Skip to main content
See deployment components for the stack, managed alternatives and the release-specific image inventory.
The one-click installer supports Debian 12/13 and Ubuntu hosts. Fresh Debian 12 and 13 arm64 Docker-in-Docker containers were verified through app startup and a clean apt update. For other distributions or derivatives, use the manual Docker setup.
Use this script to set up Formbricks on a Debian or Ubuntu server. It installs Docker when needed and configures Postgres, SpiceDB, Traefik, and an optional SSL certificate.
This setup uses Traefik as a reverse proxy, essential for directing incoming traffic to the correct container and enabling secure internet access to Formbricks. Traefik is chosen for its simplicity and automatic SSL management via Let’s Encrypt.
The bundled RustFS option in this one-click setup is designed for convenience on a single server. It is a good fit for small-scale or low-complexity self-hosted deployments, but it is not the ideal RustFS architecture for high-availability or larger production environments. For stricter production requirements, use external object storage or run a dedicated RustFS deployment separately from the Formbricks one-click stack.
The installer stores generated PostgreSQL, Hub, Cube, and AuthZed credentials in ./formbricks/.env and restricts the file to 0600. PostgreSQL has a raw password for the database plus a URL-encoded companion for connection strings. When bundled RustFS is enabled, its generated credentials are stored there too. Keep that file private, include it in your server backup plan, and avoid checking it into source control or copying it to shared locations.
For better RustFS performance and stability, prefer local SSD or NVMe-backed storage for the host volume behind rustfs-data. Avoid NFS and other network filesystems for bundled RustFS data, and use XFS on dedicated storage disks when you manage the host layout yourself.
For other operating systems or a more customized installation, please refer to the advanced installation guide with Docker.

Requirements

  • A Debian 12/13 or Ubuntu server with SSH access, curl, and a user with sudo privileges.
  • A custom domain with an A and/or AAAA record pointing to your server. Every A and AAAA record on the domain must point to this server: remove any that point elsewhere, such as an AAAA record left over from a previous host or added by default by your DNS provider.
  • Ports 80 and 443 are open in your VM’s Security Group and host firewall. Traefik serves HTTPS on port 443, which Let’s Encrypt also uses to validate the SSL certificate, and plain HTTP on port 80 (redirected to HTTPS by default). If your domain has an AAAA record, open both ports for IPv6 too.

Deployment

Run this command in your terminal:
The v6 one-click stack is based on the production Docker Compose file and includes the Formbricks Hub API, hub-worker, and Cube as part of the baseline (Cube configuration lives under formbricks/cube/). Ensure your generated formbricks/docker-compose.yml contains the worker service and a non-empty HUB_API_KEY, and that formbricks/.env contains CUBEJS_API_SECRET, AUTHZED_TOKEN, and AUTHZED_DATABASE_PASSWORD before treating the stack as ready. The installer generates these values without printing them and restricts .env to 0600. HUB_API_URL should normally stay at http://hub:8080. Bundled SpiceDB is installed for every new setup and does not require an Enterprise license check.
The v5 one-click Traefik setup also exposes Hub-backed FeedbackRecords through Formbricks at /api/v3/feedbackRecords and /v1/feedback-records. Traefik calls the internal Formbricks gateway auth endpoint first, then forwards allowed requests to Hub with the generated HUB_API_KEY. Browser callers should request short-lived gateway tokens from /api/v3/gateway/token with { "service": "feedbackRecords" }.

Script Prompts

During installation, the script asks for:
  • Permission to remove conflicting legacy packages if Docker is absent. The default keeps them in place.
  • Your app domain and HTTPS choice. For HTTPS, confirm DNS and ports 80/443, then provide a certificate email and choose whether to enable HSTS.
  • Optional SMTP and file-upload configuration.
If Docker is already installed, the script reuses it. Repository setup failures show the apt error and clean up the new Docker source. On Debian hosts affected by the old Ubuntu repository entry, the script disables that entry before updating apt and prints the path to its backup.

Update

To upgrade, start with Upgrade Formbricks, then follow the one-click runbook. Don’t run ./formbricks.sh update before you’ve pinned the target version: it pulls whatever the image tag points to, which is latest by default. To add SpiceDB to an existing installation for 6.0, see the 6.0 setup steps.

Limit request bodies on authentication routes

New installations cap request bodies on /api/auth/* at 1 MiB in Traefik. An oversized upload to a sign-in or OAuth endpoint is refused by Traefik, and a slow one is read in full by Traefik first, so it does not tie up Formbricks while it arrives. No legitimate request to these routes comes close to that size. The cap applies to requests that go through Traefik, which is every outside request once port 3000 is published on 127.0.0.1 only (see one-click update details). The update command does not add the cap to an existing installation, so add it yourself if you want it.
These labels need Traefik v3.6.15 or later. Before that version, the buffering middleware turns the empty redirects that sign-in and SSO callbacks answer with into errors (a 500 or a timeout), so nobody can sign in; Traefik v2.7 also lets paths such as /x/../api/auth/... skip the cap. Check the image: traefik: line in formbricks/docker-compose.yml and upgrade Traefik first (new installations use v3.6.25). From Traefik v2, follow the v2 to v3 migration guide.
Since v3.6.7, Traefik accepts encoded characters such as %2F in request paths, which would let a request reach /api/auth/* without passing through these routers. In formbricks/traefik.yaml, add this block under http: of both the web and the websecure entry point (create http: under web if it has none). New installations already have it.
In formbricks/docker-compose.yml, add these lines to the labels of the formbricks service, replacing your-domain.com with your domain:
Copy the tls.certresolver line from your existing formbricks router as well, with the router name changed to formbricks-auth. If your installation also has a formbricks_http router (HSTS disabled), add these lines too, or plain-HTTP requests to /api/auth/* stay uncapped:
Then apply the changes from the formbricks directory. Traefik only reads traefik.yaml when it starts, hence the restart:
A request to /api/auth/* with a body over 1 MiB now gets 413 Request Entity Too Large from Traefik.

IPv6 visitors

If your domain has an AAAA record, visitors who connect over IPv6 keep their own address only when the Compose network has IPv6. Without it, Docker hands their connections to Traefik from its bridge gateway (such as 172.18.0.1), so Formbricks sees every IPv6 visitor as that one address. They then share one rate limit, including the sign-in limit of 5 attempts per minute, and logs and response metadata record the gateway instead of the visitor. New installations and ./formbricks.sh update add this to formbricks/docker-compose.yml when the server has IPv6 and runs Docker Engine 27.0 or later:
The update keeps the previous file as docker-compose.yml.before-ipv6-network and recreates the network when it restarts the stack. With an older Docker Engine, the script leaves the network as it is and says so: before 27.0, Docker cannot create the network without an IPv6 subnet you choose. Upgrade Docker Engine, then run the update again. If you have set networks.default yourself, the script leaves it alone, so add enable_ipv6: true to it by hand. To keep IPv6 off, set enable_ipv6: false under networks.default in formbricks/docker-compose.yml; a later update leaves that setting alone. If a container that is not part of Formbricks is attached to the network, the update cannot recreate it: it puts the previous file back, says so, and finishes the update without IPv6. Detach that container and run the update again.
This relies on Docker routing IPv6 with ip6tables, which is on by default from Docker Engine 27.0. If your daemon runs with "ip6tables": false (or --ip6tables=false), IPv6 visitors still share one address, the network’s IPv6 gateway, even after this change.

Stop

To stop Formbricks, simply run the following command:
The script will automatically stop all the Formbricks related containers and brings the entire stack down.

Restart

To restart Formbricks, simply run the following command:
The script will automatically restart all the Formbricks related containers and brings the entire stack up with the previous configuration.

Cleanup RustFS init (optional)

During the one-click setup, a temporary rustfs-init service configures RustFS (bucket, policy, service user). It is idempotent and safe to leave in place; it will do nothing on subsequent starts once the configuration exists. If you prefer to remove the rustfs-init service and its references after a successful setup, run:
./formbricks.sh cleanup-minio-init is still available as a backward-compatible alias. This only removes the init job and its Compose references; it does not delete any data or affect your RustFS configuration.

Uninstall

To uninstall Formbricks, simply run the following command, but keep in mind that this will delete all your data!
The script will automatically stop all the Formbricks related containers, remove the Formbricks directory, and delete the Docker network.

Debug

Run these from the directory that contains your formbricks/ folder. To see every service, including the one-time setup jobs that already exited, run:
To read the logs of all containers, or of the services you name, run:
If the first startup stops, the installer names the step that failed and the services that exited with an error, along with the ./formbricks.sh logs <service> command for each. Fix the cause, then run ./formbricks.sh install again. It keeps the PostgreSQL password and AuthZed credentials from the first run.

Troubleshooting

If you encounter any issues, consider the following steps:
  • Inbound Rules: Make sure you have added inbound rules for ports 80 and 443 in your VM’s Security Group and host firewall: for IPv4 if your domain has an A record, and for IPv6 if it has an AAAA record. Some cloud firewalls, such as AWS Security Groups, need a separate IPv6 rule (source ::/0).
  • A/AAAA Records: Verify that your domain’s records point to this VM: A records to its IPv4 address, AAAA records to its IPv6 address. Let’s Encrypt connects over IPv6 first when an AAAA record exists, so an AAAA record pointing to another server can make certificate issuance fail even when the A record is correct. If your domain has an AAAA record, ports 80 and 443 must also be open for IPv6; with only an AAAA record, there is no IPv4 to fall back to.
  • Check Docker Instances: Run ./formbricks.sh status to check the status of every Formbricks container.
  • Check Formbricks Logs: Run ./formbricks.sh logs <service> to check the logs of one service, or ./formbricks.sh logs for the whole stack.
If you have any questions or require help, feel free to reach out to us on GitHub Discussions. 😃

Verify the installation

Run the deployment check in a dedicated workspace before collecting real responses.