Skip to content

LLM inference in C/C++

C++ ggml

tests : speed-up server test suite 3x (#26734)

Georgi Gerganov1621a3d388b7
diff --git a/tools/server/tests/conftest.py b/tools/server/tests/conftest.py
index c7ed775968..5dfde40796 100644
--- a/tools/server/tests/conftest.py
+++ b/tools/server/tests/conftest.py
@@ -15,7 +15,7 @@ def stop_server_after_each_test():
server.stop()
[email protected](scope="module", autouse=True)
-def do_something():
[email protected](scope="session", autouse=True)
+def load_server_presets():
# this will be run once per test session, before any tests
ServerPreset.load_all()
diff --git a/tools/server/tests/tests.sh b/tools/server/tests/tests.sh
index 709b5841aa..8d6681193d 100755
--- a/tools/server/tests/tests.sh
+++ b/tools/server/tests/tests.sh
@@ -14,10 +14,10 @@ fi
if [ $# -lt 1 ]
then
if [[ "${SLOW_TESTS:-0}" == 1 ]]; then
- pytest -v -x
+ pytest --durations=30 -v -x
else
- pytest -v -x -m "not slow"
+ pytest --durations=30 -v -x -m "not slow"
fi
else
- pytest "$@"
+ pytest --durations=30 "$@"
fi
diff --git a/tools/server/tests/unit/test_router.py b/tools/server/tests/unit/test_router.py
index c503ee3420..5ab62666ce 100644
--- a/tools/server/tests/unit/test_router.py
+++ b/tools/server/tests/unit/test_router.py
@@ -85,7 +85,7 @@ def _wait_for_model_status(model_id: str, desired: set[str], timeout: int = 60)
last_status = _get_model_status(model_id)
if last_status in desired:
return last_status
- time.sleep(1)
+ time.sleep(0.01)
raise AssertionError(
f"Timed out waiting for {model_id} to reach {desired}, last status: {last_status}"
)
@@ -460,7 +460,7 @@ def _wait_for_sse_event(collected: list, event_type: str, model: str, timeout: i
while time.time() < deadline:
if any(e.get("event") == event_type and e.get("model") == model for e in collected):
return True
- time.sleep(0.5)
+ time.sleep(0.01)
return False
diff --git a/tools/server/tests/utils.py b/tools/server/tests/utils.py
index 8f3a986424..3416f0bfde 100644
--- a/tools/server/tests/utils.py
+++ b/tools/server/tests/utils.py
@@ -306,6 +306,7 @@ class ServerProcess:
# wait for server to start
start_time = time.time()
+ last_print_time = start_time
while time.time() - start_time < timeout_seconds:
try:
response = self.make_request("GET", "/health", headers={
@@ -320,8 +321,10 @@ class ServerProcess:
if self.process.poll() is not None:
raise RuntimeError(f"Server process died with return code {self.process.returncode}")
- print(f"Waiting for server to start...")
- time.sleep(0.5)
+ if time.time() - last_print_time >= 1.0:
+ print(f"Waiting for server to start...")
+ last_print_time = time.time()
+ time.sleep(0.01)
raise TimeoutError(f"Server did not start within {timeout_seconds} seconds")
def stop(self) -> None: