scrape_interval: 15s
  evaluation_interval: 15s

scrape_configs:
  - job_name: 'prometheus'
    static_configs:
      - targets: ['localhost:9090']

  - job_name: 'node-exporter'
    static_configs:
      - targets:
          - 'localhost:9100'
          - '10.2.0.5:9100'
          # ... all node targets

Three additional jobs — chrony, gpsd, and node — scrape the PTP grandmaster's exporters (deployed by the ptp-experiments repo's monitoring role). Each renders only when its per-environment target list is non-empty; currently only LDN defines them. The node job is kept separate from node-exporter because the Chrony dashboard's CPU-temperature panel selects job="node" unscoped.

Container Configuration

Memory: 2GB limit, 1GB reservation
Retention: configurable via prometheus_retention
Storage: persistent volume at /prometheus
User: runs as monitoring_uid:monitoring_gid

Source: ansible/playbooks/infrastructure/monitoring/templates/prometheus.yml.j2

Configuration Reference

Parameter	Type	Description	Default
`prometheus_image`	`string`	Docker image for Prometheus	(per-env)
`prometheus_retention`	`string`	Metrics data retention period	`"30d"`
`prometheus_node_targets`	`list[string]`	Node exporter endpoints to scrape (`host:port`)	(per-env)
`prometheus_chrony_targets`	`list[string]`	chrony_exporter endpoints (PTP grandmaster, `:9123`)	`[]`
`prometheus_gpsd_targets`	`list[string]`	gpsd exporter endpoints (PTP grandmaster, `:9015`)	`[]`
`prometheus_ptp_node_targets`	`list[string]`	PTP grandmaster node_exporter endpoints (job `node`, `:9100`)	`[]`
`node_exporter_image`	`string`	Docker image for Node Exporter sidecar	(per-env)

Sources: ansible/environments/<env>/group_vars/infra_monitoring/prometheus.yml · ansible/environments/<env>/group_vars/infra_monitoring/containers.yml

Grafana

Grafana connects to Prometheus as its default datasource and provides metric visualization dashboards.

Datasource Provisioning

Grafana auto-provisions a Prometheus datasource on startup via grafana-datasources.yml.j2:

apiVersion: 1
datasources:
  - name: Prometheus
    uid: prometheus
    type: prometheus
    access: proxy
    url: http://prometheus:9090
    isDefault: true
    editable: true

The datasource carries the fixed uid prometheus because the provisioned dashboards bind to it.

Dashboard Provisioning

Dashboards are provisioned from files — any JSON dropped into playbooks/infrastructure/monitoring/files/dashboards/ is deployed and picked up automatically (the file provider rescans every 30 s; UI edits to provisioned dashboards do not persist):

Dashboard	File	Data
Chrony Detail	`chrony_grafana_dashboard.json`	`chrony` + `node` jobs (PTP grandmaster)
GPSD	`gpsd_grafana_dashboard.json`	`gpsd` job (PTP grandmaster)
Node Exporter Full	`node_exporter_full.json`	All `node-exporter`/`node` targets (grafana.com 1860)

Container Configuration

Memory: 2GB limit, 1GB reservation
Depends on: Prometheus
Provisioning: mounted read-only at /etc/grafana/provisioning
Sign-up: disabled (GF_USERS_ALLOW_SIGN_UP=false)

Configuration Reference

Parameter	Type	Description	Default
`grafana_image`	`string`	Docker image for Grafana	(per-env)
`grafana_admin_user`	`string`	Admin username for the web UI	`"admin"`
`grafana_admin_password`	`string`	Admin password (SOPS-encrypted)	(required)
`grafana_url`	`string`	Root URL for links in notifications and dashboards	(per-env)

Sources: ansible/environments/<env>/group_vars/infra_monitoring/grafana.yml · ansible/playbooks/infrastructure/monitoring/templates/grafana-datasources.yml.j2

Homepage

Homepage provides a service dashboard with status widgets, service health indicators, and quick links to all homelab services.

Container Configuration

Memory: 512MB limit, 256MB reservation
Docker socket: mounted read-only for container status widgets
Environment: receives API keys for service widgets (Sonarr, Radarr, Bazarr, Prowlarr, SABnzbd, Plex)

Configuration Reference

Parameter	Type	Description	Default
`homepage_image`	`string`	Docker image for Homepage	(per-env)
`homepage_allowed_hosts`	`string`	Comma-separated hostnames Homepage responds to	(per-env)

Dashboard Configuration

Homepage is configured via four YAML template files:

Template	Purpose
`services.yaml.j2`	Service cards with status widgets (media, monitoring, node exporters)
`bookmarks.yaml.j2`	Quick links (TrueNAS, Proxmox)
`settings.yaml.j2`	Theme (dark, slate), layout (row style, column counts)
`widgets.yaml.j2`	Global widgets (search bar, datetime)

Sources: ansible/environments/<env>/group_vars/infra_monitoring/containers.yml · ansible/playbooks/infrastructure/monitoring/templates/services.yaml.j2

Shared Configuration

These variables apply to all monitoring containers:

Parameter	Type	Description	Default
`monitoring_uid`	`string`	Container file ownership UID	`"1000"`
`monitoring_gid`	`string`	Container file ownership GID	`"1000"`
`backup_targets`	`list[string]`	Service directories under `/opt/monitoring/` to back up	(per-env)

Common Tasks

Add a new Prometheus scrape target

Edit ansible/environments/<env>/group_vars/infra_monitoring/prometheus.yml:

prometheus_node_targets:
  # ... existing targets
  - "10.2.20.60:9100"   # new VM

Deploy:
```
task ansible:deploy-monitoring ENV=wil
```

Change metrics retention

Edit ansible/environments/<env>/group_vars/infra_monitoring/prometheus.yml:
```
prometheus_retention: "90d"
```
Deploy:
```
task ansible:deploy-monitoring ENV=wil
```

Edit ansible/playbooks/infrastructure/monitoring/templates/services.yaml.j2
Add a new service entry under the appropriate section
If the service requires an API key, add the key to secrets.sops.yml and pass it through compose.yaml.j2
Deploy:
```
task ansible:deploy-monitoring ENV=wil
```