Hey 👋 Whenever we try to open Dagit in a browser after a period of inactivity, it’ll hang for several minutes and we’ll get “This site can’t be reached … took too long to respond.” errors. It usually works after retrying a couple times — as if Dagit was somehow sleeping and the process needed to be restarted. We’re using ECS for other services and have never seen that before 🤔 Is this happening to anyone else?

Are there any clues in the dagit task’s logs?

Hmm not really — this task doesn’t product logs after starting and I can see it’s been running for 5+ days 🤔 We increased it to 2 Dagit tasks to see if it helped, but it hasn’t

is there a load balancer involved?

Yeah, it’s behind an ELB which was created during the docker compose ECS setup.

what versions of various libraries do you have installed? and/or do are you able to share the image you’re using? i can see if i can reproduce the hanging or at least see if we can improve the observability tooling provided out of the box

It doesn’t seem memory or cpu bound — we gave it 2GB memory and 1 CPU, both are well below 25% utilization. I’ll double check if I can share the image (no credentials or anything) — it’s also 1.8gb. We use Pipenv to manage dependencies. I’ve attached the Pipfiles if it’s helpful.