A platform is only as good as the night it fails. FFS was built by people who run it in production, so the operational work — installing, checking, deploying, watching, rolling back — is part of the product, not an afterthought left to scripts.

Install from signed packages

FFS runs on Linux. Signed RPM packages are ready for Fedora 44 and 45 and AlmaLinux 9 and 10, from repositories with a signed index; other distributions are a port, not a rewrite — ask us. One package can stand for a whole product: a single dnf install brings a complete application with every plugin it needs.

dnf install fedem-all-repos     # the repositories and their signing keys
dnf install depotify-all        # an entire product, in one step

A site's own configuration can be packaged too, so what is installed on a server is always a version you can name — and the files an operator edited are kept on upgrade.

Check before you change

  • ffs --check-config validates the whole configuration and lists every problem, before a single site is touched.
  • Dangerous defaults are refused, not tolerated — see Fail closed.
  • A setup wizard guides a first installation and rescues a broken one: listeners, sites, certificates and groups in a browser, with every change applied as one transaction and rolled back completely if a step fails. Comments and layout of the configuration files survive a save.

Deploy with a safety net

The deployment tooling ships only signed packages, content and control scripts to a server — never source code, never development keys. After a deployment it checks the health of every site on its real addresses, and if a site does not come up, it rolls back to the previous release and says so.

See what is happening

  • Logs per site and one for the front-end, with sortable, configurable timestamps; rotated by size without losing the file handle the server holds.
  • Guard and bot-filter decisions are logged with their reason, and command-line tools show what was blocked and lift a ban that was wrong.
  • The front-end can report every WebSocket it opened and closed, and a summary when it stops.
  • Applications keep an audit log of who did what, queryable by resource, user, action and outcome.

Data where it belongs

Configuration, rules and accounts live in plain, commented, reviewable files. Application data uses SQLite by default — nothing to install, nothing to operate — and PostgreSQL where a service needs it, chosen per service. Moving an existing service to PostgreSQL happens automatically on its first start against the new database, with counts and references compared before the old file is retired. A database that is down stops only its own service.

Your infrastructure, your rules

FFS runs on your servers or in your containers. It does not phone home, needs no cloud account, and works on a network that has no internet connection at all.