[fix](trx-server): offload history persistence

src/trx-server/src/history_store.rs

@@ -226,20 +226,38 @@ pub fn flush_all(db: &mut PickleDb, rig_id: &str, histories: &Arc<DecoderHistori
     let _ = db.dump();
 }
 
+fn flush_all_rigs(db: &Mutex<PickleDb>, rig_histories: &[(String, Arc<DecoderHistories>)]) {
+    let Ok(mut guard) = db.lock() else {
+        tracing::warn!("history database mutex poisoned; skipping periodic flush");
+        return;
+    };
+    for (rig_id, histories) in rig_histories {
+        flush_all(&mut guard, rig_id, histories);
+    }
+}
+
 /// Spawn a Tokio task that flushes all rigs' histories to disk every 60 seconds.
+///
+/// Snapshot cloning, JSON serialization, and disk I/O run on Tokio's blocking
+/// pool so a large history database cannot stall an async runtime worker.
 pub fn spawn_flush_task(
     db: Arc<Mutex<PickleDb>>,
     rig_histories: Vec<(String, Arc<DecoderHistories>)>,
 ) {
     tokio::spawn(async move {
+        let rig_histories = Arc::new(rig_histories);
         let mut interval = tokio::time::interval(Duration::from_secs(60));
         interval.tick().await; // consume the immediate first tick
         loop {
             interval.tick().await;
-            if let Ok(mut guard) = db.lock() {
+            let db = Arc::clone(&db);
-                for (rig_id, histories) in &rig_histories {
+            let rig_histories = Arc::clone(&rig_histories);
-                    flush_all(&mut guard, rig_id, histories);
+            if let Err(err) = tokio::task::spawn_blocking(move || {
-                }
+                flush_all_rigs(&db, rig_histories.as_slice());
+            })
+            .await
+            {
+                tracing::warn!(error = %err, "history flush worker failed");
             }
         }
     });