Skip to main content

Before you start

This guide is for teams that already run a Docker Swarm cluster. If you have a single Linux server, the standard installation is the recommended path — Swarm adds management overhead without a benefit there. You need a working Swarm (at least one manager node) and the release package unpacked on a manager:
Collabase runs as a single application instance on Swarm, and the database is pinned to one designated node so it always finds its data.

Install

1

Label the data node

Choose the node that will hold the database and uploads, and label it:
2

Create the secrets

Swarm stores passwords as managed secrets — they never appear in configuration files. Create them once:
3

Create the proxy configuration

Generate the reverse-proxy configuration with the installer, then register it as a Swarm config:
4

Deploy the stack

5

Check the services

All services should reach 1/1 replicas within a few minutes.

Updating

Pull the new images and redeploy — Swarm rolls the services with an automatic rollback if a service fails to start:

Removing

Your data volumes and secrets are kept. To delete data too, remove the volumes listed by docker volume ls | grep collabase — this cannot be undone.

Troubleshooting

A service stays at 0/1 replicas. Ask Swarm why:
The database service will not start on another node. That is intentional — it is pinned to the node you labeled in step 1 so it never loses its data. To move it, move the volume first, then relabel. I need help. Open a ticket at collabase.featurebase.app.
Last modified on July 17, 2026