Queues

Concurrency control, rate limiting, priority, and partitioned processing for workflows.

Creating a Queue

q := rt.Queue("emails",
    turbine.WithWorkerConcurrency(5),
    turbine.WithGlobalConcurrency(10),
    turbine.WithRateLimiter(turbine.RateLimiter{Limit: 100, Period: time.Minute}),
    turbine.WithPriorityEnabled(),
    turbine.WithPartitionQueue(),
)

Queue Options

Option	Description
WithWorkerConcurrency(n)	Max concurrent workflows this instance will run from this queue
WithGlobalConcurrency(n)	Max concurrent workflows across all instances (enforced at dequeue time)
WithRateLimiter(rl)	Limit how many workflows are dequeued per time period
WithPriorityEnabled()	Dequeue lower-priority-number workflows first
WithPartitionQueue()	Process workflows with the same partition key sequentially
WithMaxTasksPerIteration(n)	Max workflows dequeued per poll cycle (default: 100)

Enqueuing Workflows

turbine.Run(rt, sendEmail, recipient, turbine.WithQueue("emails"))

You can also set priority, deduplication, and partition keys:

turbine.Run(rt, sendEmail, recipient,
    turbine.WithQueue("emails"),
    turbine.WithPriority(1),                       // lower = higher priority
    turbine.WithDeduplicationID("user-123"),        // prevent duplicates
    turbine.WithQueuePartitionKey("tenant-1"),      // partition key
)

Partitioned Queues

With WithPartitionQueue(), workflows sharing the same QueuePartitionKey run sequentially. Different partition keys run concurrently.

rt.Queue("orders", turbine.WithPartitionQueue())

// These run sequentially (same partition key)
turbine.Run(rt, processOrder, order1, turbine.WithQueue("orders"), turbine.WithQueuePartitionKey("tenant-A"))
turbine.Run(rt, processOrder, order2, turbine.WithQueue("orders"), turbine.WithQueuePartitionKey("tenant-A"))

// This runs concurrently with the above (different partition key)
turbine.Run(rt, processOrder, order3, turbine.WithQueue("orders"), turbine.WithQueuePartitionKey("tenant-B"))

Multi-Instance Setup

INFO

By default, every instance processes all queues. Use Listen to restrict which queues a specific instance handles.

rt.Listen(q)

Dashboard

See the queue example for a working demo.