Docker Compose compatibility in self hosted v4.0

  1. My goal: upgrade self hosted retool from 3.178 to the latest stable v4
  2. Issue: I don’t want to migrate to Kubernetes
  3. Retool version & hosting setup (Docker, K8s, cloud provider, etc.)

We’re trying to upgrade to the latest stable v4 and read this note in the upgrade FAQs:

Retool does not support Docker Compose for production deployments. The agent sandbox service requires resource isolation and pod-level scheduling that a single-VM stack cannot reliably provide.

If you are using Docker Compose in production, migrate to Kubernetes before upgrading.

We don’t need to use the agent sandbox, but want the security patches in stable v4. Do we really need to migrate to Kubernetes or is there a way to upgrade without it? This is an internal tool, I wouldn’t classify it as “non-production” necessarily but I’m interested in understanding that distinction more if it serves our goal.

Summary

On a self-hosted Retool 3.178 Docker Compose deployment, there's a desire to upgrade to the latest stable v4 for security patches without migrating to Kubernetes, prompting questions about whether Kubernetes is truly required and what the production vs. non-production distinction means given the FAQ warning about Docker Compose and the agent sandbox service.

AI Response

Retool does publish a supported path to upgrade a Docker Compose 3.x deployment to 4.0 without Kubernetes, but it is explicitly documented for non-production deployments only; for production, Retool directs you to migrate to Kubernetes via the Terraform blueprints. The officially supported configurations are Helm on Kubernetes and the Docker Compose blueprints — a Docker Compose production upgrade is "provided for convenience" with no support guarantees. Note that v4.0 still adds the new services (js-executor, agent-sandbox-controller, agent-sandbox-proxy, r2-agent-worker, mcp) even if you don't use agents, so the agent sandbox runs via a Postgres-backed store rather than requiring Redis/Kubernetes, and you'll need a larger VM (plan for ~8 vCPUs and 64 GiB) plus blob storage. If you don't want certain services (e.g., MCP), you can remove those blocks from the Compose file. The "production" distinction is about support guarantees and reliability rather than whether a tool is internal — treat a deployment your organization depends on day-to-day as production when weighing this.

Sources

:bookmark: Upgrade from Retool 3.x to 4.0 (Docker Compose) | Retool Docs
Official step-by-step guide for upgrading a 3.x Docker Compose deployment to 4.0 without Kubernetes, explicitly scoped to non-production use and detailing the new services, VM sizing, and blob storage requirements.
:bookmark: Scale your self-hosted deployment infrastructure | Retool Docs
States that Retool requires production self-hosted deployments to run on Kubernetes with Helm, clarifying the production vs. non-production support distinction.

The Community Team is testing out a new automation. Let us know if it's helpful (or not) by leaving a :heart:, :+1:, or :-1:. Or by marking this post as the "Solution"! Let us know if you have any feedback here. :rocket: