diff --git a/README.md b/README.md index bde7d3d..d5fc015 100644 --- a/README.md +++ b/README.md @@ -3,7 +3,7 @@ RecFlare is an implementation of RecNet — the Rec Room backend — built on Cloudflare Workers. It implements the network services the Rec Room client talks to — accounts, auth, rooms, matchmaking, economy, chat, notifications, and more — -each as an independent Worker on its own subdomain. +each as an independent Worker on their own subdomains, just how RecNet was. > **Disclaimer:** This is an unofficial, fan-made project for preservation and > experimentation. It is not affiliated with, endorsed by, or connected to Rec @@ -14,7 +14,7 @@ each as an independent Worker on its own subdomain. There are already so many multiplayer clones, why? 1. None of them are fully open source (some had leaks of old code). -2. None of them run on microservice architecture. +2. None of them run on microservice architecture, most on a single server. 3. None of them had unit tests. 4. "Upgrading the server" is not sustainable plan for growth. 5. This was fun (sort of). @@ -35,7 +35,11 @@ RecFlare is compatible with the Room with manifest `7859140924515540835`. Other client or game versions may expect different endpoints and response shapes and are not supported. -See the project page for information on how to mod the game to connect to this server. +Generally speaking any client that effectively rewrites the nameserver can be used +with this server. + +See the above project page for information on how to mod the game to connect to this +server. ## How it works @@ -105,40 +109,54 @@ apex/`ns` host and isn't listed within it. Each implemented worker has its own ## Why Cloudflare? -- **It's free to run a lot of services.** Cloudflare Workers' free tier is keyed - to usage, not to the number of Workers — so whether you deploy 1 service or all - 36, the baseline cost is the same. You only start paying once usage crosses the - free-tier limits. -- **It mirrors RecNet's architecture.** Rec Room's backend is a set of - independent microservices, not one monolith. Modeling each service as its own - Worker keeps RecFlare's structure close to the real thing — services scale, +- **Makes it easy to mirror RecNet** + Rec Room's backend is (was) a set of independent microservices, not one big monolith. + Modeling each service as its own isolated Worker keeps RecFlare's structure close to + the real thing — services scale, fail, and deploy independently — instead of collapsing everything into a single giant server. +- **It's free/cheap to run a lot of service.** + Cloudflare Workers' free tier is keyed + to usage, not to the number of Workers — so whether you deploy 1 service or all + 36, the baseline cost is the same. You only start paying once usage crosses the + free-tier limits. Additionally for development or maybe a private instance, the cost + is near zero when not in use. +- **Bundled Cloud CDN/Storage/SQL** + Cloudflare offers several cloud services we can rely on so the microservices can remain + stateless (effectively read-only). They are also scalable by default so we don't need to worry + about adding more disk space or upgrading services. If we start outgrowing the limits of these, + well, we'll cross that bridge when we get to it. + - D1 (a SQLite-compatible distributed database) + - R2 (service like S3 for mass file hosting) + - KV (service to distributed offer key/value stores) + - Durable Objects (for a notifications hub) ## Do I have to use Cloudflare? -No. The services are plain [Hono](https://hono.dev) apps, so the request-handling +Short answer, no. The services are plain [Hono](https://hono.dev) apps, so the request-handling code isn't tied to Cloudflare and can be deployed to other hosting providers — AWS (Lambda), Vercel, Netlify, Fly.io, a plain Node/Bun server, and so on. -**However** - -The catch is everything _around_ the code. RecFlare leans on Cloudflare for the -deployment and infrastructure layer — custom-domain routing per service, plus the +Long answer: the catch is everything _around_ the code. RecFlare leans on Cloudflare for the +deployment (Wrangler) and infrastructure layer — custom-domain routing per service, plus the storage bindings (D1, KV, R2, Durable Objects) the workers use. On another provider you'll need to provide equivalents (per-service routing, databases, object storage, a pub/sub or WebSocket layer) and wire up the deployment yourself. +So for example if you wanted to run on Vercel, you'd have to swap out KV for Redis, which are very similar +services but would require small code changes. + ## Prerequisites - node (modern) - pnpm - bun -- jq -- A Cloudflare account with a zone (domain) you control, for deploying +- jq/awk/sed +- A Cloudflare account with a zone (domain) you control, for deploying. -Cloudflare's free plan is good enough for testing (100k requests/day) but the -Rec Room client is pretty chatty. Frequent testing may exhaust that quota. +Cloudflare's free plan is good enough for testing (100k worker requests/day) but the +Rec Room client is pretty chatty. Frequent testing may exhaust that quota. The $5/month +Worker plan includes 10M/month. See https://developers.cloudflare.com/workers/platform/pricing/#workers @@ -152,97 +170,60 @@ just install **Configure your custom domain:** -Create a new .env file: +Create a new .env file from the template: ```bash cp .env.example .env -# edit .env and set RECFLARE_DOMAIN to your domain ``` -Or, export it: +Edit `.env` and set `RECFLARE_DOMAIN` to your domain (or declare it with `export RECFLARE_DOMAIN=rec.example.com`) -```bash -export RECFLARE_DOMAIN=rec.example.com -``` - -`just deploy` resolves the base domain at deploy time and passes it to wrangler — -each worker is attached to its custom domain via `--domain .`, -and the base domain is injected as the `DOMAIN` var so the `ns` service-discovery -document and the api share-link base URL are built at runtime. Nothing in version -control is rewritten; committed `wrangler.jsonc` files have no routes. - -Per-app subdomain overrides come from -`RECFLARE_SUBDOMAINS` (a JSON object, e.g. `'{"playersettings":"settings"}'`). +(Optional) - per-app subdomain overrides come from +`RECFLARE_SUBDOMAINS` (a JSON object, e.g. `'{"playersettings":"settings"}'`). This would be used +if you wanted to merge two services together. **Create the storage resources:** -The workers bind Cloudflare storage primitives — one shared D1 database, two KV -namespaces, two R2 buckets, and a Durable Object. Create them once against your -Cloudflare account, then record the ids in `.env`. The committed `wrangler.jsonc` -files carry `"local"` placeholders; the real ids are spliced in at deploy time, so +The workers bind Cloudflare storage primitive. Create them once against your +Cloudflare account, then record the IDs in `.env`. The committed `wrangler.jsonc` +files carry `"local"` placeholders; the real IDs are spliced in at deploy time, so nothing in version control needs editing. Authenticate wrangler first (`wrangler login`). -_D1 — one shared `recflare` database_ (bound by `api`, `accounts`, `auth`, `match`, -`rooms`): - ```bash wrangler d1 create recflare -# copy the printed database_id into .env: -# RECFLARE_D1= -``` - -Then apply the schema. The `rooms` and `auth` workers own the migrations under -their `apps//migrations/` directories. `just migrate` applies them to the -remote database (splicing `RECFLARE_D1` into the `"local"` placeholder the same way -`just deploy` does, so you don't edit any config): - -```bash -just migrate # migrate every worker that owns migrations -just migrate -F rooms # or scope to one worker -just migrate -- --local # target the local dev db instead of remote -``` - -This runs non-interactively (it sets `CI`), so wrangler's "database may be -unavailable during the migration" confirmation is auto-accepted — turbo can't -forward your answer to the task anyway. A backup is still captured. - -_KV — two namespaces_ (`RECFLARE_MATCH_PRESENCE` for `match`/`auth`, -`RECFLARE_PLAYER_SETTINGS` for `playersettings`): - -```bash wrangler kv namespace create RECFLARE_MATCH_PRESENCE wrangler kv namespace create RECFLARE_PLAYER_SETTINGS ``` -Record both ids in `.env` as a single JSON object keyed by binding name (note the -surrounding single quotes — without them the shell strips the inner quotes): +Take the IDs output from the commands and put them into `.env`. (or with CI: `RECFLARE_KV='{"RECFLARE_MATCH_PRESENCE":"","RECFLARE_PLAYER_SETTINGS":""}'`) + +Then apply the schema. `just migrate` will set up the database and populate it with data. This runs non-interactively, so be careful! ```bash -RECFLARE_KV='{"RECFLARE_MATCH_PRESENCE":"","RECFLARE_PLAYER_SETTINGS":""}' +just migrate # migrate every worker that owns migrations +just migrate -F rooms # or scope to one worker ``` -_R2 — two buckets_ (`recflare-cdn` for `cdn`, `recflare-img` for `api`/`img`): +### R2 and Durable Objects + +You only have to create the buckets: ```bash wrangler r2 bucket create recflare-cdn wrangler r2 bucket create recflare-img ``` -R2 buckets are referenced by name in the committed `wrangler.jsonc`, so there is -nothing to add to `.env`. See [`apps/img/README.md`](apps/img/README.md) for -seeding the default avatar/profile images. +### Durable Objects -_Durable Objects — no manual setup_. The `notify` worker's `RECFLARE_NOTIFICATIONS_HUB` -binding (class `NotificationsHub`) is provisioned automatically from the migration -declared in its `wrangler.jsonc` on first deploy — there is no id to create or set. +Nothing manual to do here. The object is created manually. **Run the development microservices:** > **Note:** This runs, but the name-server document still advertises the deployed > hosts, not your local instances — so service discovery won't resolve locally. > You can still call each service directly; each Wrangler instance runs on its own -> port. +> port. Maybe we can get this working somehow. @todo ```bash just dev @@ -250,17 +231,20 @@ just dev **Deploy all workers:** -```bash -just deploy -``` +This will deploy all workers to respective endpoints (*.example.com) Deploying requires `wrangler` to be authenticated against your Cloudflare account (`wrangler login`, or `CLOUDFLARE_API_TOKEN` / `CLOUDFLARE_ACCOUNT_ID` -in the environment). It also requires the storage resources to exist and their -ids to be set in `.env` — see **Create the storage resources** above. At deploy -time the deploy script splices `RECFLARE_D1` and `RECFLARE_KV` into the `"local"` -placeholders in each worker's `wrangler.jsonc`; a missing id fails the deploy with -a message naming the binding. +in the environment). + +It requires the storage to be set up above, otherwise, deployments may fail. + +You can always re-run it as often as you wish. + +```bash +just deploy # for all services +just deploy -F rooms # for a single microservice +``` ## Repository Structure @@ -308,7 +292,11 @@ See "Run the development microservices" above. It may be possible later as Wrang ### Can I use this to make my own server? -Yes, that's the point. +Yes, that's the point. Just set your custom domain and deploy it. + +### Is there an admin panel? + +Not yet. But there could be. ### Can I copy this project and modify it?