prestart.sh scaling to multiple nodes

[totalhack]

Apologies if I'm just missing something, but I'm trying to understand how prestart.sh might scale to multiple nodes. It looks like it runs before the startup of the underlying fastapi container, but what is preventing it from running multiple times across multiple nodes? If a common database was in use I imagine that would cause issues with alembic or the initial_data.py script (on first startup anyway). Thanks for your help.

[haviduck]

im scaling celeryworker, running backend solo on a local volume so it only inits on first time startup. removing initials from celery and scaling it across x nodes in a multi host cluster. works like a charm.

you could also do persistent glusterfs volumes to combat the inits.

[tiangolo]

Hello there! Indeed this would have been a problem at some point. The project has gone through several refactors and currently targets deployments with a simple Docker Compose, so the prestart.sh would only be run once.

Given that, I'll now close this issue. ☕