Monitoring Synology Containers with cAdvisor and Grafana

Something on my NAS was eating memory overnight, and DSM's Docker window only shows live values while it's open. Useless for a 3 a.m. mystery. I needed history. The usual stack is cAdvisor for per-container metrics, Prometheus to store them, and Grafana to draw them. All three run as containers on the NAS and fit in one docker-compose.yml.

You need DSM 6.2 with Docker, SSH as an administrator, and about 300 MB of spare RAM; Prometheus is the hungriest of the three.

Folders first

I keep everything under the shared docker folder so Hyper Backup picks it up:

sudo -i
mkdir -p /volume1/docker/monitoring/{prometheus,grafana}
chown -R 472:472 /volume1/docker/monitoring/grafana   # grafana runs as uid 472
chown -R 65534:65534 /volume1/docker/monitoring/prometheus   # nobody

Prometheus config

cat > /volume1/docker/monitoring/prometheus/prometheus.yml <<'EOF'
global:
  scrape_interval: 15s
scrape_configs:
  - job_name: cadvisor
    static_configs:
      - targets: ['cadvisor:8080']
EOF

The target name cadvisor works because Compose puts the services on one network with DNS by service name.

The Compose file

cat > /volume1/docker/monitoring/docker-compose.yml <<'EOF'
version: "3"
services:
  cadvisor:
    image: gcr.io/cadvisor/cadvisor:v0.36.0
    container_name: cadvisor
    privileged: true
    volumes:
      - /:/rootfs:ro
      - /var/run:/var/run:ro
      - /sys:/sys:ro
      - /volume1/@docker:/var/lib/docker:ro
    restart: unless-stopped

  prometheus:
    image: prom/prometheus:v2.20.1
    container_name: prometheus
    volumes:
      - ./prometheus:/etc/prometheus
      - ./prometheus/data:/prometheus
    command:
      - --config.file=/etc/prometheus/prometheus.yml
      - --storage.tsdb.retention.time=30d
    restart: unless-stopped

  grafana:
    image: grafana/grafana:7.1.5
    container_name: grafana
    ports:
      - "3000:3000"
    volumes:
      - ./grafana:/var/lib/grafana
    environment:
      - GF_SECURITY_ADMIN_PASSWORD=change-me
    restart: unless-stopped
EOF

Two Synology-specific details: the Docker root on DSM is /volume1/@docker, not /var/lib/docker, so it is mounted at the path cAdvisor expects; and cAdvisor needs privileged: true to read cgroup data on the DSM kernel.

Start it

cd /volume1/docker/monitoring
docker-compose up -d
docker-compose ps

All three should show Up. Only Grafana publishes a port; Prometheus and cAdvisor stay internal.

Grafana data source and dashboard

Open http://nas-ip:3000, log in as admin with the password from the Compose file, then:

1. Configuration → Data Sources → Add → Prometheus, URL http://prometheus:9090, Save & Test.
2. Create → Import → enter dashboard ID 193 (the community "Docker monitoring" dashboard), pick the Prometheus data source, Import.

You now have per-container CPU, memory, network and disk I/O with 30 days of history.

Is data flowing?

In Prometheus terms, the query below should list every running container:

container_last_seen{name!=""}

Paste it into Grafana's Explore view. If it returns rows, the pipeline works end to end.

Gotchas

• cAdvisor shows only itself. The /var/lib/docker mount is wrong. Confirm with docker info | grep "Docker Root Dir"; on DSM it prints /volume1/@docker.
• permission denied on /prometheus. The data folder must be writable by uid 65534. Re-run the chown from Step 1 after the first start creates the directory.
• Grafana loses dashboards after docker-compose down. The ./grafana bind mount is missing or owned by root. Check ls -ln /volume1/docker/monitoring/grafana shows uid 472.
• NAS feels slower. cAdvisor's default --housekeeping_interval is 1 s. Add command: ["--housekeeping_interval=10s", "--docker_only=true"] to the cadvisor service to cut its own CPU use.

Next I'd add node_exporter for host-level metrics of the NAS itself, pointed at the same Prometheus. Same Compose pattern.