We have jobs which call a a third-party API. This API seems to struggle if too many calls are fired at it too quickly. So I would like to limit the rate at which our jobs are processed, so as not to overwhelm the third-party server.
The problem is that all our jobs are scheduled in one big batch.
Any ideas how I might accomplish this?
I guess I could limit the number of worker threads to 1, but that would then limit the usefulness of HangFire for processing other kinds of jobs.
I’m also looking at ways of doing this, but one way I have tried is instead of Enquing all the jobs, I actually Schedule them with a staggered 15 seconds interval between each other. It kinda works, but I wish there was a more “official” way…
If you have a single server process, you can create an API manager shared between tasks, which will guard all API calls with a semaphore. So when the number of concurrent API requests reaches the limit, the next calls will wait. That won’t hinder other jobs’ processing, while limiting API access rate.