Umami v3 Privacy-First Web Analytics

Deploys Umami v3.5.0, an open-source, cookie-free web analytics tool, with a Postgres database. It replaces Umami's well-known default login (admin/umami) with a generated password during boot, before the dashboard goes online. Suited to anyone who wants self-hosted analytics for one or several sites.

What you get

  • Umami v3.5.0, pinned by tag and digest, serving the dashboard, the /script.js tracker and the /api/send collection endpoint
  • PostgreSQL 18 on a private network with a volume
  • A generated UMAMI_ADMIN_PASSWORD applied to the admin account on first boot; the default password umami does not work
  • Umami waits for Postgres on a cold start, migrations retry, and telemetry is off
  • A generated APP_SECRET that signs login tokens

Good for

  • Cookie-free analytics for websites, blogs and documentation
  • Custom events, funnels, retention and UTM campaign tracking
  • Analytics for several sites and teams from one dashboard
  • Public dashboards shared with clients

Cost

Measured on Railway (Umami 3.5.0): about 0.1 to 0.16 GB of RAM for Umami at idle and up to about 0.3 GB while serving the dashboard and collecting events, with Postgres at roughly 0.06 to 0.1 GB. Under a 0.5 GB limit, login, site setup and event collection worked. Database size grows with traffic.

Services in this template

  • Umami · from baranberkay96/umami-railway
  • Postgres · image ghcr.io/railwayapp-templates/postgres-ssl:18.6@sha256:4fc833b43d2aaf6ecc9b18615f5533602b723a32d0e2efe2dae8ca0f7217c863

Questions people ask

How do I log in?

When the Umami service is healthy, copy UMAMI_ADMIN_PASSWORD from its Variables tab and sign in at the public domain as admin. Then add a site under Settings, Websites, and paste the tracking code into your pages.

What if I forget the admin password?

Set a new UMAMI_ADMIN_PASSWORD (at least 8 characters), set UMAMI_ADMIN_PASSWORD_SYNC=true and redeploy. The variable is then applied on every deploy until you set the sync flag back to false.

Is data kept across redeploys, and what should I keep?

Postgres has a volume, so sites, users and events persist. Keep APP_SECRET: changing it logs everyone out and makes returning visitors count as new.

How do I upgrade?

The version is pinned in services/umami/Dockerfile (tag and digest). Railway shows an update when the template changes; to pick a version yourself, eject the service to your own repo and change the FROM line. Migrations run on boot; back up Postgres before major upgrades.

What is the licence?

Umami is MIT-licensed. This is a community template, not affiliated with Umami Software.

Full template README

Deploy and Host Umami with Railway

Umami is an open-source, privacy-focused alternative to Google Analytics: page views, visitors, referrers, events, funnels and more, without cookies or personal data. This community template runs Umami v3.5.0 with Postgres, and replaces the public default login (admin/umami) with a generated password before the dashboard goes online.

About Hosting Umami

Umami needs one Node.js service and a PostgreSQL database. Its database migration creates an administrator with the well-known password umami, so a fresh public instance can be taken over by anyone who logs in first. This template sets your own generated password on that account during boot, before the server listens, and checks it on every deploy. The Umami version is pinned by tag and digest, Umami waits for Postgres on a cold start instead of crash-looping, migrations retry while the database finishes starting, and telemetry is off. Measured on Railway: about 0.1-0.3 GB of RAM for Umami and under 0.1 GB for Postgres, so it runs on every plan, Free included.

Common Use Cases

  • Privacy-friendly, cookie-free analytics for websites, blogs and documentation
  • Product and marketing analytics: custom events, funnels, retention, UTM campaigns
  • Analytics for several sites and teams from one self-hosted dashboard
  • Replacing Google Analytics where consent banners or data residency matter
  • Sharing public dashboards with clients

Dependencies for Umami Hosting

  • Umami v3.5.0 (ghcr.io/umami-software/umami, official image)
  • PostgreSQL 18 (Railway's postgres-ssl image)

Deployment Dependencies

Implementation Details

Service Role Public
Umami Dashboard, API and the /script.js tracker plus /api/send collection endpoint yes
Postgres Websites, users, sessions and events; private only, with a volume no

First login

  1. Wait until the Umami service is healthy. The first deploy builds a thin image on top of Umami's official one and runs the database migrations, which takes a few minutes.
  2. Open the Umami service → Variables and copy UMAMI_ADMIN_PASSWORD (generated, 24 characters).
  3. Open the Umami service's public domain and sign in as admin with that password. The default password umami does not work.

Adding a website. In Umami go to Settings → Websites → Add website, then copy the tracking code into the head section of your site. It loads https://{your-umami-domain}/script.js and sends page views to /api/send.

Admin password. UMAMI_ADMIN_PASSWORD is applied once, while the account still has the seeded default password. After that, change your password in Umami's settings as usual; redeploys keep it. Forgot it? Set a new UMAMI_ADMIN_PASSWORD, set UMAMI_ADMIN_PASSWORD_SYNC=true and redeploy; the variable is then applied on every deploy until you set it back to false. Invalid values never break the deploy: a password shorter than 8 characters keeps the stored one, and on a fresh install locks the admin login until you set a valid one (the deploy log says so).

Keep APP_SECRET. It signs login tokens and is part of the anonymous visitor hash. Changing it logs everyone out and counts returning visitors as new.

Custom domain. Add it in the Umami service's Settings → Networking and use it in your tracking code. A custom domain on your own site's domain (for example stats.{your-domain}) is also less likely to be blocked by ad blockers.

Resources and plans. Measured on Railway (Umami 3.5.0, 2026-11-14): Umami uses about 0.1-0.16 GB of RAM at idle and up to about 0.3 GB while serving the dashboard and collecting events; Postgres uses about 0.06-0.1 GB. Under a 0.5 GB limit (the Free plan's per-service cap) login, website setup and event collection all worked. That is roughly $2-4 a month at Railway's usage prices for a small site, so every plan works: Free, Trial, Hobby and Pro. Database size grows with traffic; Free and Trial volumes are small, so busy sites belong on Hobby or Pro.

Scaling. One Umami replica handles a lot of traffic, since each page view is one small insert. For more throughput raise the service's CPU and memory or add replicas (the service has no volume). For very large sites Umami also supports ClickHouse and Redis; they are not part of this template.

Upgrading Umami. The version is pinned in services/umami/Dockerfile (tag and digest). When this template is updated, Railway shows an update for the Umami service. To pick a version yourself, eject the service to your own GitHub repo and change the FROM line to another release of ghcr.io/umami-software/umami. Migrations run automatically on boot; back up Postgres before major upgrades.

License. Umami is MIT-licensed. This is a community template, not affiliated with Umami Software.

Why Deploy Umami on Railway?

Railway runs Umami and its Postgres database in one project with a private network, HTTPS domains, logs and usage-based pricing. This template gives you a locked-down, cookie-free analytics dashboard in a few minutes, at a low monthly cost, with no default credentials left behind.

Guides for this template

View Umami v3 Privacy-First Web Analytics on railway.com

Similar templates

TemplateCategoryDeploymentsDeploys
Umami
An open source, privacy-focused alternative to Google Analytics
Analytics4,236
Metabase
Fast analytics and integrated tooling for companies exploring their data
Analytics2,642
Plausible Analytics
An easy to use, privacy-friendly, open-source Google Analytics alternative
Analytics915
Formbricks
An open source surveys & experience management solution
Analytics693
SearXNG | Open Source Search API for AI Agents
Self-Host SearXNG. Open source, AI-ready search, 70+ engines combined
Analytics140
PostHog Product Analytics SuiteTracklayer
Self-hosted product analytics, session replay, feature flags and A/B tests.
Analytics14