docker compose up on any VPS. You can also deploy to a
managed platform — but read this first, because Replayfy uses five datastores
and no platform manages all of them.
Render
The repo ships arender.yaml
blueprint that provisions the API + dashboard + managed Postgres + Redis. You
supply the three external datastore URLs (Mongo, ClickHouse, S3) as environment
variables after the blueprint loads.
1
Click the button
Render reads
render.yaml and previews the services + managed Postgres/Redis.2
Fill the secrets
Set the
sync: false vars: DATABASE_URL_MONGO (Atlas), CLICKHOUSE_URL,
the S3_* block (R2/S3), and — after the first deploy assigns your
*.onrender.com URLs — APP_BASE_URL, API_BASE_URL, CORS_ORIGINS, then
redeploy.Railway
Railway manages Postgres, Redis, and MongoDB (three of the four), so only ClickHouse and object storage are external. There’s no repo blueprint file — you build a project once (the two app services + the managed databases), then Generate Template from Project and publish it; the “Deploy on Railway” button points at that published template. Railway does not rundocker-compose.yml
directly; map each compose service to a Railway service.
Brimble
Brimble has no one-click deploy button. Deploy with its CLI and attach managed datastores from the dashboard:CLICKHOUSE_URL + the S3_* vars at external services. Brimble is a good
fit for the static dashboard + a Node API; ClickHouse, object storage, and full
orchestration are on you.
The reliable path
If a clean one-click matters more than a managed control plane, putdocker compose up on a single VPS — it runs all five datastores together, is
the path we test, and needs no external services. See
Self-hosting.