Graceful Shutdown

The framework supports graceful shutdown, ensuring all connections, timers, and state are cleaned up before the process exits.

---

app.stop() — Stopping the App

Triggers a graceful shutdown of all services.

app.stop();

Shutdown Sequence

Step	What Happens
1	Timers cancelled — All internal timers (metrics, dashboard, gossip, Raft) are stopped.
2	Task registry stops — No new scheduled tasks are created. Pending tasks are cancelled.
3	Listeners close — TCP and HTTP listeners stop accepting new connections.
4	Connections drain — All active peer-to-peer and client connections are closed gracefully via their strands.
5	State cleanup — Registries are cleared, pools are stopped, WAL is flushed.

---

Signal Handling

The framework automatically registers handlers for SIGINT and SIGTERM that call app.stop(). You do not need to handle signals manually.

---

wait_for_registries_clear(timeout) — Verifying Clean Shutdown

After stopping, wait for all registries to be fully cleared.

using namespace framework;

bool all_clear = app.get_state()->wait_for_registries_clear(std::chrono::seconds(30));
if (!all_clear) {
    LOG_WARN_SVC(config, "Some connections did not close within the timeout");
}

Param	Description
timeout	Maximum time to wait for registries to clear.
Returns	true if all registries cleared, false if timeout was reached.

---

is_stopping() — Checking Shutdown State

if (app.get_state()->is_stopping()) {
    // The app is in the process of shutting down
}