# Concurrency limits for async service endpoints: how many callback jobs can run in parallel?

**URL:** <https://community.appfarm.io/t/concurrency-limits-for-async-service-endpoints-how-many-callback-jobs-can-run-in-parallel/2452>\
**Category:** Ask the community\
**Created:** [September 25, 2026, 11:55am UTC](https://community.appfarm.io/t/concurrency-limits-for-async-service-endpoints-how-many-callback-jobs-can-run-in-parallel/2452 "2026-09-25T11:55:56Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![michael](https://dub1.discourse-cdn.com/flex013/user_avatar/community.appfarm.io/michael/32/2883_2.png) [@michael](https://community.appfarm.io/u/michael)\
**Post date:** [September 25, 2026, 11:55am UTC](https://community.appfarm.io/t/concurrency-limits-for-async-service-endpoints-how-many-callback-jobs-can-run-in-parallel/2452/1 "2026-09-25T11:55:56Z")

</div>

Hi!

We need to set up a service that runs an external process, then returns data to us for processing.

**Our intended setup:** A scheduled service sends each project to an external API (fire-and-forget). The API processes it for 2–3 minutes and then posts the result back to a service endpoint in our solution. That endpoint has Run Async enabled and processes and saves the results, which takes about 5–20 minutes depending on the project’s size. We need to run this for about 600 projects per night.

If we send them between 00:00 and 06:00 this means one project every 36 seconds and 34 threads at max..

My question is: **how many concurrent threads can a service have before it starts throwing 429 errors?**

---

<div class="post-metadata">

**Author:** ![kristian](https://dub1.discourse-cdn.com/flex013/user_avatar/community.appfarm.io/kristian/32/58_2.png) [@kristian](https://community.appfarm.io/u/kristian)\
**Post date:** [September 25, 2026, 5:35pm UTC](https://community.appfarm.io/t/concurrency-limits-for-async-service-endpoints-how-many-callback-jobs-can-run-in-parallel/2452/2 "2026-09-25T17:35:14Z")

</div>

Hi!

Please read here about that behaviour. 2-3 in parallell is the default:

> **[Behaviour of Schedules - Service Schedules | Appfarm Documentation](https://docs.appfarm.io/reference/operations/schedules#behaviour-of-schedules)**
>
> Service schedules are used to automate services and run them at set times. Common use cases are running batch and recurring jobs such as synchronizing data with an external system, polling for changes, and sending notifications. | Create and manage...

This sounds like something to discuss with Appfarm. The resources and horizontal scaling may be increased.