Monitoring Overview¶

Stack¶

The monitoring stack runs as Docker containers in the monitoring network on VPS #1.

Component	Version	Role	Port
Prometheus	v3.10	Metrics aggregation, alerting rules	9090
Grafana	v12.4	Dashboards, unified alerting	3000 (proxied via nginx)
Loki	v3.6	Log aggregation	3100
Alloy	v1.14	Log collector (Docker auto-discovery)	—
node-exporter	v1.10	Host system metrics (CPU, memory, disk, network)	9100
cAdvisor	v0.55	Container metrics (CPU, memory, I/O)	8080
postgres-exporter	v0.19	PostgreSQL metrics	9187
redis-exporter	v1.82	Redis metrics	9121
nginx-exporter	v1.4	Nginx reverse proxy metrics	9113
celery-exporter	v0.10	Celery task queue metrics	9808
blackbox-exporter	v0.26	External probes (SSL, HTTP health)	9115

Architecture¶

                              ┌──────────────┐
                              │   Grafana    │ ← dashboards + alerts
                              │  :3000       │
                              └──┬───────┬───┘
                                 │       │
                    ┌────────────┘       └────────────┐
                    ▼                                  ▼
             ┌──────────────┐                  ┌──────────────┐
             │  Prometheus  │                  │     Loki     │
             │  :9090       │                  │  :3100       │
             └──────┬───────┘                  └──────┬───────┘
                    │                                  │
        ┌───────────┼───────────┐                      │
        ▼           ▼           ▼                      ▼
   exporters   blackbox    cAdvisor              ┌──────────────┐
   (node,      (SSL +      (container            │    Alloy     │
    pg, redis,  HTTP        metrics)              │  (log shim)  │
    nginx,      probes)                           └──────────────┘
    celery)                                             │
                                                   Docker socket
                                                   (auto-discover)

Access¶

Grafana UI: https://monitoring.groupe-suffren.com
Prometheus UI: Internal only (port 9090, not exposed via nginx)

Configuration¶

All configs live in the Aether repo at monitoring/ and deploy to /opt/docker/monitoring/:

monitoring/
├── docker-compose.yml
├── prometheus/
│   ├── prometheus.yml          # scrape targets
│   └── alerts/                 # 6 rule files
├── loki/loki-config.yml
├── alloy/config.alloy
├── blackbox/blackbox.yml
└── grafana/provisioning/
    ├── datasources/
    ├── dashboards/json/        # 5 pre-built dashboards
    └── alerting/               # contact points, policies, rules

Resource Limits¶

Container	Image	Memory Limit
`monitoring_prometheus`	`prom/prometheus:v3.10.0`	384m
`monitoring_grafana`	`grafana/grafana:12.4`	384m
`monitoring_node_exporter`	`prom/node-exporter:v1.10.2`	64m
`monitoring_cadvisor`	`gcr.io/cadvisor/cadvisor:v0.55.1`	128m
`monitoring_postgres_exporter`	`prometheuscommunity/postgres-exporter:v0.19.1`	48m
`monitoring_redis_exporter`	`oliver006/redis_exporter:v1.82.0`	48m
`monitoring_blackbox_exporter`	`prom/blackbox-exporter:v0.26.0`	64m
`monitoring_loki`	`grafana/loki:3.6`	384m
`monitoring_alloy`	`grafana/alloy:v1.14.1`	384m
`monitoring_nginx_exporter`	`nginx/nginx-prometheus-exporter:1.4`	32m
`monitoring_celery_exporter`	`danihodovic/celery-exporter:0.10.10`	128m

Retention¶

Data	Retention	Source
Prometheus metrics	15d	`monitoring/docker-compose.yml`
Loki logs	168h (7 days)	`monitoring/loki/loki-config.yml`
Backup daily	7 days	`backups/scripts/backup.sh`
Backup weekly	28 days	`backups/scripts/backup.sh`
Backup archive (RPPS)	Indefinite	`backups/scripts/backup.sh`