NAS device silently reverted a manually applied fleet update overnight

One NAS device's own container-management add-on carries a separate, per-container auto-update setting that operates independently of the fleet's own git-driven update tracking. It silently recreated a management agent from its original bundled template overnight, undoing a version bump that had been applied hours earlier through the correct process — no error, no notification, just quietly back to the old version by morning.

Fixed by disabling that specific auto-update setting for the affected container and re-applying the update. A sibling device was found to have an even broader version of the same setting (applied to every container on the host, not just one) and was patched narrowly for the same container while leaving the broader policy for a separate, deliberate decision.