Diagnosing a crashed / hung deployed app

Every template installs a zero-dependency diagnostics bootstrap at startup (app_diagnostics.py for Python, diagnostics.ts for Node) that routes crash, unhandled-exception, and shutdown stack traces to stderr — the app's log stream. When an app crashes or hangs, read those logs to find the cause.

Reading the logs

On SSO workspaces (e.g. dogfood staging) the databricks apps logs CLI is blocked (websocket 302). Read the app's own log viewer in the browser instead:

https://<app-url>/logz

Open it while signed in (SSO). You'll see the stdout/stderr stream, including the diagnostics records. On a non-SSO workspace the CLI works: