@tduncan / Nvidia-Optimization-Stack-VLLM / commits / d35eb6c440

[CI] Exclude KV-connector subtree from broad source dependencies (#51046)

Signed-off-by: NickLucche <nicolo.lucchesi@mistral.ai>

Nicolò Lucchesi committed Aug 7, 2026 at 01:51 UTC d35eb6c44071ea806018841c490f0d2f3219c485

13 files changed +59

.buildkite/test_areas/basic_correctness.yaml

@@ -8,6 +8,7 @@ steps:

8 device: h200_18gb

9 source_file_dependencies:

10 - vllm/

11 + - "!vllm/distributed/kv_transfer/"

12 - tests/basic_correctness/test_basic_correctness

13 - tests/basic_correctness/test_cpu_offload

14 - tests/basic_correctness/test_mem.py

.buildkite/test_areas/benchmarks.yaml

@@ -8,6 +8,7 @@ steps:

8 device: h200_18gb

9 source_file_dependencies:

10 - vllm/

11 + - "!vllm/distributed/kv_transfer/"

12 - tests/benchmarks/

13 commands:

14 - pytest -v -s benchmarks/

.buildkite/test_areas/distributed.yaml

+12

@@ -9,6 +9,7 @@ steps:

9 num_devices: 2

10 source_file_dependencies:

11 - vllm/distributed

12 + - "!vllm/distributed/kv_transfer/"

13 - tests/distributed

14 commands:

15 - pytest -v -s distributed/test_comm_ops.py

@@ -23,6 +24,7 @@ steps:

24 num_devices: 2

25 source_file_dependencies:

26 - vllm/distributed/

27 + - "!vllm/distributed/kv_transfer/"

28 - vllm/engine/

29 - vllm/executor/

30 - vllm/worker/worker_base.py

@@ -46,6 +48,7 @@ steps:

48 - image-build-amd

49 source_file_dependencies:

50 - vllm/distributed/

51 + - "!vllm/distributed/kv_transfer/"

52 - vllm/engine/

53 - vllm/executor/

54 - vllm/worker/worker_base.py

@@ -63,6 +66,7 @@ steps:

66 source_file_dependencies:

67 - vllm/compilation/

68 - vllm/distributed/

69 + - "!vllm/distributed/kv_transfer/"

70 - vllm/engine/

71 - vllm/executor/

72 - vllm/worker/worker_base.py

@@ -83,6 +87,7 @@ steps:

87 num_devices: 2

88 source_file_dependencies:

89 - vllm/distributed/

90 + - "!vllm/distributed/kv_transfer/"

91 - vllm/engine/

92 - vllm/executor/

93 - vllm/worker/worker_base.py

@@ -106,6 +111,7 @@ steps:

111 num_devices: 4

112 source_file_dependencies:

113 - vllm/distributed/

114 + - "!vllm/distributed/kv_transfer/"

115 - tests/distributed/test_torchrun_example.py

116 - tests/distributed/test_torchrun_example_moe.py

117 - examples/rl/

@@ -138,6 +144,7 @@ steps:

144 num_devices: 4

145 source_file_dependencies:

146 - vllm/distributed/

147 + - "!vllm/distributed/kv_transfer/"

148 - tests/v1/distributed

149 - tests/v1/engine/test_engine_core_client.py

150 - tests/distributed/test_utils

@@ -159,6 +166,7 @@ steps:

166 num_devices: 4

167 source_file_dependencies:

168 - vllm/distributed/

169 + - "!vllm/distributed/kv_transfer/"

170 - tests/distributed/test_pynccl

171 - tests/distributed/test_events

172 - tests/distributed/test_symm_mem_allreduce.py

@@ -182,6 +190,7 @@ steps:

190 - examples/features/torchrun/torchrun_dp_example_offline.py

191 - vllm/config/parallel.py

192 - vllm/distributed/

193 + - "!vllm/distributed/kv_transfer/"

194 - vllm/v1/engine/llm_engine.py

195 - vllm/v1/executor/uniproc_executor.py

196 - vllm/v1/worker/gpu_worker.py

@@ -200,6 +209,7 @@ steps:

209 num_devices: 4

210 source_file_dependencies:

211 - vllm/

212 + - "!vllm/distributed/kv_transfer/"

213 commands:

214 # NOTE: don't test llama model here, it seems hf implementation is buggy

215 # see https://github.com/vllm-project/vllm/pull/5689 for details

@@ -247,6 +257,7 @@ steps:

257 optional: true # TODO: revert once infra issue solved

258 source_file_dependencies:

259 - vllm/distributed/

260 + - "!vllm/distributed/kv_transfer/"

261 - vllm/engine/

262 - vllm/executor/

263 - vllm/model_executor/models/

@@ -262,6 +273,7 @@ steps:

273 num_devices: 4

274 source_file_dependencies:

275 - vllm/distributed/

276 + - "!vllm/distributed/kv_transfer/"

277 - vllm/engine/

278 - vllm/executor/

279 - vllm/model_executor/models/

.buildkite/test_areas/engine.yaml

@@ -98,6 +98,7 @@ steps:

98 - vllm/compilation/

99 - vllm/config/

100 - vllm/distributed/

101 + - "!vllm/distributed/kv_transfer/"

102 - vllm/engine/

103 - vllm/envs.py

104 - vllm/forward_context.py

.buildkite/test_areas/entrypoints.yaml

@@ -22,6 +22,7 @@ steps:

22 working_dir: "/vllm-workspace/tests"

23 source_file_dependencies:

24 - vllm/

25 + - "!vllm/distributed/kv_transfer/"

26 - tests/entrypoints/llm

27 commands:

28 - export VLLM_WORKER_MULTIPROC_METHOD=spawn

@@ -43,6 +44,7 @@ steps:

44 working_dir: "/vllm-workspace/tests"

45 source_file_dependencies:

46 - vllm/

47 + - "!vllm/distributed/kv_transfer/"

48 - tests/entrypoints/serve

49 - tests/entrypoints/scale_out

50 commands:

@@ -65,6 +67,7 @@ steps:

67 working_dir: "/vllm-workspace/tests"

68 source_file_dependencies:

69 - vllm/

70 + - "!vllm/distributed/kv_transfer/"

71 - tests/entrypoints/openai

72 - tests/entrypoints/test_chat_utils

73 commands:

@@ -85,6 +88,7 @@ steps:

88 working_dir: "/vllm-workspace/tests"

89 source_file_dependencies:

90 - vllm/

91 + - "!vllm/distributed/kv_transfer/"

92 - tests/entrypoints/openai

93 - tests/entrypoints/test_chat_utils

94 commands:

@@ -106,6 +110,7 @@ steps:

110 working_dir: "/vllm-workspace/tests"

111 source_file_dependencies:

112 - vllm/

113 + - "!vllm/distributed/kv_transfer/"

114 - tests/tool_use

115 - tests/entrypoints/tool_parsers

116 - tests/entrypoints/anthropic

@@ -130,6 +135,7 @@ steps:

135 working_dir: "/vllm-workspace/tests"

136 source_file_dependencies:

137 - vllm/

138 + - "!vllm/distributed/kv_transfer/"

139 - tests/entrypoints/openai/responses

140 commands:

141 - pytest -v -s entrypoints/openai/responses

@@ -141,6 +147,7 @@ steps:

147 working_dir: "/vllm-workspace/tests"

148 source_file_dependencies:

149 - vllm/

150 + - "!vllm/distributed/kv_transfer/"

151 - tests/entrypoints/speech_to_text

152 commands:

153 - export VLLM_WORKER_MULTIPROC_METHOD=spawn

@@ -153,6 +160,7 @@ steps:

160 working_dir: "/vllm-workspace/tests"

161 source_file_dependencies:

162 - vllm/

163 + - "!vllm/distributed/kv_transfer/"

164 - tests/entrypoints/multimodal

165 commands:

166 - export VLLM_WORKER_MULTIPROC_METHOD=spawn

@@ -165,6 +173,7 @@ steps:

173 working_dir: "/vllm-workspace/tests"

174 source_file_dependencies:

175 - vllm/

176 + - "!vllm/distributed/kv_transfer/"

177 - tests/entrypoints/pooling

178 commands:

179 - export VLLM_WORKER_MULTIPROC_METHOD=spawn

.buildkite/test_areas/expert_parallelism.yaml

@@ -47,6 +47,7 @@ steps:

47 num_devices: 4

48 source_file_dependencies:

49 - vllm/distributed/

50 + - "!vllm/distributed/kv_transfer/"

51 - vllm/engine/

52 - vllm/executor/

53 - vllm/compilation/

.buildkite/test_areas/misc.yaml

@@ -9,6 +9,7 @@ steps:

9 source_file_dependencies:

10 - vllm/config/

11 - vllm/distributed/

12 + - "!vllm/distributed/kv_transfer/"

13 - vllm/engine/

14 - vllm/inputs/

15 - vllm/logger.py

@@ -45,6 +46,7 @@ steps:

46 source_file_dependencies:

47 - vllm/config/

48 - vllm/distributed/

49 + - "!vllm/distributed/kv_transfer/"

50 - vllm/engine/

51 - vllm/entrypoints/pooling/

52 - vllm/inputs/

@@ -101,6 +103,7 @@ steps:

103 source_file_dependencies:

104 - vllm/config/

105 - vllm/distributed/

106 + - "!vllm/distributed/kv_transfer/"

107 - vllm/engine/

108 - vllm/inputs/

109 - vllm/lora/

@@ -161,6 +164,7 @@ steps:

164 source_file_dependencies:

165 - vllm/config/

166 - vllm/distributed/

167 + - "!vllm/distributed/kv_transfer/"

168 - vllm/engine/

169 - vllm/inputs/

170 - vllm/model_executor/

@@ -228,6 +232,7 @@ steps:

232 source_file_dependencies:

233 - vllm/config/

234 - vllm/distributed/

235 + - "!vllm/distributed/kv_transfer/"

236 - vllm/engine/

237 - vllm/inputs/

238 - vllm/model_executor/

@@ -287,6 +292,7 @@ steps:

292 - vllm/assets/

293 - vllm/config/

294 - vllm/distributed/

295 + - "!vllm/distributed/kv_transfer/"

296 - vllm/engine/

297 - vllm/inputs/

298 - vllm/model_executor/

.buildkite/test_areas/models_basic.yaml

@@ -8,6 +8,7 @@ steps:

8 device: h200_18gb

9 source_file_dependencies:

10 - vllm/

11 + - "!vllm/distributed/kv_transfer/"

12 - tests/models/test_initialization.py

13 - tests/models/registry.py

14 commands:

@@ -35,6 +36,7 @@ steps:

36 timeout_in_minutes: 35

37 source_file_dependencies:

38 - vllm/

39 + - "!vllm/distributed/kv_transfer/"

40 - tests/models/test_terratorch.py

41 - tests/models/transformers/test_backend.py

42 - tests/models/test_registry.py

@@ -82,6 +84,7 @@ steps:

84 timeout_in_minutes: 20

85 source_file_dependencies:

86 - vllm/

87 + - "!vllm/distributed/kv_transfer/"

88 - tests/models/test_utils.py

89 - tests/models/test_vision.py

90 - tests/models/test_adapters.py

.buildkite/test_areas/models_language.yaml

@@ -8,6 +8,7 @@ steps:

8 device: h200_18gb

9 source_file_dependencies:

10 - vllm/

11 + - "!vllm/distributed/kv_transfer/"

12 - tests/models/language

13 commands:

14 # Test standard language models, excluding a subset of slow tests

@@ -60,6 +61,7 @@ steps:

61 timeout_in_minutes: 65

62 source_file_dependencies:

63 - vllm/

64 + - "!vllm/distributed/kv_transfer/"

65 - tests/models/language/generation

66 commands:

67 # Install fast path packages for testing against transformers

@@ -88,6 +90,7 @@ steps:

90 timeout_in_minutes: 65

91 source_file_dependencies:

92 - vllm/

93 + - "!vllm/distributed/kv_transfer/"

94 - tests/models/language/generation

95 commands:

96 - uv pip install --system --no-build-isolation 'git+https://github.com/state-spaces/mamba@v2.3.0'

@@ -101,6 +104,7 @@ steps:

104 optional: true

105 source_file_dependencies:

106 - vllm/

107 + - "!vllm/distributed/kv_transfer/"

108 - tests/models/language/generation

109 commands:

110 # Install fast path packages for testing against transformers

@@ -115,6 +119,7 @@ steps:

119 optional: true

120 source_file_dependencies:

121 - vllm/

122 + - "!vllm/distributed/kv_transfer/"

123 - tests/models/language/generation_ppl_test

124 commands:

125 - pytest -v -s models/language/generation_ppl_test

@@ -126,6 +131,7 @@ steps:

131 optional: true

132 source_file_dependencies:

133 - vllm/

134 + - "!vllm/distributed/kv_transfer/"

135 - tests/models/language/pooling

136 commands:

137 - pytest -v -s models/language/pooling -m 'not core_model'

@@ -144,6 +150,7 @@ steps:

150 optional: true

151 source_file_dependencies:

152 - vllm/

153 + - "!vllm/distributed/kv_transfer/"

154 - tests/models/language/pooling_mteb_test

155 commands:

156 - pytest -v -s models/language/pooling_mteb_test

.buildkite/test_areas/models_multimodal.yaml

+11

@@ -8,6 +8,7 @@ steps:

8 device: h200_18gb

9 source_file_dependencies:

10 - vllm/

11 + - "!vllm/distributed/kv_transfer/"

12 - tests/models/multimodal

13 commands:

14 - pytest -v -s models/multimodal/generation/test_common.py -m core_model -k "qwen2"

@@ -26,6 +27,7 @@ steps:

27 device: h200_18gb

28 source_file_dependencies:

29 - vllm/

30 + - "!vllm/distributed/kv_transfer/"

31 - tests/models/multimodal

32 commands:

33 - pytest -v -s models/multimodal/generation/test_common.py -m core_model -k "qwen3 or gemma"

@@ -45,6 +47,7 @@ steps:

47 timeout_in_minutes: 40

48 source_file_dependencies:

49 - vllm/

50 + - "!vllm/distributed/kv_transfer/"

51 - tests/models/multimodal

52 commands:

53 - pytest -v -s models/multimodal/generation/test_common.py -m core_model -k "not qwen2 and not qwen3 and not gemma"

@@ -62,6 +65,7 @@ steps:

65 timeout_in_minutes: 75

66 source_file_dependencies:

67 - vllm/

68 + - "!vllm/distributed/kv_transfer/"

69 - tests/models/multimodal

70 commands:

71 - pytest -v -s models/multimodal -m core_model --ignore models/multimodal/generation/test_common.py --ignore models/multimodal/generation/test_ultravox.py --ignore models/multimodal/generation/test_qwen2_5_vl.py --ignore models/multimodal/generation/test_qwen2_vl.py --ignore models/multimodal/generation/test_whisper.py --ignore models/multimodal/generation/test_mm_prefix_lm.py --ignore models/multimodal/generation/test_memory_leak.py --ignore models/multimodal/generation/test_vit_cudagraph.py --ignore models/multimodal/processing

@@ -83,6 +87,7 @@ steps:

87 timeout_in_minutes: 125

88 source_file_dependencies:

89 - vllm/

90 + - "!vllm/distributed/kv_transfer/"

91 - tests/models/multimodal

92 - tests/models/registry.py

93 device: cpu-medium

@@ -96,6 +101,7 @@ steps:

101 device: h200_18gb

102 source_file_dependencies:

103 - vllm/

104 + - "!vllm/distributed/kv_transfer/"

105 - tests/models/multimodal

106 - tests/models/registry.py

107 commands:

@@ -132,6 +138,7 @@ steps:

138 optional: true

139 source_file_dependencies:

140 - vllm/

141 + - "!vllm/distributed/kv_transfer/"

142 - tests/models/multimodal/generation

143 - tests/models/multimodal/test_mapping.py

144 commands:

@@ -168,6 +175,7 @@ steps:

175 optional: true

176 source_file_dependencies:

177 - vllm/

178 + - "!vllm/distributed/kv_transfer/"

179 - tests/models/multimodal/generation

180 commands:

181 - pytest -v -s models/multimodal/generation/test_common.py -m 'split(group=0) and not core_model'

@@ -178,6 +186,7 @@ steps:

186 optional: true

187 source_file_dependencies:

188 - vllm/

189 + - "!vllm/distributed/kv_transfer/"

190 - tests/models/multimodal/generation

191 commands:

192 - pytest -v -s models/multimodal/generation/test_common.py -m 'split(group=1) and not core_model'

@@ -188,6 +197,7 @@ steps:

197 device: h200_18gb

198 source_file_dependencies:

199 - vllm/

200 + - "!vllm/distributed/kv_transfer/"

201 - tests/models/multimodal/pooling

202 commands:

203 - pytest -v -s models/multimodal/pooling -m 'not core_model'

@@ -200,4 +210,5 @@ steps:

210 - image-build-amd

211 source_file_dependencies:

212 - vllm/

213 + - "!vllm/distributed/kv_transfer/"

214 - tests/models/multimodal/pooling

.buildkite/test_areas/pytorch.yaml

@@ -13,6 +13,7 @@ steps:

13 - vllm/compilation/

14 - vllm/config/

15 - vllm/distributed/

16 + - "!vllm/distributed/kv_transfer/"

17 - vllm/engine/

18 - vllm/env_override.py

19 - vllm/envs.py

@@ -53,6 +54,7 @@ steps:

54 - vllm/compilation/

55 - vllm/config/

56 - vllm/distributed/

57 + - "!vllm/distributed/kv_transfer/"

58 - vllm/engine/

59 - vllm/env_override.py

60 - vllm/envs.py

@@ -85,6 +87,7 @@ steps:

87 - vllm/compilation/

88 - vllm/config/

89 - vllm/distributed/

90 + - "!vllm/distributed/kv_transfer/"

91 - vllm/engine/

92 - vllm/env_override.py

93 - vllm/envs.py

@@ -118,6 +121,7 @@ steps:

121 - vllm/compilation/

122 - vllm/config/

123 - vllm/distributed/

124 + - "!vllm/distributed/kv_transfer/"

125 - vllm/engine/

126 - vllm/env_override.py

127 - vllm/envs.py

@@ -157,6 +161,7 @@ steps:

161 - vllm/compilation/

162 - vllm/config/

163 - vllm/distributed/

164 + - "!vllm/distributed/kv_transfer/"

165 - vllm/engine/

166 - vllm/env_override.py

167 - vllm/envs.py

.buildkite/test_areas/rust_frontend.yaml

@@ -102,6 +102,7 @@ steps:

102 source_file_dependencies:

103 - rust/

104 - vllm/distributed/

105 + - "!vllm/distributed/kv_transfer/"

106 - vllm/engine/

107 - vllm/executor/

108 - vllm/v1/engine/

.buildkite/test_areas/weight_loading.yaml

@@ -10,6 +10,7 @@ steps:

10 optional: true

11 source_file_dependencies:

12 - vllm/

13 + - "!vllm/distributed/kv_transfer/"

14 - tests/weight_loading

15 commands:

16 - bash weight_loading/run_model_weight_loading_test.sh -c weight_loading/models.txt