Architecture¶

Record Store is one Rust workspace. Protocol crates call shared application services; they never reach into storage internals directly.

Request path¶

flowchart TB
    Client[S3 client] -->|SigV4 :7600| S3[S3 protocol]
    Console[Console :7602] -->|Bearer token :7601| API[Management API]
    S3 --> Svc[Object and bucket services]
    API --> Svc
    Svc --> Store[Filesystem store]
    Svc --> Meta[(Metadata catalog)]
    Store --> Disk[(objects/)]

Both protocol surfaces go through the same service layer, so an object written over S3 and an object written through the console are the same object under the same rules.

Console path¶

flowchart LR
    Browser --> Console[Console server :7602]
    Console -->|Bearer token| API[Management API :7601]
    API --> RS[Record Store]

The browser only ever talks to the console's own origin. The console server attaches the management credential server-side, so it lives in an HTTP-only cookie the page cannot read, no CORS configuration is required, and the browser never reaches the management API, the stored objects, or the metadata catalog.

Crates¶

Crate	Responsibility
`record-store-core`	Validated domain model: names, keys, checksums, ranges
`record-store-service`	Shared bucket and object application services
`record-store-s3`	S3 protocol, SigV4, XML, multipart, versioning
`record-store-api`	Management HTTP API and management roles
`record-store-storage`	Streaming filesystem backend and recovery journal
`record-store-metadata`	Durable indexed catalog and ordered migrations
`record-store-auth`	Encrypted credentials and authorization policies
`record-store-audit`	Durable bounded security audit trail
`record-store-sharing`	Share and embed capabilities
`record-store-events`	Durable events and signed webhook delivery
`record-store-lifecycle`	Incremental expiration worker
`record-store-config`	Configuration loading and validation
`record-store-observability`	Structured tracing initialization
`record-store-erasure`	Reed-Solomon library, not currently wired in

See Repository Structure for the layout including applications and the console.

On-disk layout¶

<data-directory>/
├── metadata/catalog.redb
├── metadata/credentials.redb
├── metadata/audit.redb
├── metadata/events.redb
├── metadata/lifecycle.redb
├── objects/<2 hex>/<2 hex>/<object UUID>
├── system/
└── tmp/

Payloads are addressed by generated UUID. No bucket name or object key ever becomes part of a filesystem path.