Method destroy_broken

components/scheduler/src/vm_session.rs:411–421

Disposes of a worker whose channel died: drains what's left of its messages and returns it as broken (the pool spawns a replacement). The drained events are returned so the guild handler can apply their side effects (db writes for acks etc.) before clearing claim state.

(mut self, pool: &VmWorkerPool)

    /// The drained events are returned so the guild handler can apply their
    /// side effects (db writes for acks etc.) before clearing claim state.
    pub fn destroy_broken(mut self, pool: &VmWorkerPool) -> Vec<SessionEvent> {
        let mut events = Vec::new();
        while let Ok(msg) = self.worker.rx.try_recv() {
            events.extend(self.handle_worker_msg(Some(msg)));
        }

        self.worker.session_state = None;
        pool.return_worker(self.worker, true);

        events
    }

    pub fn worker_id(&self) -> u64 {
        self.worker.worker_id