How upc/ravada works
Project type and purpose
• This is ravada, an open-source Remote Virtual Desktops Manager (a full-stack web application plus background daemon) for provisioning and accessing QEMU/KVM VMs via a browser/SPICE client.
• The README confirms its goal: web UI for VM life-cycle management, snapshot/clone, LDAP/SQL auth, remote console (SPICE).
File structure
• Top-level “dockerfy” holds two containers: back (rvd_back daemon) and front (web server).
• “lib/Ravada” contains Perl modules implementing the core API, domain drivers (KVM, LXC), authentication, booking, volume, networking, WebSocket.
• “script/rvd_back” and “script/rvd_front” are entrypoints for the backend daemon and frontend web server.
• “public/” holds static assets (CSS, JS, fonts, images) and AngularJS modules (“public/js/booking”).
• “templates/” and “public/templates/” store server-side (Mojolicious) and Angular/HTML templates.
• “sql/” holds schemas for MySQL and SQLite plus upgrade scripts.
• “etc/” holds configuration files (ravada.conf, systemd units), XML definitions for VM images/templates.
README for additional insights
• Backend written in Perl, uses Mojolicious (the presence of .ep templates and WebSocket modules)
• Supports LDAP, SQL auth, KVM hypervisor driver by default, plug-in architecture for other hypervisors.
• Typical deployment uses a DB (MySQL/PostgreSQL/SQLite), an HTTP front, and a daemon that talks to libvirt.