I'm also using PostgreSQL as a queue of tasks for a side project and it has been OK---although it has barely enough load for me to say that it's "really OK" or

I'm also using PostgreSQL as a queue of tasks for a side project and it has been OK---although it has barely enough load for me to say that it's "really OK" or whether it will turn out to be a problem in the future. I'm convinced I can go this way for a really long time without having to rearchitect the system or add more bloat to it. Anyhow, in my case, PostgreSQL is there just there to persist task state, and the database does very little magic. In fact, I also have a SQLite-backed implementation that allows service unit tests to be fast. Most of the work around managing such queue is in-process (and in Rust) because I wanted to keep deployments and operations simple. If interested, I wrote this thing a few months ago on how it works: https://jmmv.dev/2023/06/iii-iv-task-queue.html https://jmmv.dev/2023/06/iii-iv-task-queue.html