A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 09:45 UTC
[Bugfix][Failing Test] Fix test_events.py (#18460)
Signed-off-by: rabi <ramishra@redhat.com>
Rabi Mishra committed
May 21, 2025 at 17:27 UTC
61acfc45bcf44de6e5a14c82906bbb7a33940443
2 files changed
+6
-5
.buildkite/test-pipeline.yaml
+2
@@ -138,6 +138,7 @@ steps:
138
- vllm/core/
139
- tests/distributed/test_utils
140
- tests/distributed/test_pynccl
141
+ - tests/distributed/test_events
142
- tests/spec_decode/e2e/test_integration_dist_tp4
143
- tests/compile/test_basic_correctness
144
- examples/offline_inference/rlhf.py
@@ -156,6 +157,7 @@ steps:
157
- pytest -v -s distributed/test_utils.py
158
- pytest -v -s compile/test_basic_correctness.py
159
- pytest -v -s distributed/test_pynccl.py
160
+ - pytest -v -s distributed/test_events.py
161
- pytest -v -s spec_decode/e2e/test_integration_dist_tp4.py
162
# TODO: create a dedicated test section for multi-GPU example tests
163
# when we have multiple distributed example tests
tests/distributed/test_events.py
+4
-5
@@ -119,13 +119,12 @@ def test_topic_filtering(publisher_config):
119
"""
120
publisher_config.replay_endpoint = None
121
122
- cfg = publisher_config.model_copy()
123
- cfg.topic = "foo"
124
- pub = EventPublisherFactory.create(cfg)
122
+ publisher_config.topic = "foo"
123
+ pub = EventPublisherFactory.create(publisher_config)
124
125
from .conftest import MockSubscriber
127
- sub_foo = MockSubscriber(cfg.endpoint, None, "foo")
128
- sub_bar = MockSubscriber(cfg.endpoint, None, "bar")
126
+ sub_foo = MockSubscriber(publisher_config.endpoint, None, "foo")
127
+ sub_bar = MockSubscriber(publisher_config.endpoint, None, "bar")
128
129
try:
130
time.sleep(0.1)