From 2c7f7fdaaee966d3291659f2158e7f18fa015a0f Mon Sep 17 00:00:00 2001
From: Gahow Wang <gahow.wang@gmail.com>
Date: Sat, 23 May 2026 21:02:26 +0800
Subject: [PATCH] replayer: restore optional max_inflight_sessions for
 backwards compat

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---
 replayer/replay.py | 34 +++++++++++++++++++++-------------
 1 file changed, 21 insertions(+), 13 deletions(-)

diff --git a/replayer/replay.py b/replayer/replay.py
index 13f336a..3da2a9d 100644
--- a/replayer/replay.py
+++ b/replayer/replay.py
@@ -59,6 +59,7 @@ class ReplayConfig:
     request_timeout_s: float = 600.0
     request_limit: int | None = None
     model_name: str = "default"
+    max_inflight_sessions: int | None = None  # cap on concurrent sessions; None = unlimited
 
 
 def _build_prompt_token_ids(req: TraceRequest) -> list[int]:
@@ -215,21 +216,28 @@ async def _run_session(
     earliest_ts: float,
     sweep_start: float,
     sink: IncrementalMetricSink,
+    session_sem: asyncio.Semaphore | None = None,
 ) -> list[RequestMetrics]:
-    for req in state.turns:
-        # Wait until this request's trace timestamp
-        target_wall = (req.timestamp_s - earliest_ts)
-        elapsed = time.perf_counter() - sweep_start
-        if elapsed < target_wall:
-            await asyncio.sleep(target_wall - elapsed)
+    if session_sem is not None:
+        await session_sem.acquire()
+    try:
+        for req in state.turns:
+            # Wait until this request's trace timestamp
+            target_wall = (req.timestamp_s - earliest_ts)
+            elapsed = time.perf_counter() - sweep_start
+            if elapsed < target_wall:
+                await asyncio.sleep(target_wall - elapsed)
 
-        token_ids = _build_prompt_token_ids(req)
-        metric = await _dispatch_request(
-            client=client, config=config, req=req,
-            prompt_token_ids=token_ids, sem=request_sem,
-        )
-        state.metrics.append(metric)
-        await sink.append(metric)
+            token_ids = _build_prompt_token_ids(req)
+            metric = await _dispatch_request(
+                client=client, config=config, req=req,
+                prompt_token_ids=token_ids, sem=request_sem,
+            )
+            state.metrics.append(metric)
+            await sink.append(metric)
+    finally:
+        if session_sem is not None:
+            session_sem.release()
 
     return state.metrics