[CI/Test] Add expert parallelism coverage to external LB tests (#53497)
Signed-off-by: Tony Lin <tony.lin@intel.com> Co-authored-by: Kevin H. Luu <khluu000@gmail.com>
Tony Lin committed
Sep 4, 2026 at 15:56 UTC
9cd956c7e6cf54efa366b803cafa15ec6c2df827
1 file changed
+22
-3
tests/v1/distributed/test_external_lb_dp.py
+22
-3
@@ -133,11 +133,19 @@ def default_server_args():
133
]
134
135
136
-@pytest.fixture(scope="module", params=[1, 4])
136
+@pytest.fixture(
137
+ scope="module",
138
+ params=[(1, False), (4, False), (1, True)],
139
+ ids=["api-1-standard", "api-4-standard", "api-1-ep-enabled"],
140
+)
141
def server_manager(request, default_server_args):
138
- api_server_count = request.param
142
+ api_server_count, enable_expert_parallel = request.param
143
+ server_args = default_server_args.copy()
144
+ if enable_expert_parallel:
145
+ server_args.append("--enable-expert-parallel")
146
+
147
server_manager = ExternalLBServerManager(
140
- MODEL_NAME, DP_SIZE, api_server_count, default_server_args
148
+ MODEL_NAME, DP_SIZE, api_server_count, server_args
149
)
150
151
with server_manager:
@@ -167,10 +175,21 @@ def _get_parallel_config(server: RemoteOpenAIServer):
175
return vllm_config["parallel_config"]
176
177
178
+def _assert_expert_parallel_config(
179
+ servers: list[tuple[RemoteOpenAIServer, list[str]]],
180
+):
181
+ for server, server_args in servers:
182
+ expected = "--enable-expert-parallel" in server_args
183
+ parallel_config = _get_parallel_config(server)
184
+ assert parallel_config["enable_expert_parallel"] is expected
185
+
186
+
187
def test_external_lb_server_info(server_manager):
188
servers = server_manager.servers
189
api_server_count = server_manager.api_server_count
190
191
+ _assert_expert_parallel_config(servers)
192
+
193
for i, (server, _) in enumerate(servers):
194
print(f"Testing {i=}")
195