BullMQ

Use this skill when work touches BullMQ Redis queues, workers, flows, retries/scheduling, NestJS BullMQ, or Pro-vs-OSS decisions.

Workflow

1. Inspect the local BullMQ surface before changing code:

- Package versions for bullmq, @nestjs/bullmq if present, Redis clients (ioredis, redis), Node/Bun, and TypeScript. - Connection shape: options object vs shared client vs adapters (createNodeRedisClient, createBunRedisClient); prefix; Redis DB index. - Producers (Queue, FlowProducer) vs consumers (Worker); process isolation (inline vs sandboxed file / worker threads). - Default job options, schedulers, rate limits, dedupe usage, Nest module registration if present.

1. Refresh current official docs when the task depends on latest APIs, version drift, Pro features, adapters, or Nest wrappers. Start from [source-map.md](references/source-map.md).
2. Route the work to the focused references:

- Setup, Redis connections, Queue, Worker, QueueEvents: [setup-connections.md](references/setup-connections.md). - Job lifecycle and JobsOptions: [jobs-options.md](references/jobs-options.md). - Parent-child flows / FlowProducer: [flows.md](references/flows.md). - Retries, delays, job schedulers, rate limits, dedupe: [retries-scheduling-limits.md](references/retries-scheduling-limits.md). - Ops, testing, NestJS, Pro vs OSS, polyglot: [production-ops-integrations.md](references/production-ops-integrations.md).

1. Preserve the repository's existing connection factories, Nest modules, and deployment topology unless the user explicitly asks for a migration.
2. Verify behavior at the narrowest useful boundary (processor unit), then queue/worker integration.

Core Judgment

• Treat Redis as shared infrastructure. Reuse connection options or factories deliberately; Workers and QueueEvents still duplicate connections for blocking commands, so the client must support duplicate().
• Attach error listeners on Queue, Worker, QueueEvents, and FlowProducer. Unhandled connection errors crash Node.
• Design processors as idempotent. Jobs can retry, stall after lock loss, or run more than once—side effects need safe keys or transactional checks.
• Set attempts and backoff intentionally. Throw UnrecoverableError for poison/permanent failures; throw Error objects (not strings) from processors.
• Prefer FlowProducer for parent-child dependency graphs. Do not hand-roll “wait for children” with races across queues. Flow job opts omit repeat, deduplication, and debounce. Pick child failure opts deliberately (failParentOnFailure, ignoreDependencyOnFailure, removeDependencyOnFailure, continueParentOnFailure).
• Rate limiting is global per queue across workers. Do not resurrect removed patterns (QueueScheduler, pre-v3 groupKey rate limits) on BullMQ 5.x. Group rate limits are Pro-only.
• Deduplication needs a stable deduplication.id. Choose simple vs throttle (ttl) vs debounce (extend/replace + delay) vs keepLastIfActive for the product semantics.
• Priority: lower number wins among prioritized jobs (1..2097151); unset/0 is unprioritized and is processed before any prioritized job.
• Tune removeOnComplete / removeOnFail (age or count keep policies). Unbounded completed/failed sets grow Redis without bound. Never use ioredis keyPrefix; use BullMQ prefix (brace hash tags on Cluster).
• Close workers gracefully on shutdown (worker.close()). Killing processes mid-lock causes stalls and duplicate work unless ops expect that window.
• Prefer queue.upsertJobScheduler over legacy repeat / QueueScheduler for cron/interval factories. Prefer @nestjs/bullmq over legacy @nestjs/bull.
• Do not recommend BullMQ Pro-only APIs (groups, group rate limits, batches, observables) unless the project already uses Pro. Call the boundary out explicitly.
• Keep processors thin and non-blocking on the event loop. Use sandboxed processors / useWorkerThreads when CPU-heavy or crash-isolating work is required. Job data is plaintext in Redis — avoid secrets or encrypt fields.

Verification

Prefer repository-owned commands. For meaningful BullMQ changes, cover the relevant subset:

• Typecheck Queue/Worker/FlowProducer options, job data/returnvalue types, and Nest injection tokens if used.
• Unit-test processor success, retryable failure, UnrecoverableError, and progress updates in isolation.
• Integration test against Redis: add → process → complete/fail; assert job state transitions.
• Delayed/scheduler tests with short intervals; assert no duplicate schedules after redeploy when that matters (upsertJobScheduler).
• Rate-limit / dedupe tests: assert ignored/replaced/deferred behavior and any deduplicated events.
• Flow tests: children complete before parent; parent reads getChildrenValues; child failure policy (failParentOnFailure / ignoreDependencyOnFailure / removeDependencyOnFailure / continueParentOnFailure) matches product intent.
• Stall/lock smoke when changing lockDuration, stalledInterval, maxStalledCount, or long-running processors.
• Graceful-shutdown smoke: SIGTERM/close path finishes in-flight work without unexpected orphaned locks beyond the stall window.
• NestJS module smoke: queue registration, WorkerHost processors, and FlowProducer injection when Nest is in play.
• Redis growth check when changing retention (removeOnComplete/removeOnFail) or adding high-throughput queues.

Report which checks ran, which did not, and any Pro/OSS or package-version assumptions that remain.