fix(gateway): 改进流式传输稳定性

- stream_pump 读取错误时记录完整错误链并输出 warn 日志
- hub body 转发从 try_send 改为 async send，支持背压避免丢帧
- tunnel_stale_timeout 默认值从 900ms 提升到 10s，减少误判过期

apps/aether-gateway/src/execution_runtime/stream_pump.rs

@@ -8,6 +8,7 @@ use async_stream::stream;
 use axum::body::Bytes;
 use base64::Engine as _;
 use futures_util::{Stream, StreamExt};
+use tracing::warn;
 
 use crate::execution_runtime::ndjson::encode_stream_frame_ndjson;
 use crate::execution_runtime::DirectUpstreamStreamExecution;
@@ -87,13 +88,22 @@ pub(crate) fn build_direct_execution_frame_stream(
                     }
                 }
                 Err(err) => {
+                    let message = format_error_chain(&err);
+                    warn!(
+                        event_name = "stream_pump_body_read_error",
+                        log_type = "ops",
+                        status_code,
+                        upstream_bytes,
+                        error = %message,
+                        "upstream body stream read error"
+                    );
                     let frame = StreamFrame {
                         frame_type: StreamFrameType::Error,
                         payload: StreamFramePayload::Error {
                             error: ExecutionError {
                                 kind: ExecutionErrorKind::Internal,
                                 phase: ExecutionPhase::StreamRead,
-                                message: err.to_string(),
+                                message,
                                 upstream_status: Some(status_code),
                                 retryable: false,
                                 failover_recommended: false,
@@ -136,6 +146,17 @@ pub(crate) fn build_direct_execution_frame_stream(
     }
 }
 
+fn format_error_chain(err: &(dyn std::error::Error + 'static)) -> String {
+    let mut message = err.to_string();
+    let mut source = err.source();
+    while let Some(cause) = source {
+        message.push_str(": ");
+        message.push_str(&cause.to_string());
+        source = cause.source();
+    }
+    message
+}
+
 #[cfg(test)]
 mod tests {
     use std::collections::BTreeMap;

apps/aether-gateway/src/tunnel/embedded/hub.rs

@@ -278,12 +278,13 @@ impl LocalStream {
         }
     }
 
-    fn push_body_chunk(&self, payload: Bytes) -> bool {
+    async fn push_body_chunk(&self, payload: Bytes) -> bool {
         if self.terminal.load(Ordering::Acquire) {
             return false;
         }
         self.body_tx
-            .try_send(LocalBodyEvent::Chunk(payload))
+            .send(LocalBodyEvent::Chunk(payload))
+            .await
             .is_ok()
     }
 
@@ -644,7 +645,7 @@ impl HubRouter {
                 self.route_response_headers(proxy_conn_id, header, data);
             }
             protocol::RESPONSE_BODY => {
-                self.route_response_body(proxy_conn_id, header, data);
+                self.route_response_body(proxy_conn_id, header, data).await;
             }
             protocol::STREAM_END => {
                 self.finish_proxy_stream(proxy_conn_id, header.stream_id);
@@ -720,7 +721,12 @@ impl HubRouter {
         }
     }
 
-    fn route_response_body(&self, proxy_conn_id: u64, header: protocol::FrameHeader, data: &[u8]) {
+    async fn route_response_body(
+        &self,
+        proxy_conn_id: u64,
+        header: protocol::FrameHeader,
+        data: &[u8],
+    ) {
         let Some(local_id) = self.lookup_local_stream(proxy_conn_id, header.stream_id) else {
             return;
         };
@@ -738,7 +744,7 @@ impl HubRouter {
             None => return,
         };
 
-        if !stream.push_body_chunk(Bytes::from(payload)) {
+        if !stream.push_body_chunk(Bytes::from(payload)).await {
             self.cancel_local_stream(local_id, "local relay response congested");
         }
     }

apps/aether-proxy/src/config.rs

@@ -55,7 +55,7 @@ pub const DEFAULT_LOG_RETENTION_DAYS: u64 = 7;
 pub const DEFAULT_LOG_MAX_FILES: usize = 30;
 pub const DEFAULT_TUNNEL_PING_INTERVAL_MS: u64 = 250;
 pub const DEFAULT_TUNNEL_CONNECT_TIMEOUT_MS: u64 = 800;
-pub const DEFAULT_TUNNEL_STALE_TIMEOUT_MS: u64 = 900;
+pub const DEFAULT_TUNNEL_STALE_TIMEOUT_MS: u64 = 10_000;
 pub const DEFAULT_TUNNEL_SCALE_CHECK_INTERVAL_MS: u64 = 1_000;
 pub const DEFAULT_TUNNEL_SCALE_UP_THRESHOLD_PERCENT: u32 = 70;
 pub const DEFAULT_TUNNEL_SCALE_DOWN_THRESHOLD_PERCENT: u32 = 35;