Homelab monitoring, Part 2: Every VM, the NAS and the Wi-Fi

Add every Proxmox VM and container, your TrueNAS and your UniFi devices to the monitoring from Part 1.

intermediate1 hupdated 2026-10-13monitoringprometheusproxmoxtruenasunifizfsDownload script.sh

Before you start

The monitoring stack from Part 1.
Any of: a Proxmox host, a TrueNAS SCALE NAS (24.10 or newer), a UniFi network. Each section stands on its own; skip what you don't have.
Fill in Your values below and every command on this page updates to match.

Your values

Fill these in and every command below updates to match. Nothing is sent anywhere.

Monitoring container IDFrom Part 1Proxmox host addressThe IP of your Proxmox serverProxmox host nameSame name as in Part 1NAS addressThe IP of your TrueNASNAS nameHow it is labelled on dashboards and alertsNAS apps poolA pool that is always availableUniFi controllerUniFi OS Server :11443, or https://<gateway IP>

Run this at your own risk. These steps worked on my hardware; yours may differ. Take a backup first, read each command before running it, and see the disclaimer.

In Part 1 you built Prometheus and Grafana and connected the Proxmox host. Now you add the rest of the homelab:

Source	What you get	Exporter
Proxmox	CPU, RAM, disk and network of every VM and container	pve-exporter
TrueNAS	CPU, RAM, temperatures, and pool usage and health	node_exporter + a small script
UniFi	Access points, switches, connected clients	Unpoller

Each section stands on its own. Skip what you don't have.

A. Proxmox: all VMs and containers (pve-exporter)

The node_exporter from Part 1 only sees the host itself. pve-exporter asks the Proxmox API about every guest. It gets its own read-only user.

A1. Create a read-only user and token (Proxmox host shell)

bash

pveum user add prometheus@pve --comment "Monitoring (read-only)"
pveum acl modify / --users prometheus@pve --roles PVEAuditor
pveum user token add prometheus@pve exporter --privsep 0

The last command prints a table with a value like a1b2c3d4-…. Copy it now; Proxmox shows it only once.

Option	Why
Role `PVEAuditor`	May read everything, change nothing
`--privsep 0`	The token gets the same rights as the user. Without it the token has no rights and the exporter sees nothing.

A2. Save the token (monitoring container)

bash

pct enter 140        # from the Proxmox host shell
nano /opt/monitoring/prometheus/pve.yml

Type this, with your token value, then save (Ctrl + O, Enter, Ctrl + X):

yaml

pve:
  user: prometheus@pve
  token_name: exporter
  token_value: a1b2c3d4-xxxx-xxxx-xxxx-xxxxxxxxxxxx
  verify_ssl: false

pve: is the name of this Proxmox server. The job below uses this name again. If you have two Proxmox hosts, add a second block (for example pve2:) with its own token.

A3. Add the exporter and the job

New services go into a second compose file, compose.override.yaml. Docker reads it automatically together with compose.yaml.

bash

cat > /opt/monitoring/compose.override.yaml <<'EOF'
services:
  pve-exporter:
    image: prompve/prometheus-pve-exporter:latest
    container_name: pve-exporter
    restart: unless-stopped
    volumes:
      - ./prometheus/pve.yml:/etc/prometheus/pve.yml:ro
EOF

cat >> /opt/monitoring/prometheus/prometheus.yml <<'EOF'

  - job_name: pve
    metrics_path: /pve
    params:
      cluster: ['1']
      node: ['1']
    static_configs:
      - targets: ['192.168.1.2']
        labels:
          module: pve
    relabel_configs:
      - source_labels: [__address__]
        target_label: __param_target
      - source_labels: [module]
        target_label: __param_module
      - source_labels: [__param_target]
        target_label: instance
      - target_label: __address__
        replacement: pve-exporter:9221
EOF

cd /opt/monitoring && docker compose up -d
curl -s -X POST localhost:9090/-/reload && echo "reload OK"

module: must match the name in pve.yml. With a second host, add another - targets: entry with module: pve2.

B. TrueNAS: system and pool health

On TrueNAS, node_exporter runs as an app. It does not know about pool usage and health, so a small script writes them into a file, and node_exporter publishes that file.

B1. Folder and script (TrueNAS: System → Shell)

bash

mkdir -p /mnt/tank/apps/node-exporter/textfile
chown -R 568:568 /mnt/tank/apps/node-exporter
chmod 755 /mnt/tank/apps/node-exporter /mnt/tank/apps/node-exporter/textfile

cat > /mnt/tank/apps/node-exporter/zpool-metrics.sh <<'EOF'
#!/bin/bash
# ZFS pool size, usage and health for node_exporter's textfile collector
DIR=/mnt/tank/apps/node-exporter/textfile
TMP=$(mktemp "$DIR/.zpool.XXXXXX")
zpool list -Hp -o name,size,alloc,free,cap,health | while read -r name size alloc free cap health; do
  echo "zpool_size_bytes{pool=\"$name\"} $size"
  echo "zpool_allocated_bytes{pool=\"$name\"} $alloc"
  echo "zpool_free_bytes{pool=\"$name\"} $free"
  echo "zpool_capacity_percent{pool=\"$name\"} ${cap%\%}"
  echo "zpool_healthy{pool=\"$name\",state=\"$health\"} $([ "$health" = ONLINE ] && echo 1 || echo 0)"
done | sort > "$TMP"
chmod 644 "$TMP" && mv "$TMP" "$DIR/zpool.prom"
EOF
bash /mnt/tank/apps/node-exporter/zpool-metrics.sh && cat /mnt/tank/apps/node-exporter/textfile/zpool.prom

The last line shows one block of lines per pool, including zpool_healthy{…,state="ONLINE"} 1.

Put the folder on a pool that's always available (your apps pool), not on a USB disk.

B2. Run the script every 5 minutes

System → Advanced Settings → Cron Jobs → Add:

Field	Value
Description	`ZFS pool metrics`
Command	`bash /mnt/tank/apps/node-exporter/zpool-metrics.sh`
Run As User	`root`
Schedule	Custom: Minutes `/5`, all other fields ``
Hide Standard Output / Error	✅ ✅

B3. Install node_exporter as an app

Apps → Discover Apps → ⋮ (top right) → Install via YAML. Name: node-exporter. Paste:

yaml

services:
  node-exporter:
    image: prom/node-exporter:latest
    restart: unless-stopped
    network_mode: host
    pid: host
    user: "568:568"
    command:
      - --path.rootfs=/host
      - --collector.textfile.directory=/textfile
    volumes:
      - /:/host:ro,rslave
      - /mnt/tank/apps/node-exporter/textfile:/textfile:ro

Line	Why
`network_mode: host`	Publishes on the NAS's own IP, port 9100
`pid: host`, `/:/host`	Lets the exporter see the real system, not just its container
`user: "568:568"`	The TrueNAS apps user. The default user, `nobody`, cannot read the textfile folder.

Installing node_exporter via YAML in TrueNAS

Check from the monitoring container:

bash

curl -s http://192.168.1.10:9100/metrics | grep '^zpool_'

B4. Add the NAS to Prometheus

The NAS becomes a second target of the node job from Part 1. Open the file:

bash

nano /opt/monitoring/prometheus/prometheus.yml

Under job_name: node, add the NAS below the existing Proxmox target, with the same indentation:

yaml

      - targets: ['192.168.1.10:9100']
        labels:
          host: nas

Save, then reload:

bash

curl -s -X POST localhost:9090/-/reload && echo "reload OK"

C. UniFi: Wi-Fi and switches (Unpoller)

C1. A read-only account in UniFi

In the UniFi Network app: Settings → Admins & Users → Create New:

Field	Value
Name / username	`unpoller`
Restrict to local access only	✅ (no Ubiquiti cloud account)
Role	View Only for the Network app
Password	a long random one

C2. Password, exporter and job (monitoring container)

Add the password to the .env file:

bash

nano /opt/monitoring/.env

Add a second line, save:

expected output

UNPOLLER_PASSWORD=the-password-you-chose

Then add to the override file and the job list. The controller address comes from Your values. For UniFi OS Server, use port 11443. For a UniFi gateway (UDM/UCG), use https:// and its IP.

bash

cat >> /opt/monitoring/compose.override.yaml <<'EOF'

  unpoller:
    image: ghcr.io/unpoller/unpoller:latest
    container_name: unpoller
    restart: unless-stopped
    environment:
      - UP_UNIFI_DEFAULT_URL=https://192.168.1.15:11443
      - UP_UNIFI_DEFAULT_USER=unpoller
      - UP_UNIFI_DEFAULT_PASS=${UNPOLLER_PASSWORD}
      - UP_UNIFI_DEFAULT_VERIFY_SSL=false
      - UP_UNIFI_DEFAULT_SAVE_SITES=true
      - UP_UNIFI_DEFAULT_SAVE_DPI=false
      - UP_INFLUXDB_DISABLE=true
      - UP_PROMETHEUS_DISABLE=false
      - UP_PROMETHEUS_NAMESPACE=unpoller
EOF

cat >> /opt/monitoring/prometheus/prometheus.yml <<'EOF'

  - job_name: unpoller
    static_configs:
      - targets: ['unpoller:9130']
EOF

cd /opt/monitoring && docker compose up -d
curl -s -X POST localhost:9090/-/reload && echo "reload OK"
sleep 40
docker compose logs unpoller --since 1m | tail -5

The log should mention your site and devices, and no 401 or login failed.

Check: everything UP?

bash

curl -s 'localhost:9090/api/v1/targets?state=active' |
  jq -r '.data.activeTargets[] | "\(.health)\t\(.labels.job)\t\(.labels.host // .labels.instance)"' | sort -k2 | column -t

Expected: one line per target, all up:

expected output

up  node        nas
up  node        pve
up  prometheus  localhost:9090
up  pve         192.168.1.2
up  unpoller    unpoller:9130

Beginner trap: in the Prometheus web UI you can type queries like up{job="pve"}. With curl on the command line, the { } confuses curl. Use curl -G … --data-urlencode 'query=up{job="pve"}' as in this series, or the api/v1/targets call above.

In Part 3 these numbers become dashboards.

Troubleshooting

Problem	Fix
`pve` UP, but no VM data	Token created without `--privsep 0`. Delete it (`pveum user token remove prometheus@pve exporter`) and create it again as in A1.
`pve` DOWN, log says 401	Wrong token in `pve.yml`, or the module name doesn't match the job's `module:` label.
No `zpool_` lines from the NAS	Run the script by hand (B1) and check `ls -l /mnt/tank/apps/node-exporter/textfile`. The file must be readable by everyone (`-rw-r--r--`).
Unpoller: login failed	Wrong password, or the account isn't local. Check C1.
Reload prints an error	A YAML indentation mistake in `prometheus.yml`. Every `- job_name` needs exactly 2 spaces in front.

Homelab monitoring, Part 2: Every VM, the NAS and the Wi-Fi

Before you start

Your values

A. Proxmox: all VMs and containers (pve-exporter)

A1. Create a read-only user and token (Proxmox host shell)

A2. Save the token (monitoring container)

A3. Add the exporter and the job

B. TrueNAS: system and pool health

B1. Folder and script (TrueNAS: System → Shell)

B2. Run the script every 5 minutes

B3. Install node_exporter as an app

B4. Add the NAS to Prometheus

C. UniFi: Wi-Fi and switches (Unpoller)

C1. A read-only account in UniFi

C2. Password, exporter and job (monitoring container)

Check: everything UP?

Troubleshooting

Did this guide work for you?