biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[CI][Bugfix] Fix flaky tests/v1/kv_connector/unit/test_multi_connector.py::test_multi_example_connector_consistency (#33555)

Browse Source 
Signed-off-by: NickLucche <nlucches@redhat.com>
This commit is contained in:
Nicolò Lucchesi
2026-02-02 12:01:29 +01:00
committed by GitHub
parent a502831d36
commit 528b3076af
1 changed files with 4 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
tests/v1/kv_connector/unit/test_multi_connector.py
View File

@@ -151,7 +151,8 @@ def test_multi_example_connector_consistency():
kv_transfer_config=kv_transfer_config,
)
# Run generation - this should trigger saving KV cache
_ = llm.generate(PROMPTS, SAMPLING_PARAMS)
# Use a single prompt to avoid race conditions depending on the order of scheduling
_ = llm.generate(PROMPTS[0], SAMPLING_PARAMS)
# --- Verification ---
@@ -221,7 +222,7 @@ def test_multi_example_connector_consistency():
# Run generation again - this should trigger loading from the first
# connector.
_ = llm.generate(PROMPTS, SAMPLING_PARAMS)
_ = llm.generate(PROMPTS[1], SAMPLING_PARAMS)
events = get_connector_events()
# get_num_new_matched_tokens will return new tokens from the first
@@ -247,7 +248,7 @@ def test_multi_example_connector_consistency():
# Run generation again - this should trigger loading from the first
# connector.
_ = llm.generate(PROMPTS, SAMPLING_PARAMS)
_ = llm.generate(PROMPTS[0], SAMPLING_PARAMS)
events = get_connector_events()
# get_num_new_matched_tokens will be called for both connectors but will