Add configurable perf_record_delay for profiling during benchmarks

packages/ucache_bench/run.py

@@ -25,6 +25,7 @@
 import os
 import pathlib
 import subprocess
+import threading
 from typing import List, Optional
 
 
@@ -144,6 +145,19 @@ def run_cmd(
         return ""
 
 
+def profile_server() -> None:
+    """Record a system-wide perf profile for 5 seconds.
+
+    Skips if perf.data already exists.
+    """
+    if os.path.exists("perf.data"):
+        return
+    print("DCPERF_PERF_RECORD=1, starting perf record...")
+    subprocess.run(
+        ["perf", "record", "-a", "-g", "-o", "perf.data", "--", "sleep", "5"]
+    )
+
+
 def run_server(args: argparse.Namespace) -> None:
     """Run the UcacheBench server.
 
@@ -255,9 +269,18 @@ def run_server(args: argparse.Namespace) -> None:
     if args.verbose:
         server_cmd.append("--verbose=true")
 
+    if "DCPERF_PERF_RECORD" in os.environ and os.environ["DCPERF_PERF_RECORD"] == "1":
+        delay = args.perf_record_delay
+        print(f"DCPERF_PERF_RECORD=1, will start perf record after {delay}s delay")
+        t_prof = threading.Timer(delay, profile_server)
+        t_prof.start()
+
     stdout = run_cmd(server_cmd, timeout=None, for_real=args.real)
     print(stdout)
 
+    if "DCPERF_PERF_RECORD" in os.environ and os.environ["DCPERF_PERF_RECORD"] == "1":
+        t_prof.cancel()
+
 
 def run_client(args: argparse.Namespace) -> None:
     """Run the UcacheBench client.
@@ -541,6 +564,15 @@ def init_parser() -> argparse.ArgumentParser:
         help="Timeout in seconds for waiting for clients (0 = no timeout)",
     )
 
+    # Profiling configuration
+    server_parser.add_argument(
+        "--perf-record-delay",
+        type=int,
+        default=120,
+        help="Delay in seconds before starting perf record (when DCPERF_PERF_RECORD=1). "
+        "Should account for client startup and warmup time.",
+    )
+
     server_parser.add_argument(
         "--verbose", action="store_true", help="Enable verbose logging"
     )

packages/ucache_bench/server/UcacheBenchServer.cpp

@@ -45,6 +45,11 @@ void UcacheBenchServer::setupCacheLib() {
   cacheConfig.setAccessConfig(
       {config_.hash_power, config_.hashtable_lock_power});
 
+  // Configure number of CacheLib shards if specified
+  if (config_.cachelib_num_shards > 0) {
+    cacheConfig.setNumShards(config_.cachelib_num_shards);
+  }
+
   // Generate alloc sizes (factor 1.25, min allocation size)
   // This provides a good distribution of allocation classes for cache items
   // Max alloc size increased to 64KB to support production traffic distribution