Reduce workers from 6 to 3, should be more than enough with reduced r…

…esponsibilities (no /ingest)

ai_ta_backend/main.py

@@ -59,7 +59,8 @@ def index() -> Response:
   Returns:
       JSON: _description_
   """
-  response = jsonify({"Choo Choo": "Welcome to your Flask app 🚅"})
+  response = jsonify(
+      {"hi there, this is a 404": "Welcome to UIUC.chat backend 🚅 Read the docs here: https://docs.uiuc.chat/ "})
   response.headers.add('Access-Control-Allow-Origin', '*')
   return response
 

run.sh

@@ -5,4 +5,4 @@
 # 200 MB object store memory.. necessary to statically allocate or will crash in Railway env restrictions.
 # ray start --head --num-cpus 6 --object-store-memory 300000000
 export PYTHONPATH=${PYTHONPATH}:$(pwd)/ai_ta_backend
-exec gunicorn --workers=6 --threads=20000 --worker-class=gthread ai_ta_backend.main:app --timeout 1800
+exec gunicorn --workers=3 --threads=20000 --worker-class=gthread ai_ta_backend.main:app --timeout 1800