How does your team advise for scaling concurrency in Agents? My LocalAgent crashes once too many flows are submitted and concurrently running (roughly 300) -- and I believe this is because too many subprocesses are created, causing my computer to kill the parent process.

This is a big issue for me as well (using the docker agent each flow loads an entire docker container, which means if you have more than a handful you run out of ram and CPU). Id recommend showing support for this issue: https://github.com/PrefectHQ/server/pull/90 which is requesting concurrency limits.

Ah nice. I'll post something in there as well. Thank you!

If you manage to find an interim solution or workaround, please let me know, Ive been butting heads with this for a while!

I have a fix for LocalAgent and will write that up for you now!

samreay 🙂

Alright so I wrote up a comment, but after reading the rest of that issue, I'm not sure it applies there. That post is about concurrent flow runs managed by the Server/Cloud, whereas we are talking about managing flow runs for a given Agent.

Awesome, good idea there, lets see if we can get a dev feedback soon because I would dearly like it if I could trust my environment to be stable! @Zanie, any thoughts on this proposed lightweight solution?

Hey guys, I'll aim to talk to the team about this today and get back to you.