Begin adding prometheus metrics

Dockerfile

@@ -259,6 +259,9 @@ COPY /src/requirements.txt /opt/backstory/src/requirements.txt
 RUN pip install -r /opt/backstory/src/requirements.txt
 RUN pip install 'markitdown[all]' pydantic
 
+# Prometheus
+RUN pip install prometheus-client prometheus-fastapi-instrumentator
+
 SHELL [ "/bin/bash", "-c" ]
 
 RUN { \

docker-compose.yml

@@ -149,6 +149,23 @@ services:
     volumes:
       - ./cache:/root/.cache
 
+  prometheus:
+    image: prom/prometheus
+    container_name: prometheus
+    restart: "always"
+#    env_file:
+#      - .env
+#    devices:
+#      - /dev/dri:/dev/dri
+    ports:
+    - 9090:9090 # Prometheus
+    networks:
+      - internal
+    volumes:
+#      - ./prometheus.yml:/etc/prometheus/prometheus.yml
+      - ./cache/prometheus:/prometheus
+
+
 networks:
   internal:
     driver: bridge

src/server.py

@@ -37,6 +37,8 @@ try_import("uvicorn")
 try_import("numpy")
 try_import("umap")
 try_import("sklearn")
+try_import("prometheus_client")
+try_import("prometheus_fastapi_instrumentator")
 
 import ollama
 import requests
@@ -48,6 +50,8 @@ import uvicorn # type: ignore
 import numpy as np # type: ignore
 import umap # type: ignore
 from sklearn.preprocessing import MinMaxScaler # type: ignore
+from prometheus_client import Summary # type: ignore
+from prometheus_fastapi_instrumentator import Instrumentator # type: ignore
 
 from utils import (
     rag as Rag,
@@ -66,6 +70,8 @@ rags = [
     # { "name": "LKML", "enabled": False, "description": "Full associative data for entire LKML mailing list archive." },
 ]
 
+REQUEST_TIME = Summary('request_processing_seconds', 'Time spent processing request')
+
 system_message_old = f"""
 Launched on {datetime.now().isoformat()}.
  
@@ -204,7 +210,7 @@ def parse_args():
 # %%
 
 # %%
-def is_valid_uuid(value):
+def is_valid_uuid(value: str) -> bool:
     try:
         uuid_obj = uuid.UUID(value, version=4)
         return str(uuid_obj) == value
@@ -236,6 +242,8 @@ class WebServer:
 
     def __init__(self, llm, model=MODEL_NAME):
         self.app = FastAPI(lifespan=self.lifespan)
+        Instrumentator().instrument(self.app)
+        Instrumentator().expose(self.app)
         self.contexts = {}
         self.llm = llm
         self.model = model
@@ -658,11 +666,12 @@ class WebServer:
             return JSONResponse({"status": "healthy"})
 
         @self.app.get("/{path:path}")
-        async def serve_static(path: str):
+        async def serve_static(path: str, request: Request):
             full_path = os.path.join(defines.static_content, path)
             if os.path.exists(full_path) and os.path.isfile(full_path):
                 logger.info(f"Serve static request for {full_path}")
                 return FileResponse(full_path)
+            
             logger.info(f"Serve index.html for {path}")
             return FileResponse(os.path.join(defines.static_content, "index.html"))
         
@@ -785,6 +794,7 @@ class WebServer:
         logger.info(f"Context {context_id} is not yet loaded.")
         return self.load_or_create_context(context_id)
         
+    @REQUEST_TIME.time()
     async def generate_response(self, context : Context, agent : Agent, prompt : str, options: Tunables | None) -> AsyncGenerator[Message, None]:
         if not self.file_watcher:
             raise Exception("File watcher not initialized")

src/utils/setup_logging.py

@@ -26,6 +26,7 @@ def setup_logging(level=defines.logging_level) -> logging.Logger:
 
     # Now reduce verbosity for FastAPI, Uvicorn, Starlette
     for noisy_logger in ("uvicorn", "uvicorn.error", "uvicorn.access", "fastapi", "starlette"):
+    #for noisy_logger in ("starlette"):
         logging.getLogger(noisy_logger).setLevel(logging.WARNING)
     
     logger = logging.getLogger(__name__)