Dokploy vs CapRover: The Veteran Swarm PaaS Against the 2026 Favourite, Feature by Feature

CapRover (born as CaptainDuckDuck in 2017) was the self-hosted PaaS before the category had a name: a web dashboard and CLI over Docker Swarm with Nginx in front, one-click apps, Let's Encrypt certificates and clustering. Dokploy arrived in 2024 with Traefik, Docker Compose as a first-class citizen, managed databases with backups, previews and a template catalogue, and now has more than twice CapRover's GitHub stars. This comparison uses the projects' own repositories and documentation on 31 October 2026 and ends with what it takes to move from one to the other on a MassiveGRID server from the Dokploy hosting page.

Quick verdict

Stay on CapRover if you run single-container applications and one-click apps, know its captain-definition format and Nginx template, and value a tool whose behaviour has not changed much in years. Choose Dokploy if your applications are Docker Compose stacks, you want databases with scheduled S3 backups, preview deployments, notifications, team permissions and a modern dashboard, or you are starting fresh in 2026. Both are free and open source, both cluster with Docker Swarm, and both run on the same 2 vCPU, 4 GB server.

What they share

A web dashboard and a CLI on one server, deployments from a Git repository or an image, automatic Let's Encrypt certificates per domain, one-click application catalogues, and Docker Swarm underneath, so both can add worker nodes and distribute services. Both install with one command on a fresh Ubuntu or Debian server, and both have an API. CapRover's documented minimum is 1 GB of RAM; Dokploy's installer asks for 2 GB and 30 GB of disk. In practice either wants the 4 GB Starter preset once applications and a database run beside it.

Deploy model and Docker Compose

The Compose row is the one that decides most migrations. If your stack is a docker-compose.yml with an app, a worker, a database and a cache, Dokploy deploys it as written; CapRover asks you to split it into apps and loses the dependency and networking semantics along the way. Our Dokploy vs Coolify post has the same observation about CapRover in the three-way context.

Proxy, TLS and clustering

CapRover fronts everything with Nginx and a customisable Nginx template per app, which experienced Nginx users like and newcomers find opaque. Dokploy uses Traefik, configured by the dashboard, with per-app middlewares (redirects, headers, rate limits) editable in the Advanced tab. Both obtain Let's Encrypt certificates automatically.

Both cluster with Docker Swarm. CapRover has done so since its early releases and handles image distribution with its own registry or an external one; Dokploy's Cluster section needs a container registry (Docker Hub, GitHub Container Registry, ECR) before it unlocks, and adds remote servers over SSH and a dedicated build server as alternatives to a full Swarm. For a single server neither needs any of this; for a build that disturbs production, Dokploy's build server is the simpler fix.

Databases, backups, teams and notifications

Rows from Dokploy's own comparison table and CapRover's documentation on 31 October 2026. The single-admin model is the quiet dealbreaker for teams: CapRover has one password for the whole server, Dokploy has accounts, roles and two-factor authentication.

Community and maintenance

GitHub on 31 October 2026: CapRover 15,200 stars, Dokploy 38,200. CapRover is the older project by seven years and its release cadence has slowed; that is stability to some users and a risk to others, because security fixes and new Docker behaviour take longer to land. Dokploy ships minor releases frequently (v0.30.x at the time of writing), has a paid Cloud and Enterprise business behind it, and a growing template repository. Both have Apache-2.0 cores; Dokploy's repository also carries a proprietary licence file for Enterprise components.

Where CapRover is better

• Maturity and predictability. Nine years of the same deployment model; the edge cases are documented in forum threads.
• Nginx. Teams that already tune Nginx get a template they understand instead of Traefik's configuration.
• Simplicity for single-container apps. A Dockerfile, a captain-definition, a domain, done.
• No registry required to cluster. CapRover ships with its own registry option for Swarm.
• Smaller minimum. CapRover's docs accept 1 GB of RAM, though a real workload wants more on either tool.

Where Dokploy is better

• Docker Compose as a first-class citizen, which is how most applications are described today.
• Databases with scheduled S3 backups, volume backups and an instance backup that restores to a new server.
• Teams: accounts, permissions, 2FA and passkeys, SSO on Enterprise.
• Preview deployments, notifications, scheduled jobs, a web terminal and more build types.
• Remote and build servers over SSH in addition to Swarm, and a hosted dashboard option (Dokploy Cloud) if you want one.
• Development pace and community size, which decide how quickly tomorrow's Docker or Node change is handled.

Migrating from CapRover to Dokploy

1. Inventory your apps in the CapRover dashboard: which have a Dockerfile, which use a captain-definition pointing at an image, which are one-click apps, and which have persistent directories.
2. Install Dokploy on a new server (or a second one; the Dokploy hosting page presets the configurator and MassiveGRID installs Dokploy on request). Do not install it on the CapRover host; both want ports 80 and 443.
3. Recreate applications: Dockerfile apps become Dokploy applications with the Dockerfile build type; image-based captain-definitions become applications deployed from the image; one-click apps become Dokploy templates or Compose stacks (every common one, PostgreSQL, Redis, MinIO, Plausible, n8n and more, is in the catalogue).
4. Move data: dump databases from the CapRover apps and restore into Dokploy database services; copy persistent directories into named volumes with rsync or a tarball.
5. Translate Nginx customisations into Traefik middlewares (redirects, headers, basic auth, rate limits) in each application's Advanced tab, and any custom Nginx locations into application routes.
6. Environment variables copy across one app at a time; webhooks are recreated by enabling Auto Deploy.
7. Switch DNS once Traefik has issued certificates, keep CapRover running a week, then decommission it.

Budget an afternoon for a handful of apps. The server underneath can be the same size: the Starter preset (2 vCPU, 4 GB, 64 GB NVMe) is $9.58 a month on shared cores or $19.80 on dedicated cores; Growth (4 vCPU, 8 GB) is $19.16 or $39.60. Our self-hosted PaaS hosting page puts both tools next to Coolify and Dokku with the server each one needs.

Frequently Asked Questions

Is CapRover still maintained in 2026?

Yes, but at a slower pace than Dokploy or Coolify: releases are less frequent and the feature set has been stable for years. For single-container apps that is fine; for teams who want Compose, previews, backups and user accounts, the newer platforms have moved ahead.

Does CapRover support Docker Compose?

Only in a limited way: a Compose file is translated into separate CapRover apps with restrictions (no build directives, limited options), and the networking and dependency semantics of Compose are lost. Dokploy deploys Compose files as written, in Compose or Swarm Stack mode.

Can I run CapRover and Dokploy on the same server?

Not comfortably: both want ports 80 and 443 for their proxies. Install Dokploy on a second server, migrate app by app, switch DNS, then retire the CapRover host. A MassiveGRID Starter preset at $9.58 a month covers the overlap period.