Repository navigation
Replies: 1 comment
|
Both work. A few details are worth getting right. 1. Shared Postgres. Its own database and role in your existing instance is fine. Umami needs PostgreSQL 12.14+ (README) and, on every container start, runs 2. Script and collect on example.com. That's what those two variables are for. In the Docker image the middleware rewrites For visitor IPs, Umami prefers This is from reading the v3.4.1 source, not from running your exact Caddy setup. |
Uh oh!
There was an error while loading. Please reload this page.
Small business site, one 4 GB VPS, Docker Compose, Caddy in front, Postgres 17 already running for the site with a nightly pg_dump sidecar. Under 1,000 pageviews a month, replay off. Two points of the planned setup I would like your view on:
Umami runs in the site's existing Postgres container, in its own database and role, instead of a second Postgres, so one instance gets patched and the existing backup job covers it. Fine for Umami, or do you recommend its own Postgres?
Dashboard on stats.example.com. Tracker script and collect endpoint served from example.com itself: Caddy routes two paths on example.com to the Umami container, named by TRACKER_SCRIPT_NAME and COLLECT_API_ENDPOINT, so the site's CSP (script-src 'self' with nonce, connect-src 'self') needs no new host, and data-host-url stays unset. Is this the setup you recommend, or is there a better one?
Thank you.
All reactions