Queues

Cloudflare Queues is a message queue: a reliable waiting line where one Worker drops off messages and another Worker picks them up to process later. It guarantees delivery, so nothing is lost even if the receiver is busy or briefly fails.

A "message" is just a small chunk of data — for example "send a welcome email to user 123" or "resize this uploaded photo." Instead of doing that slow work while the user waits, you put a message in the queue and respond immediately. The work happens in the background.

If your Worker tries to do everything during a single request, slow tasks make users wait and a single failure loses the work. A queue decouples "accepting the work" from "doing the work," making your app faster and far more reliable.

Use Queues whenever work can happen later instead of right now, or when one part of your system needs to hand a job to another safely.

You create a queue with Wrangler, then wire one Worker as the producer (sends messages) and another handler as the consumer (processes them). A single Worker can be both.

1. Create a queue

   Make a new queue by name. This is where messages will wait.

   bashcontent_copyCopy

   npx wrangler queues create my-queue

2. Add bindings in config

   In wrangler.jsonc, declare the queue as a producer (so the Worker can send to it) and as a consumer (so the Worker is invoked to process batches).

   jsonccontent_copyCopy

   {
     "queues": {
       "producers": [
         { "queue": "my-queue", "binding": "MY_QUEUE" }
       ],
       "consumers": [
         {
           "queue": "my-queue",
           "max_batch_size": 10,
           "max_batch_timeout": 5
         }
       ]
     }
   }

3. Deploy

   Publish your Worker. Use wrangler tail to watch messages being consumed live.

   bashcontent_copyCopy

   npx wrangler deploy
   npx wrangler tail

export default {
  // PRODUCER: runs on each HTTP request and sends a message
  async fetch(request, env, ctx) {
    await env.MY_QUEUE.send({
      url: request.url,
      time: Date.now(),
    });
    return new Response("Queued!");
  },

  // CONSUMER: runs in the background with a batch of messages
  async queue(batch, env, ctx) {
    for (const message of batch.messages) {
      console.log("Processing:", message.body);
      // ...do the slow work here...
      message.ack(); // mark this message as done
    }
  },
};

Good limits to remember

• Up to 10,000 queues per account
• Max message size: 128 KB
• Max batch: 100 messages (or 256 KB total)
• Max batch wait: 60 seconds
• Throughput: up to 5,000 messages/second per queue
• Retries: up to 100; retention configurable up to 14 days