I'm Peter. I built Nomploy because I liked the Dokploy/Coolify experience (git push, domains, SSL, databases, backups from a UI), but I kept hitting a wall when I needed more than one server. Docker Swarm felt like a dead end and Kubernetes felt like too much for what I run. So I forked Dokploy and swapped the orchestrator for HashiCorp Nomad. Everything you deploy runs as a Nomad job. Your docker-compose files are translated to Nomad HCL, or you can deploy a native Nomad job or a Nomad Pack directly. What it does beyond a single-box PaaS: - Grows from one node to an HA Nomad + Consul control plane from the UI, over a WireGuard mesh - Autoscaling groups: provisions and removes Hetzner VMs on its own based on cluster pressure, with node pools, min/max limits and cron schedules - HA load balancing: a pool of Traefik nodes, multi-A DNS with health-based pruning, and certificates shared through Consul KV - App autoscaling on CPU/memory, canary/zero-downtime deploys, and secrets in Nomad Variables (never written into the job spec) - A self-hosted registry option (zot backed by S3/R2) so images built from source run on every node Install on a fresh Debian/Ubuntu VPS: curl -sSL https://nomploy.com/install.sh | sh On licensing: upstream Dokploy moved some features into a source-available licence. Nomploy doesn't include any of that code. The whole project is AGPL-3.0. I run it on my own Hetzner cluster in production. Rough edges I know about: per-allocation stats show as unavailable on some cgroup v2 setups, and Hetzner is the only cloud provider for autoscaling so far. I'd really like feedback, especially from people running Nomad already: what would stop you from using this?

Nomploy

Self-hosted PaaS on Nomad, Dokploy fork with real clustering

Self-hosted PaaS on Nomad, Dokploy fork with real clustering