iii-queue-processing（iii 队列处理）

Queue Processing

Comparable to: BullMQ, Celery, SQS

Key Concepts

Use the concepts below when they fit the task. Not every queue setup needs all of them.

• Named queues are declared in iii-config.yaml under queue_configs
• Standard queues process jobs concurrently; FIFO queues preserve ordering
• TriggerAction.Enqueue({queue}) dispatches a job to a named queue
• Failed jobs auto-retry with exponential backoff up to max_retries
• Jobs that exhaust retries land in a dead letter queue for inspection
• Each consumer function receives the job payload and a messageReceiptId
• Fan-out is achieved by having one producer trigger multiple distinct consumer functions via separate enqueue calls

Architecture

Producer function
  → TriggerAction.Enqueue({ queue: 'task-queue' })
    → Named Queue (standard or FIFO)
      → Consumer registerFunction handler
        → success / retry with backoff
          → Dead Letter Queue (after max_retries)

iii Primitives Used

Reference Implementation

See ../references/queue-processing.js for the full working example — a producer that enqueues jobs and a consumer that processes them with retry logic.

Also available in Python: ../references/queue-processing.py

Also available in Rust: ../references/queue-processing.rs

Common Patterns

Code using this pattern commonly includes, when relevant:

• registerWorker(url, {workerName}) — worker initialization
• registerFunction(id, handler) — define the consumer
• trigger({function_id, payload, action: TriggerAction.Enqueue({queue})}) — enqueue a job
• payload.messageReceiptId — track or acknowledge the job
• trigger({function_id: 'state::set', payload}) — persist results after processing
• const logger = new Logger() — structured logging per job

Adapting This Pattern

Use the adaptations below when they apply to the task.

• Choose FIFO queues when job ordering matters (e.g. sequential pipeline steps)
• Set max_retries and concurrency in queue config to match your workload
• Chain multiple queues for multi-stage pipelines (queue A consumer enqueues to queue B)
• For idempotency, check state before processing to avoid duplicate work on retries
• Use fan-out by enqueuing to multiple consumer functions when a single event requires parallel processing (e.g. payment + notification + audit)

Engine Configuration

Named queues are declared in iii-config.yaml under queue_configs with per-queue max_retries, concurrency, type, and backoff_ms. Fan-out is a pattern (one producer triggers multiple consumer functions), not a queue config key. See ../references/iii-config.yaml for the full annotated config reference.

Pattern Boundaries

• If the task only needs fire-and-forget without retries or ordering, prefer iii-trigger-actions with TriggerAction.Void().
• If failed jobs need special handling or alerting, prefer iii-dead-letter-queues for the DLQ consumer.
• If the task is step-by-step orchestration with branching, prefer iii-workflow-orchestration.
• Stay with iii-queue-processing when the primary need is reliable async job execution with retries.

When to Use

• Use this skill when the task is primarily about iii-queue-processing in the iii engine.
• Triggers when the request directly asks for this pattern or an equivalent implementation.

Boundaries

• Never use this skill as a generic fallback for unrelated tasks.
• You must not apply this skill when a more specific iii skill is a better fit.
• Always verify environment and safety constraints before applying examples from this skill.