From d90c8a0d9e588806cde863db2e3383d7195e43ce Mon Sep 17 00:00:00 2001
From: Ruslan Dautkhanov <dautkhanov@gmail.com>
Date: Fri, 22 May 2026 09:37:39 -0600
Subject: [PATCH] fix: XD cold-start scenario races JVM startup on CI runners
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The XD ``test_cold_start_scenario`` introduced in #12 used a blind
``time.sleep(0.25)`` followed by a direct ``JavaGateway()`` connect
attempt. On CodSpeed CI runners (and any host slower than a modern
laptop) the JVM hadn't yet bound its listen socket when the connect
fired, surfacing as ``ConnectionRefusedError: [Errno 111] Connection
refused`` and failing the whole codspeed workflow.

Switch XD's body to the existing ``fresh_jvm`` context manager with
``readiness_retries=5``. ``fresh_jvm`` polls connection readiness
with ``check_connection`` and only proceeds once the JVM is actually
accepting calls — same convention the rest of the test suite uses.
This also drops a bunch of try/finally boilerplate since
``fresh_jvm`` handles spawn/shutdown.

Locally on M-series + JDK 21: XD median ~355 ms (unchanged from
post-#12 baseline). The variance is higher in cold-start measurement
by nature; what matters is the elimination of hard failures on
slower hosts.

Co-authored-by: Isaac
---
 .../src/py4j/tests/perf/scenarios/macro.py    | 34 ++++++-------------
 1 file changed, 10 insertions(+), 24 deletions(-)

diff --git a/py4j-python/src/py4j/tests/perf/scenarios/macro.py b/py4j-python/src/py4j/tests/perf/scenarios/macro.py
index c37fcb92..d14373a7 100644
--- a/py4j-python/src/py4j/tests/perf/scenarios/macro.py
+++ b/py4j-python/src/py4j/tests/perf/scenarios/macro.py
@@ -478,30 +478,16 @@ class XD_FullColdStart(MacroScenario):
     iterations_per_round = 1
 
     def measure(self, gateway):
-        # Import inside measure() so the cost of import isn't paid
-        # repeatedly; jvm helpers are already imported by the fixture
-        # in any sane run.
-        from py4j.tests.perf.jvm import spawn_jvm, shutdown_jvm
-        process = spawn_jvm()
-        try:
-            # Brief sleep mirrors fresh_jvm's startup_sleep so we
-            # exercise the same path; without it on a fast machine the
-            # first connect attempt can race ahead of the listening
-            # socket.
-            import time
-            time.sleep(0.25)
-            gw = None
-            try:
-                gw = JavaGateway()
-                gw.jvm.java.lang.System.currentTimeMillis()
-            finally:
-                if gw is not None:
-                    try:
-                        gw.shutdown()
-                    except Exception:
-                        pass
-        finally:
-            shutdown_jvm(process, None)
+        # Delegate to fresh_jvm so JVM startup readiness is polled
+        # rather than blind-slept — CodSpeed CI runners are slower
+        # than a local laptop and the original 0.25 s sleep was
+        # racing the JVM's listening socket (causing
+        # ConnectionRefusedError under CI). fresh_jvm uses a retry
+        # loop matching the rest of the test suite's conventions and
+        # cleanly tears down the subprocess in __exit__.
+        from py4j.tests.perf.jvm import fresh_jvm
+        with fresh_jvm(readiness_retries=5) as gw:
+            gw.jvm.java.lang.System.currentTimeMillis()
 
 
 ALL_MACRO_CLASSES = [