nomploy

The self-hosted PaaS that grows past one server.
Heroku-style deploys from a UI, running on a real HashiCorp Nomad cluster: HA, autoscaling and load balancing included.

Quick start · Docs · Features · vs. Dokploy

curl -sSL https://nomploy.com/install.sh | sh

One command on a fresh Debian/Ubuntu VPS gives you a working panel. Start on a single €5 box, then add nodes from the UI when you need them.

Why nomploy?

Tools like Dokploy and Coolify make self-hosting easy, until you need a second server. Kubernetes solves that, but it's a lot to run for a handful of apps. nomploy sits in between: the same push-to-deploy experience, on Nomad, a scheduler that's simple enough for one person to run.

• Start small, grow to a cluster. Go from one node to an HA Nomad + Consul control plane from the Cluster tab, over an encrypted WireGuard mesh.
• Autoscaling at every level. Apps scale on CPU/memory, and autoscaling groups create and remove cloud VMs (Hetzner) based on cluster pressure.
• No single point of failure at the edge. An active/active Traefik pool with health-checked DNS and shared certificates.
• Bring what you have. Deploy Docker Compose files (translated to Nomad), native Nomad jobs, or Nomad Packs.
• Fully open source. AGPL-3.0, with no source-available "enterprise" modules and no paid feature gates.

	Single-server PaaS (Dokploy, Coolify)	nomploy	Kubernetes
Deploy from a UI, git push, SSL	✅	✅	add-ons
Multi-node HA cluster	limited	✅	✅
Node (VM) autoscaling	—	✅	✅
HA ingress / load balancer	—	✅	✅
Easy for one person to run	✅	✅	—

What is it?

nomploy is a free, self-hostable Platform as a Service (PaaS) that deploys and manages your applications and databases on a HashiCorp Nomad cluster.

It is a fork of Dokploy that swaps the orchestrator from Docker Swarm to Nomad, while keeping Dokploy's UI, git integration, domains/SSL, backups, monitoring and notifications. nomploy as a whole is licensed under the GNU AGPL-3.0; the upstream enterprise (source-available) modules are not included — see Attribution & License.

✨ Features

• Nomad orchestration — deploys run as Nomad jobs; the compose files you already know are translated to Nomad HCL (ports, env, health checks, resources, replicas and autoscaling via x-nomad-scaling).
• Three ways to deploy — translated Docker Compose, a native Nomad HCL job file deployed verbatim, or a Nomad Pack from a custom registry or the hosted pack registry at packs.nomploy.com.
• Highly-available clusters — grow from one node to an HA Nomad + Consul control plane from the Cluster tab: add/remove servers and workers over an encrypted WireGuard mesh, drain nodes for maintenance, upgrade Nomad (including the control plane) from the UI, see each node's private IP, and watch cluster DNS health.
• Cluster autoscaling with node-pool groups — ASG-style autoscaling groups, each backed by its own Nomad node pool and launch template (cloud provider, server type, region, image, SSH key, min/max, CPU/memory reservation thresholds and cooldown). Each group scales independently on its pool's pressure; target a group from a job's node_pool. Supports reactive scaling, a manual desired count and scheduled actions (cron), and it provisions, joins, drains and destroys cloud VMs automatically — without removing a node whose pool still has running work.
• Cloud providers — register a cloud credential once (e.g. Hetzner) in Settings → Cloud and reference it from any autoscaling group or the one-click Add node action; tokens never live on individual groups.
• App & service autoscaling — horizontal autoscaling on CPU/memory targets for applications and compose services.
• Runtime secrets & config files — per application or compose, store secret environment variables and whole config files in Nomad Variables; they are injected as env or mounted as files at runtime and never appear in the job spec.
• Bring-your-own registry — register any OCI registry (your own zot/registry:2, GHCR, ECR, …) in Settings → Registry; credentials are distributed to every node via Consul KV + consul-template, so private images pull cluster-wide with no credentials in job specs.
• GPU workloads — request NVIDIA GPUs for a job via Nomad device plugins.
• Nomad dashboard — view jobs, allocations, nodes, logs and cluster resources; scale or stop jobs and exec into an allocation's task from the UI. Pick which server's Nomad cluster to view with a per-server selector.
• One-click Nomad bootstrap — install Docker + Consul + Nomad + CNI on a managed server over SSH, straight from the UI.
• Applications & databases — Node.js, PHP, Python, Go, Ruby, …; MySQL, PostgreSQL (incl. pgvector), MongoDB, MariaDB, libSQL and Redis, persisted on real Nomad-managed Docker volumes.
• Ingress via Traefik + Consul Catalog — services register in Consul with Traefik tags and are routed automatically, with Let's Encrypt TLS.
• High-availability ingress ("Load Balancer") — run an active/active Traefik pool on every node tagged nomploy_lb (the hub is excluded) from the Load Balancer tab. The pool serves every route from the Consul catalog and shares TLS certs through Consul KV (auto-resynced from the hub as they renew). It gets a generated DNS hostname whose A records are health-pruned to the healthy nodes' public IPs (Cloudflare, AWS-ALB style — CNAME your app domains to it), and the tab has time-range throughput/latency graphs (2xx/4xx/5xx, Prometheus), consolidated searchable logs tagged by instance, and per-cert expiry.
• DNS providers (Cloudflare) — register a DNS credential in Settings → DNS Providers; used for ACME DNS-01 (as an additional resolver, HTTP-01 stays the default) and to manage the Load Balancer's records.
• Live metrics & scaling suggestions — telemetry-based CPU/memory per project, environment and service, shown as used-vs-reserved with sparklines, plus utilization/right-sizing suggestions and an optional daily digest.
• Backups & restore — scheduled backups of managed databases (PostgreSQL, MySQL, MariaDB, MongoDB, libSQL) and the panel's own database to any S3-compatible store (AWS S3, Cloudflare R2, …), with per-backup last-run health and one-command restore.
• Docker Compose, templates, multi-server, real-time monitoring, notifications (Slack/Discord/Telegram/email), and a tRPC API.
• Self-hosted — runs on your own VPS.

🚀 Getting Started

On a fresh Linux VPS (Debian/Ubuntu or RHEL family), run:

curl -sSL https://nomploy.com/install.sh | sh

This installs Docker, Consul, Nomad, the CNI plugins, Traefik, Postgres, Redis and the nomploy app, then prints the URL to open.

The app image is published to ghcr.io/nomploy/nomploy. If the container package is private, either make it public in its GitHub package settings or run docker login ghcr.io on the server before installing. Override the image with NOMPLOY_IMAGE=… if needed.

To add Nomad to an existing managed server instead, use the Bootstrap Nomad button in that server's Nomad settings inside the dashboard.

📚 Documentation

Full guides live in docs/:

• Getting Started — install, first login, first deploy.
• Architecture — Nomad, Consul, Traefik and the WireGuard overlay.
• Cluster management — high availability: add/remove nodes, drain, DNS health.
• Deploying — Compose, native Nomad HCL, and Nomad Pack (custom registries or the hosted registry at packs.nomploy.com).
• Autoscaling — scale apps and services on CPU/memory, and the cluster itself with node-pool autoscaling groups.
• Load Balancer — HA Traefik pool, DNS health-prune, shared certs, and the metrics/logs tabs.
• Container registry — bring-your-own registry + cluster-wide auth.
• GPU workloads — requesting NVIDIA GPUs.

🧭 How nomploy differs from Dokploy

	Dokploy	nomploy
Orchestrator	Docker Swarm	HashiCorp Nomad
Service discovery / ingress	Traefik (Docker provider)	Traefik + Consul Catalog
Deploy artifact	Swarm stack / compose	Nomad HCL job (from compose), native HCL, or Nomad Pack
Cluster / node autoscaling	—	ASG-style autoscaling groups on Nomad node pools
Ingress HA	Single Traefik	Active/active Traefik pool + DNS health-prune, shared certs via Consul KV
Secrets & config files	Plaintext env in the stack	Nomad Variables, kept out of the job spec
Pack registry	—	Hosted at packs.nomploy.com, plus any custom registry
Enterprise modules (SSO, audit, custom roles, white-label)	Source-available add-on	Removed; free-tier equivalents

Swarm → Nomad migration status

The Docker Swarm backend has been removed from every runtime path:

• Deploys, reloads, stop/start, scaling and rollbacks for applications, databases and compose services run as Nomad jobs (nomad job run / scale / stop -purge).
• Logs, exec/terminal, volume & database backups, restores, scheduled tasks and container monitoring resolve a service's container through its Nomad allocation (the com.hashicorp.nomad.alloc_id label), not Swarm services.
• The panel updates itself — the UI's Reload/Update re-submits nomploy's own Nomad job (the Nomad equivalent of Swarm's docker service update --image), and update available is detected from the GHCR image digest.
• Traefik routes via the Consul Catalog provider on host networking, and the dashboard is toggled through api.insecure in traefik.yml.
• Adding a server provisions a Nomad worker over the WireGuard mesh, not a Swarm node; server setup/validation no longer touch docker swarm or the overlay network.

Cleanup of Swarm-era remnants is largely done: the unused *Swarm DB columns have been dropped, and GPU scheduling now uses Nomad device plugins rather than node labels. The stack compose type remains in the schema enum for backward compatibility but is unused and unreachable.

🤝 Contributing

See the Contributing Guide.

📝 Attribution & License

nomploy is a fork of Dokploy — Copyright © Dokploy Technology, Inc. Original project: https://github.com/dokploy/dokploy

nomploy as a whole is licensed under the GNU AGPL-3.0 (see LICENSE). Portions derived from Dokploy remain under Apache-2.0 (see LICENSE-APACHE); that grant is preserved and Dokploy's notices are retained. Apache-2.0 permits redistributing a modified work under a compatible copyleft license such as the AGPL — see LICENSING.md for how the two fit together.

Upstream Dokploy was dual-licensed: most code under Apache-2.0, plus enterprise modules under the separate Dokploy Source Available License (DSAL) in /proprietary directories. nomploy does not ship any DSAL-licensed code — those modules were removed and replaced with original implementations. Full details and attribution are in the NOTICE file.

Contributions are accepted under the Contributor License Agreement, which keeps a commercial-licensing option open alongside the AGPL.