pub async fn quiesce_for_reboot()Expand description
Settle the mesh before a deliberate reset: let queued transmissions finish, then force every frame-counter boundary to flash.
A reboot commanded over the mesh has two obligations the reset itself would otherwise destroy. The MAC acknowledgment of the frame that carried the command is sitting in the TX queue — a reset-class command is answered by that acknowledgment and nothing else, so resetting first leaves the administrator retrying into silence. And the RX replay boundary that admitted the command is normally persisted only every persist-block of frames — resetting inside the block re-opens the window, and the administrator’s retries of the very same command are then accepted again after boot, one reboot per retry.
The TX drain is bounded and best-effort — an unlucky channel loses the acknowledgment, which the sender’s retry ladder is built for. The counter flush is not optional and happens last, so any frame accepted while the drain waited is inside the persisted boundary too.
A node that never came up has accepted nothing and holds no acknowledgment; the call returns immediately.