nodemon restart after file deletion does not restore application state

Direct answer

Nodemon does not provide a built‑in option to run user‑defined logic before it restarts the child process when a watched file is deleted.

Confirmed facts

1. When nodemon detects a deletion of a watched file it emits a restart event and immediately spawns a new child process.
2. The --exec flag replaces the application process with the command you give; it does not run a separate preparatory step.
3. The --signal flag can send SIGTERM/SIGINT to the child process, allowing graceful shutdown handling, but the signal is not tied specifically to the file‑deletion event.

Practical steps for this case

To run verification or backup restoration before a deletion‑triggered restart, use one of the following patterns that do not require fork­ing nodemon:

1. Wrapper script with --exec – Create a small script that performs the backup check and then launches your app. Start nodemon with nodemon --exec "./start‑app.sh" node app.js. The wrapper runs on every restart (including deletion events) and can exit with a non‑zero code to abort the restart if verification fails.
2. Signal‑based graceful shutdown – Configure nodemon with --signal SIGTERM. In your application, listen for process.on('SIGTERM', async () => { await verifyBackup(); process.exit(0); });. When nodemon restarts, it sends SIGTERM, giving your app a chance to run verification before exiting.
3. External file‑watcher – Use a separate watcher (e.g., chokidar or fs.watch) to run the backup script. On success, trigger nodemon’s restart by touching a dummy file or sending a signal (e.g., kill -SIGUSR1 $(pgrep -f nodemon)) if you have enabled a custom signal.

Missing diagnostic detail

Recommendations may vary with nodemon version. Please confirm the nodemon version you are using (e.g., nodemon -v) so we can verify that the --signal behavior described above applies to your release.