Queues, Cache & Events

Queues

Jobs are typed — no any blobs. A job implements the Job trait with a handle method, and dispatch supports dispatch, dispatchSync, chain, delay, onQueue, and onConnection.

cargo artisan make:job SendWelcomeEmail
cargo artisan queue:work --queue=default --max-jobs=10

Retries use an exponential backoff loop on tokio::spawn workers — no external backoff crate.

Cache

Cache::remember("posts.featured", 3600, || async { fetch_featured().await }).await?;

MemoryStore is a lock-guarded HashMap with TTL support. RedisStore is a real implementation behind the opt-in redis feature, with atomic SET NX insert-if-absent and Lua compare-and-delete lock release. Without the feature — or when Redis is unreachable — operations return a typed StoreUnavailable rather than panicking.

Events and listeners

• Event trait plus Listener, where Queue { enable: true } makes a listener async.
• dispatch and dispatchAfterResponse cover immediate and post-response dispatch.
• Framework events include JobAttempted { exception } and QueueBusy { connectionName }.

Scheduling

cargo artisan schedule:list --json
cargo artisan schedule:run
cargo artisan schedule:pause     # emits SchedulePaused
cargo artisan schedule:resume    # emits ScheduleResumed

Frequencies cover daily, everyMinute, cron expressions, skipIfStillRunning, and onOneServer. Scheduling is hand-rolled: a 5-field cron parser driven by a 60-second tokio ticker, with no cron crate dependency.

flowchart LR
    A["dispatch(job)"] --> B{"Queue::route"}
    B --> C["database driver"]
    B --> D["redis driver"]
    C --> E["worker loop"]
    D --> E
    E --> F{"handle ok?"}
    F -->|yes| G["JobAttempted"]
    F -->|no| H["backoff retry"] --> I["failed_jobs"]