front: 'silent' wait-mode still keepalives (SSE comments), driven by the front

The point of the keepalive is to hold the client connection open while the engine
is stuck loading — so 'silent' must NOT fall back to the dead legacy path (which
sends nothing until the engine's first byte and lets short-timeout clients/proxies
disconnect). Now ALL streaming inference goes through the front keepalive path:

  * silent    — SSE comment lines (': …') only: keeps the socket alive, emits no
                chunk/content/status (invisible to event parsers).
  * invisible — empty-content chunk + x_queue_info (default).
  * visible   — visible status text.
  * thinking  — reasoning channel (when mode != silent).

The front stays responsive even when the engine is GIL-blocked, so it can drive
these regardless of engine state.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RdMufYvtTbtGDWsiZVoXce

codai/config.py

@@ -111,7 +111,10 @@ class ModelsConfig:
     #   "invisible" (default) — empty-content SSE chunk + x_queue_info metadata
     #                           (holds the connection; no message-content pollution)
     #   "visible"             — short visible status text (appears in the content)
-    #   "silent"              — send nothing (legacy behaviour)
+    #   "silent"              — keep the connection alive via SSE comments only
+    #                           (no chunk, no content, no status) — still prevents a
+    #                           client/proxy idle-timeout disconnect while the engine
+    #                           is stuck loading
     # When the request has thinking enabled the keepalive is sent on the reasoning
     # channel instead (no pollution), unless the mode is "silent". Global default;
     # override per-model via the models.json entry's "wait_status_mode".

codai/frontproxy/app.py

@@ -1208,6 +1208,11 @@ class FrontProxy:
         import time as _t
 
         def _ka(msg: str) -> bytes:
+            # "silent" still keeps the connection alive — but as an SSE COMMENT
+            # (a ':' line), which keeps the socket/stream open without emitting any
+            # chat.completion.chunk (no event for parsers, no content, no metadata).
+            if mode == "silent":
+                return (": " + msg + "\n\n").encode()
             if thinking:
                 delta = {"reasoning_content": msg}
             elif mode == "visible":
@@ -1248,7 +1253,6 @@ class FrontProxy:
                             await _asyncio.wait_for(_asyncio.shield(_acq), timeout=_KA)
                             break
                         except _asyncio.TimeoutError:
-                            if mode != "silent":
                             yield _ka("queued — waiting for a free slot")
                         except QueueFull:
                             yield (b'data: {"error":"Server busy: the generation '
@@ -1270,7 +1274,6 @@ class FrontProxy:
                         rp_resp = await self._long.send(rp_req, stream=True)
                     except Exception as exc:
                         if not _last:
-                            if mode != "silent":
                             yield _ka("engine starting up")
                             await _asyncio.sleep(_RETRY)
                             continue
@@ -1283,7 +1286,6 @@ class FrontProxy:
                             await rp_resp.aclose()
                         except Exception:
                             pass
-                        if mode != "silent":
                         yield _ka("model loading")
                         await _asyncio.sleep(_RETRY)
                         continue
@@ -1594,17 +1596,17 @@ class FrontProxy:
                 headers=dict(self._filter_headers(r.headers, _DROP_RESP)),
                 media_type=r.headers.get("content-type"))
 
-        # Streaming inference: emit a wait-keepalive (status / reasoning) while the
-        # front acquires a queue slot and the engine loads the model, so the client
-        # doesn't time out and disconnect. Mode is per-model / global
-        # (silent|invisible|visible); "silent" keeps the legacy silent path below.
+        # Streaming inference always goes through the keepalive path: the WHOLE point
+        # is to hold the connection open (from the front, which stays responsive even
+        # when the engine is stuck/GIL-blocked loading) while we acquire a queue slot
+        # and the engine loads the model. The mode only changes the payload —
+        # "silent" still keeps the socket alive (SSE comments), "invisible"/"visible"
+        # add status, thinking uses the reasoning channel.
         if (method == "POST" and _router.is_inference_path(path)
                 and body_bytes is not None and self._peek_stream(body_bytes)):
-            _mode = self._wait_status_mode(model)
-            if _mode != "silent":
             return await self._stream_with_keepalive(
-                    request, engine, path, body_bytes, model, _mode,
-                    self._peek_thinking(body_bytes))
+                request, engine, path, body_bytes, model,
+                self._wait_status_mode(model), self._peek_thinking(body_bytes))
 
         # Front-managed generation queue (text only). Acquire a per-model slot
         # before dispatching: if all max_instances slots are busy this awaits