CapRover

What CapRover actually is

CapRover is a free, self-hosted platform-as-a-service: you install it on a server you own, and it handles building, deploying, routing, HTTPS, and day-two management of containerized apps behind a web dashboard. The stack underneath is deliberately familiar — Docker, Docker Swarm for clustering, nginx for routing, and Let's Encrypt for certificates — wrapped in a control plane so that you rarely touch those pieces directly. Deploy from the dashboard, the caprover deploy CLI, a Git webhook, or an existing CI workflow; the result is the same app URL with a certificate attached.

The project began in 2017 as CaptainDuckDuck, written by Kasra Bigdeli after he worked out what he was paying Heroku versus what a VPS costs (his 2017 freeCodeCamp write-up puts the gap at roughly 4x), and rebranded to CapRover around 2019. The GitHub organization lists its location as BigZee Ventures LLC, with the license file copyright held by that entity; the code is Apache 2.0. Today it has around 15,000 GitHub stars and a companion repository of community-maintained one-click app definitions — WordPress, MySQL, MongoDB, monitoring tools, and dozens more deployable as ready-made services.

The app-level controls cover what real deployments need without a YAML tour: environment variables, persistent directories for state, port mappings, per-app log views, health checks, and one-click HTTP-to-HTTPS redirects. Captain Claw, the CLI, mirrors the dashboard for scripted use, and webhooks from GitHub or any Git host trigger rebuilds on push. There is a live public demo (password published on the homepage), which we appreciate — being able to click through a vendor's actual dashboard before installing anything is rarer than it should be. A database with persistent storage, correct environment variables, and HTTPS in front of it is minutes of work on a raw Docker host and roughly one click here. For a solo developer or small team shipping side projects or client sites, that is the whole pitch: Heroku-style ergonomics without the Heroku bill, running on infrastructure you can move.

Pricing: what you really pay

CapRover the software is free forever — that is on the front page of caprover.com, alongside "no vendor lock-in" and "production ready," and the Apache 2.0 license backs it. There is no paid tier of the core, no feature gating, and no license key. Funding is voluntary: the project points companies at OpenCollective to fund ongoing development and maintenance.

Your real costs sit elsewhere. First, the server: the project's own getting-started guide says a machine with a public IP runs "as low as $5 a month," and a wildcard DNS record "as low as $2 a year" — both from pages we read on caprover.com, and both are third-party prices that will vary by provider. The DigitalOcean one-click install is the recommended path, and the docs mention $100 of referral credit for new accounts. Add the domain, and that is your infrastructure bill. We would budget a little more than the floor: builds are memory-hungry, and the docs warn that 512 MB of RAM may not be enough for the build process.

Second, your time. CapRover removes a lot of Docker busywork, but it is still a server you patch, back up, and monitor. Third, there is CapRover Pro (pro.caprover.com), a hosted management layer for running multiple CapRover instances; the login page is live but does not publish pricing, so we cannot quote it — if fleet management matters to you, contact the project.

In our view the price-to-capability ratio is the strongest thing about CapRover. A comparable commercial PaaS bills per app, per seat, or per request; here the marginal cost of the tenth app is zero dollars and some RAM.

Performance & infrastructure

CapRover runs where you put it. Single node out of the box, Docker Swarm when you need more than one; published images cover AMD64 and ARM64 (the ARMv7 image is no longer built), and the tested stack is Ubuntu 24.04 with Docker 25 or newer. nginx terminates traffic and routes by hostname, Let's Encrypt certificates are issued and renewed from the dashboard with one-click HTTP-to-HTTPS redirects, and apps get persistent directories, environment variable management, port control, and basic health checks.

Because the platform is your VPS, the performance envelope is whatever you buy. That is a feature, not a dodge: you pick the CPU, the disk, and the datacenter region, and you can move providers without rewriting anything. On our infrastructure benchmarks page we track which hosts give the best raw value per core and per gigabyte — a natural pairing with a PaaS whose per-app cost is zero. Scaling is horizontal at the app level (replica counts) and at the cluster level (Swarm nodes); there is no autoscaler wired to traffic, so plan capacity yourself.

Builds run inside Docker on your machine, which means build speed follows your server's CPU and memory. This is the most common cause of slow or failed first deploys on undersized instances, and it is worth sizing the box for build load, not just steady-state serving.

Where it falls short

The honest caveat is project velocity. CapRover is largely a solo-maintained project — a 2022 GitHub discussion about stale development drew the maintainer's confirmation that feature work had slowed and that he is the main developer. The project is not abandoned (releases continue and the site is current), but anyone betting a production platform on CapRover should go in with eyes open: bus factor is low, and the roadmap is one person's capacity. Contributing code or funding via OpenCollective is the practical mitigation.

Second, polish and edges. Compared with a commercial PaaS, you will meet rough corners: documentation has gaps, error messages assume Docker knowledge, and some flows (custom domains with wildcards, non-standard ports, Dockerfile quirks) need manual work. The community Slack and GitHub issues are the support system; there is no support contract and no SLA.

There is also the maintenance contract nobody signs: security updates are yours to apply, at the Docker daemon, the host OS, and the CapRover layer. The project ships releases, but pulling them, reading the changelog, and testing against your apps is operator work. Teams that treat a self-hosted PaaS as install-and-forget tend to discover this during their first vulnerable-package audit. CapRover is app deployment, not a full infrastructure platform: no built-in monitoring stack beyond basic stats, no managed backups for your volumes (that is your server provider's or your own cron's job), no multi-tenant isolation for untrusted users, and clustering via Swarm is functional rather than a Kubernetes substitute. If you need scheduled autoscaling, regional failover, or compliance certifications, look at managed clouds in our category listings.

Who should pick this

We pick CapRover for developers and small teams who want Heroku-style deploys on infrastructure they control. Strong fits: side projects and client sites where a $5-10 VPS beats a per-app PaaS bill; agencies standardizing WordPress, Ghost, or database deployments across a few servers; homelab and internal-tool setups where the live demo's simplicity matters more than enterprise features; and anyone with container-based apps who does not want to hand-write nginx configs and certbot cron jobs.

The nearest alternatives frame the choice well. Dokku is the other popular self-hosted PaaS and is lighter, more Unix-philosophy, and Git-push oriented; CapRover is friendlier, with a real dashboard and the one-click catalog. Heroku and Render give you less operational work and an SLA in exchange for per-app pricing that scales badly. Kubernetes is the answer to a different question entirely. Our pick for someone who wants a control plane and does not want to learn a PaaS's internals is CapRover; our pick for someone who lives in the terminal and wants the thinnest possible layer is Dokku. We would not pick it for a large engineering org needing Kubernetes-grade orchestration, for teams without anyone comfortable with Docker when things break, or for workloads requiring contractual uptime commitments. And if the idea of a mostly-solo-maintained dependency in your critical path makes you uneasy, either fund the project until that changes or choose a platform with a bigger bench.

Key takeaways

• Free, self-hosted PaaS (Apache 2.0, BigZee Ventures LLC) wrapping Docker, Swarm, nginx, and Let's Encrypt behind a dashboard.
• One-click app catalog (WordPress, MySQL, MongoDB, and more) is the fastest path from VPS to running service.
• Real costs are the server (from about $5/mo per the project's docs), a wildcard domain, and your own ops time — not the software.
• Deploy via dashboard, CLI, webhook, or Git; automatic HTTPS included.
• Development is largely solo-maintained and slower-moving than commercial alternatives — fund it or factor in the bus factor.
• Best fit: developers and small teams wanting PaaS ergonomics on their own VPS; poor fit: orgs needing Kubernetes, SLAs, or managed compliance.

Sources

• https://caprover.com/
• https://caprover.com/docs/get-started.html
• https://caprover.com/docs/one-click-apps
• https://github.com/caprover/caprover
• https://raw.githubusercontent.com/caprover/caprover/master/LICENSE
• https://github.com/caprover/caprover/discussions/1544
• https://www.freecodecamp.org/news/how-i-cut-my-heroku-cost-by-400-5b9d0220ce13/
• https://github.com/githubsaturn