A queue holds a list of work items waiting to be processed. Instead of one big run that does everything, you add items to a queue and let a process work through them one at a time. This is the right pattern for high-volume, repeatable work – invoices to process, records to update, messages to send – and it gives you retries, priorities and progress tracking out of the box.
1. Create a queue
- Go to Library › Queues and create a new queue.
- Give it a name and an optional description.
- Optionally set:
- Unique references – prevent the same item being added twice.
- Max retries – how many times a failed item is retried before it's left as failed.
- SLA window – how long an item should take, so you can track items that are running late.
2. Add items
Add items to the queue from the queue page, or have a process add them with the queue actions. Each item can carry a reference (your own identifier), a priority (high, normal or low), and optional defer and due dates so it's only worked at the right time.
3. How items are worked
A process picks up the next item, does the work, and marks it successful or failed. Failed items are retried up to the queue's limit. As this happens, the queue page shows live counts of New, In progress, Successful and Failed items, so you can see the backlog clear.
4. Queue variables
A queue can define its own variables that override the global ones while its items are being worked – handy when the same process needs slightly different settings per queue. Queue variables can include secret values that point at a stored credential.
5. Automatic dispatch with pools
A queue can be bound to a bot pool so that, whenever items arrive, the pool's bots automatically pick up the work. You choose the worker process and how many bots may work the queue at once.