Features Pricing Guides Alternatives Blog Docs My account Français Buy a license GitHub
Integrations / Lightweight Monitoring for Dokploy

Lightweight Monitoring for Dokploy

Dokploy runs your apps as Docker Swarm services behind Traefik. Maintenant watches the services, the tasks, the nodes, and the endpoints they serve — from one container.

· 2 min read

Dokploy is built on Docker Swarm with Traefik in front, which is a good architecture and an awkward one to monitor. Most lightweight tools list containers; on Swarm that is the wrong unit. A service with three replicas is not three unrelated containers, and a task that failed and was rescheduled is not a container that vanished.

Maintenant detects the runtime at boot and reports in Swarm’s own terms.

Install on the manager node

Deploy it through Dokploy as a Docker Compose service, on the manager:

services:
  maintenant:
    image: ghcr.io/kolapsis/maintenant:latest
    ports:
      - "8080:8080"
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro
      - /proc:/host/proc:ro
      - maintenant-data:/data
    environment:
      MAINTENANT_ADDR: "0.0.0.0:8080"
      MAINTENANT_DB: "/data/maintenant.db"
    restart: unless-stopped
volumes:
  maintenant-data:

The manager placement matters: Swarm state is only readable from a manager socket. Give it a domain in Dokploy so Traefik terminates TLS, and put authentication in front — Maintenant deliberately ships none.

What it shows on a Dokploy fleet

  • Services, with desired versus running replicas, so a partially-scheduled service is visible as such;
  • Tasks, including the ones that failed and were rescheduled — the history Swarm keeps and most dashboards drop;
  • Nodes, their availability and the tasks placed on them;
  • Endpoints, checked from outside, so a service that is running but returning 502 through Traefik still alerts;
  • Certificates, auto-detected from those endpoints;
  • Cron and background jobs, through heartbeat URLs, with alerts on missed deadlines;
  • Image updates, by digest comparison against the registry.

The combination that matters on a PaaS is the second and the fourth: Swarm will happily reschedule a task forever while every replica returns errors. Watching the service alone tells you it is “running”.

Multi-node and licensing

Community covers the machine it runs on, which for a single-server Dokploy install is the whole thing. Once you add worker nodes, enroll them with a lightweight agent — up to 20 machines on Personal, unlimited on Pro. If the servers belong to a client rather than to you, Pro is the edition that grants commercial use; there is a page about exactly that.

For the mechanics of Swarm monitoring independent of Dokploy, read the Docker Swarm monitoring guide.

Ready to switch to Maintenant ?

One container, zero config. Full monitoring in 30 seconds.