Describe the bug

Running bench update --reset --no-backup fails during the supervisor reload step.
The failure occurs when bench attempts to restart the erp-workers supervisor group, which does not exist on this system. This results in ERROR (no such file) for all worker processes and causes the entire update command to exit with a non-zero status.

This happens even though:

• App updates
• Migrations
• Asset builds
  all complete successfully.

To Reproduce

Steps to reproduce the behavior:

1. Use a bench setup without ERPNext (Frappe only).

2. Switch Frappe branch from develop to version-16:

   bench switch-branch version-16 --update

3. Run:

   bench update --reset --no-backup

4. Observe failure during supervisor reload.

Expected behavior

bench update should:

• Skip restarting supervisor groups that do not exist, or
• Treat missing supervisor groups as a warning instead of a fatal error.

The update process should complete successfully if core steps (migrate, build assets, etc.) succeed.

Actual behavior

The update fails with the following error:

erp-workers:erp-frappe-schedule: ERROR (no such file)
erp-workers:erp-frappe-short-worker-0: ERROR (no such file)
erp-workers:erp-frappe-long-worker-0: ERROR (no such file)

bench.exceptions.CommandFailedError: sudo supervisorctl restart erp-workers:

OS

• Linux: Debian GNU/Linux 12 (bookworm)

Version Information

• Bench Version: 5.29.0

• Bench Branch: PyPI (latest)

• Frappe Version: v16.1.1

• ERPNext Version: N/A

• Custom Apps:

  • eef

Additional context

• This bench setup does not include ERPNext.
• Supervisor web processes (erp-web) restart successfully.
• The erp-workers supervisor group does not exist on this system.
• No manual modifications were made to supervisor configuration.
• The issue appeared after switching the Frappe branch from develop to version-16.

Possible solution

• Check for the existence of a supervisor group before attempting to restart it.
• Handle missing supervisor groups gracefully (warning instead of error).
• Make supervisor restart failures non-fatal during bench update.