Debug & Profiling

Lila includes an in-memory diagnostic engine (lila.core.debug). When debugging is enabled, a pure ASGI middleware records the last 100 requests in RAM (no disk I/O, no SQLite table) with their timing and status.

Enabling Debug Mode & Dashboard

The debugging features are only active when your application is in debug mode (debug=True, or DEBUG=True in .env).

The Debug Dashboard (/debug) is mounted only when you also pass debug_html=True during App initialization in main.py:

main.py 

from app.config import DEBUG
from lila.core.app import App

app = App(
    routes=[router],
    debug=DEBUG,        # bool, not the string "False"
    debug_html=DEBUG,   # mounts GET/DELETE /debug and /debug/
)

What gets tracked?

For every non-static request, DebugStore appends one entry to the in-memory history (collections.deque(maxlen=100)):

  • Time: local HH:MM:SS of the request.
  • Path & Method: endpoint accessed and HTTP method.
  • IP Address: client IP (or unknown).
  • Status Code: the HTTP response status.
  • Duration: total wall time in milliseconds.

RAM/CPU figures come from DebugStore.get_system_metrics(). Lila uses psutil when installed (pip install "lila-framework[metrics]") and native OS fallbacks otherwise; values are never invented.

The Debug Dashboard

With debug=True, debug_html=True, Lila mounts:

  • GET /debug (and /debug/) — HTML dashboard.
  • GET /debug/?fetch=true — JSON object with metrics and history (true, True or 1).
  • DELETE /debug/ — clears the in-memory history. Send the page CSRF token in X-CSRF-Token (or the _csrf cookie token).

The dashboard is not mounted when debug_html is false, and nothing is recorded when debug is off.

Fetching Data Programmatically

Hit the same endpoint with the fetch query parameter:

Terminal 

curl http://localhost:8000/debug/?fetch=true

Reading the Store in Code


from lila.core.debug import DebugStore

history = DebugStore.get_all()          # list[dict], newest first
metrics = DebugStore.get_system_metrics()
DebugStore.clear()                      # same as DELETE /debug/

Clearing Logs

Send a DELETE request to the dashboard URL (or call DebugStore.clear()):

Terminal 

curl -X DELETE http://localhost:8000/debug/