Concepts
| Term | Meaning |
|---|---|
| Administrator | A provider account. Can manage nodes, customers, templates, plugins and settings. Must use two-factor sign-in. |
| Customer | A normal account that owns servers. Limited by an optional quota. |
| Collaborator | A user invited to one server with some of view, console, files, backups, manage. |
| Node | A Linux host running Docker and the Fledge agent. Has a capacity (memory, CPU, disk) and a location label. |
| Server | One game server container plus its data volume, owned by a customer and placed on a node. |
| Template | A recipe for a kind of server: image, startup command, ports, environment, variables, default resources. Templates are versioned. |
| Job | A durable unit of work for a node (create, start, stop, backup, file operations...). The agent polls for jobs and reports results. |
| Allocation | A host port reserved for a server on a node. A server's base port plus offsets from its template and extra ports. |
| Add-on | A mod or plugin file installed into a server's folder by a catalog plugin, tracked by the panel. |
| Plugin | A sandboxed program that extends the panel (today: add-on catalogs, event hooks and payment providers). |
| Schedule | A cron or interval rule that runs a chain of steps on a server (command, wait, backup, power). |
| Limits | What a customer may own. Layered: panel defaults, plans, and per-customer overrides. Can be switched off. See Limits. |
| Plan | Something you sell or give: a server plan creates one server per subscription, an account plan adds to a customer's limits. See Plans. |
| Subscription | A customer's hold on a plan, paid through a payment provider (or given for free). Its status controls the server. See Billing. |
| Store | Where customers buy plans. Closed until everything it needs works. See The store. |
| Payment provider | A plugin that talks to Stripe or another provider. Fledge keeps the orders and entitlements. |
| Failover | Re-creating a server from its newest backup on another node after its node has been silent too long. |
| Resilience | The page that shows backup freshness, failover readiness and history. |
Lifecycle of a server
mermaid
sequenceDiagram
participant U as User
participant A as API
participant N as Node agent
U->>A: Create server (template, owner, resources)
A->>A: Pick a node with room, reserve ports
A->>N: create job (via polling)
N->>N: Make the data volume, start the container
N->>A: Result and heartbeat (running)
U->>A: Console, files, backups
A->>N: More jobsEverything the panel asks of a node is a job. Jobs are leased, can be retried, and are idempotent where it matters, so a lost connection does not lose or duplicate work.
