Self-hosting Coolify: an open-source PaaS on your own VPS

From the outside, a small web app is just a few pages and a database. Keeping it running reliably means handling domains, HTTPS, reverse proxies, builds, environment variables, databases, logs, and backups. Docker Compose handles one app well. With three to five services, an ad hoc approach starts getting messy: every app has its own configuration, deployment process, and place to check logs.

Vercel, Heroku, and Netlify address that pain: push code to Git and you get a new deployment, a domain, HTTPS, and pull-request previews. The trade-off is usage-based pricing and data living outside your server. Coolify keeps the “git push to deploy” experience while running on your own VPS.

A clear caveat: I checked the commands and features in this article against the official documentation on October 27, 2026. I have not tested the installation below on a real VPS, so check the latest documentation before running the commands.

What is Coolify?

Coolify is an open-source, self-hostable PaaS for deploying applications, databases, and services to your own servers. It replaces Heroku, Netlify, and Vercel in one important respect: instead of renting infrastructure from a PaaS provider, you run Coolify on your VPS, and it manages servers, applications, and databases over SSH.

The project uses the Apache-2.0 license. As of October 27, 2026, its GitHub repository had approximately 64,987 stars and 5,833 forks. The latest release was v4.5.0, published on October 23, 2026, and the latest code push was on October 26, 2026. Those numbers indicate active maintenance, not proof of quality. Check the latest release before installing.

Coolify takes code from GitHub, GitLab, Bitbucket, and Gitea, and builds it with Nixpacks, Railpack, Dockerfiles, Docker Compose, or prebuilt images. Pushes can trigger automatic deployments, pull requests get separate previews, and a failed deployment can be rolled back to a retained image. There are also more than 360 one-click services, with persistent storage and automatically generated credentials.

Who should use it, and who does not need it?

In my view, Coolify is best suited to small teams, freelancers, homelabs, and businesses that want control over their data at a fixed VPS cost. When you manage several websites, internal tools, and separate staging and production environments, a dashboard for deployments, logs, and scheduled backups saves a lot of repetitive SSH work.

Three situations do not call for Coolify: you need Kubernetes-style multi-region autoscaling and automatic failover; you run just one application, where plain Docker Compose is lighter; or you do not want responsibility for server updates, backups, and monitoring, in which case a managed platform still makes more sense.

Architecture and the deployment process

The most important point: Coolify is not a new runtime. It is a control plane that turns SSH and Docker into a PaaS experience. Understanding this helps you evaluate both the benefits and the risks.

The control plane consists of the dashboard, API, CLI, and MCP: the place to declare servers, applications, databases, domains, environment variables, backup schedules, and team permissions. A destination server receives deployments over SSH and runs workloads using Docker. The two layers can share a machine, but should not do so for important systems, for reasons covered in the installation section.

A typical deployment starts with a code push, pull request, or webhook. Coolify receives the event and builds an image on the destination server using your chosen builder. It then assigns a domain, configures the reverse proxy, provisions HTTPS automatically, and runs health checks. Containers run on a private network. Build and runtime logs, a container terminal, and resource monitoring are available in the dashboard.

For pull requests, Coolify creates previews to inspect changes. If a new deployment fails, you can roll back to a retained image. Note that rollback restores the application; it does not automatically restore overwritten database data without a separate backup.

Coolify runs on Linux and depends on Docker on the destination server, reliable SSH, a domain pointing to the server, and open ports 80/443 for HTTPS provisioning. It is not a multi-node orchestrator like Kubernetes and does not promise autoscaling or automatic failover. In other words, it manages multiple servers rather than turning them into a self-balancing cluster. You add destination servers when you need more places to run workloads, but load distribution and failover still require separate design.

Installing Coolify on a VPS

Prerequisites

The official documentation requires at least 2 CPUs, 2 GB of RAM, 10 GB of free disk space, and amd64 or arm64. Smaller machines, such as 1 CPU with 512 MB of RAM, may run it but are not recommended. I would not economize that far if builds run on the same server, because builds can easily exhaust memory.

Listed operating systems include Debian/Ubuntu, the RHEL family (CentOS, Fedora, RHEL, AlmaLinux, Rocky), SUSE, Arch, Alpine, and 64-bit Raspberry Pi OS. More important than the specification is starting with a fresh server that is not running other services. Existing services occupying ports 80/443, or a preconfigured Docker installation, are common sources of conflicts. Before installing, take a VPS snapshot, point your domain to the VPS IP, and prepare root or sudo access.

Installing with the official command

curl -fsSL https://cdn.coollabs.io/coolify/install.sh | bash

Before piping an internet script into bash, download and read it. The installer has permission to change your system; you should know what it will do. Run it only on a fresh server and keep the installer’s final output, which tells you the next step.

Verification and first login

docker ps

Expected result: Coolify-related containers appear in the list and are running. Then open the dashboard at the address printed by the installer—usually the domain you pointed at the VPS, or the IP and port the installer provides. On the first visit, create the initial administrator account.

If the dashboard is unreachable, check in this order: firewall and open ports, whether Docker is running, and the installer logs. Most problems lie in one of those three places, rather than a broken Coolify installation.

Deploying your first application

There are four steps: add a destination server; connect the Git repository and select a builder (Nixpacks/Railpack for typical apps, Dockerfile/Docker Compose if your project already has them, or a prebuilt image); set the domain, enable automatic HTTPS, and enter environment variables and secrets; then deploy and watch the build logs, runtime logs, and health checks.

If possible, separate the Coolify server from workload servers from the beginning. If both share a VPS, a failure takes down both your management dashboard and your running applications. The dashboard’s rollback button cannot help when the host itself is down. Treat the Docker host—not the dashboard—as the first layer to monitor, back up, and restore.

Common problems

An existing server can cause port 80/443 conflicts or conflicts with Docker: use a fresh server or free the ports first. An incorrect domain IP prevents HTTPS provisioning: check DNS before blaming Coolify. An undersized server causes slow builds or memory exhaustion: meet the recommended minimum. Builds fail because environment variables are missing or a Dockerfile needs adjustment: read the build logs rather than blindly redeploying.

Evaluation: benefits, trade-offs, and security

The clearest benefit is a PaaS experience on your own infrastructure: git-push deployments, pull-request previews, rollback, backups, scheduled tasks, notifications, an API, CLI, and MCP. Costs are easier to predict by VPS rather than usage, while data and configuration stay within infrastructure you control. More than 360 one-click services help you launch databases, queues, and internal tools without rewriting Compose files.

The biggest trade-off is owning operations. Coolify, Docker, OS, certificate, backup, and monitoring maintenance all fall to the operator. Coolify does not replace Kubernetes for multi-cluster systems, autoscaling, or complex failover. Deep Docker or network failures still require Docker debugging skills; clicking around a dashboard is not enough.

This article gives no performance or cost benchmark against Vercel or Heroku: the documentation I checked did not publish one, and I have not measured it. The only planning baseline is the minimum of 2 CPUs, 2 GB of RAM, 10 GB of disk, and the recommendation to separate Coolify from workload servers.

For security, protect the dashboard first: strong administrator credentials, multi-factor authentication if your installed version supports it, management IP restrictions where possible, and team-based permissions rather than a shared root account. Manage secrets through Coolify environment variables; do not commit them to the repository. Closely control destination-server SSH keys and revoke them when team members leave.

Updates are part of security too. Schedule Coolify, Docker, and OS updates, and always back up Coolify configuration and workload data before major upgrades. More importantly, test restoring those backups before a real incident. An untested backup is not yet a backup you can rely on.

Comparison with alternatives in the same space

The table below compares application deployment options using decisions an operator actually has to make.

The short version: choose Coolify for PaaS convenience while keeping your own server; Compose for just 1–2 apps; managed PaaS when you do not want to operate servers; and Kubernetes when you genuinely need a multi-node cluster.

Where does Coolify fit in practice?

For small businesses and internal IT teams, Coolify brings websites, internal CRMs, operations tools, and separate staging and production environments into one dashboard, with data under the business’s control. Separate destination servers also reduce the risk of accidentally deploying a test build to production.

For a homelab, one dashboard for a blog, media server, databases, and self-hosted tools makes experimentation faster. Pull-request previews and one-click services let you try things without cluttering your main server, and removal is tidier than manually cleaning up containers.

For small-team DevOps, Coolify standardizes git-push deployments, pull-request previews, quick rollback, centralized logs, and staging/production separation. It is enough standardization without building a complicated GitOps pipeline.

For AI prototypes and full-stack applications, Coolify can deploy quickly from a Git repository and attach databases and one-click services for a demo. One caveat: GPU-heavy AI workloads are outside the scope of a typical Coolify VPS and require separate hardware and driver evaluation.

For system administration, Coolify manages multiple servers from one dashboard, monitors resources, and schedules backups and recurring tasks, instead of requiring separate SSH sessions for routine commands. Its value is consolidating repetitive operations with clear logs and schedules.

Conclusion

Self-host Coolify if you are a small team, freelancer, homelab operator, or business wanting PaaS on your own VPS, already know Docker, and accept responsibility for operations: updates, backups, monitoring, and SSH security are yours. Avoid it if you run just one simple static app, require multi-region autoscaling and failover, or do not want server responsibility.

Next, snapshot your VPS, install on a fresh server using the latest documentation, deploy a test app and a one-click service, and check HTTPS, logs, backups, and rollback before adding production workloads. Do not put production on it before successfully testing a backup restore at least once. For the alternatives on either side, see deploying Astro to Cloudflare Pages for managed hosting and self-hosting Stoat with Docker Compose (Vietnamese) for plain Compose. Coolify sits between them; choose it only when you understand the trade-off.

References

• Coolify on GitHub: https://github.com/coollabsio/coolify
• Coolify installation documentation: https://coolify.io/docs/installation
• Coolify homepage: https://coolify.io