[Docker][KVConnector] Build mooncake-transfer-engine from source (#42114)

Signed-off-by: Zhewen Li <zhewenli@inferact.ai>
Signed-off-by: khluu <khluu000@gmail.com>
Co-authored-by: Zhewen Li <zhewenli@inferact.ai>
Co-authored-by: khluu <khluu000@gmail.com>
This commit is contained in:
Zhewen Li
2026-05-16 00:26:25 -07:00
committed by GitHub
co-authored by Zhewen Li khluu
parent 32b7177909
commit 657b42b592
2 changed files with 41 additions and 0 deletions
+19
View File
@@ -7,6 +7,9 @@ env:
CUDA_ARCH_AARCH64: "8.0 8.7 8.9 9.0 10.0 11.0 12.0+PTX"
CUDA_ARCH_X86_CU129: "7.5 8.0 8.6 8.9 9.0 10.0 12.0"
CUDA_ARCH_AARCH64_CU129: "8.0 8.7 8.9 9.0 10.0 12.0"
MOONCAKE_WHEEL_AARCH64_2_35: "https://vllm-wheels.s3.amazonaws.com/mooncake/mooncake_transfer_engine-0.3.10.post2-0da9dfea3-cp312-cp312-manylinux_2_35_aarch64.whl"
MOONCAKE_WHEEL_AARCH64_2_39: "https://vllm-wheels.s3.amazonaws.com/mooncake/mooncake_transfer_engine-0.3.10.post2-0da9dfea3-cp312-cp312-manylinux_2_39_aarch64.whl"
MOONCAKE_WHEEL_X86_64: "https://vllm-wheels.s3.amazonaws.com/mooncake/mooncake_transfer_engine-0.3.10.post2-0da9dfea3-cp312-cp312-manylinux_2_35_x86_64.whl"
steps:
- input: "Provide Release version here"
@@ -136,6 +139,8 @@ steps:
--build-arg CUDA_VERSION=13.0.2 \
--build-arg torch_cuda_arch_list="${CUDA_ARCH_X86}" \
--build-arg INSTALL_KV_CONNECTORS=true \
--build-arg MOONCAKE_WHEEL_AARCH64="${MOONCAKE_WHEEL_AARCH64_2_35}" \
--build-arg MOONCAKE_WHEEL_X86_64="${MOONCAKE_WHEEL_X86_64}" \
--build-arg BUILD_BASE_IMAGE=nvidia/cuda:13.0.2-devel-ubuntu22.04 \
--target vllm-openai \
--progress plain \
@@ -162,6 +167,8 @@ steps:
--build-arg CUDA_VERSION=13.0.2 \
--build-arg torch_cuda_arch_list="${CUDA_ARCH_AARCH64}" \
--build-arg INSTALL_KV_CONNECTORS=true \
--build-arg MOONCAKE_WHEEL_AARCH64="${MOONCAKE_WHEEL_AARCH64_2_35}" \
--build-arg MOONCAKE_WHEEL_X86_64="${MOONCAKE_WHEEL_X86_64}" \
--build-arg BUILD_BASE_IMAGE=nvidia/cuda:13.0.2-devel-ubuntu22.04 \
--target vllm-openai \
--progress plain \
@@ -185,6 +192,8 @@ steps:
--build-arg CUDA_VERSION=12.9.1 \
--build-arg torch_cuda_arch_list="${CUDA_ARCH_X86_CU129}" \
--build-arg INSTALL_KV_CONNECTORS=true \
--build-arg MOONCAKE_WHEEL_AARCH64="${MOONCAKE_WHEEL_AARCH64_2_35}" \
--build-arg MOONCAKE_WHEEL_X86_64="${MOONCAKE_WHEEL_X86_64}" \
--target vllm-openai \
--progress plain \
-f docker/Dockerfile .
@@ -210,6 +219,8 @@ steps:
--build-arg CUDA_VERSION=12.9.1 \
--build-arg torch_cuda_arch_list="${CUDA_ARCH_AARCH64_CU129}" \
--build-arg INSTALL_KV_CONNECTORS=true \
--build-arg MOONCAKE_WHEEL_AARCH64="${MOONCAKE_WHEEL_AARCH64_2_35}" \
--build-arg MOONCAKE_WHEEL_X86_64="${MOONCAKE_WHEEL_X86_64}" \
--target vllm-openai \
--progress plain \
-f docker/Dockerfile .
@@ -234,6 +245,8 @@ steps:
--build-arg GDRCOPY_OS_VERSION=Ubuntu24_04 \
--build-arg torch_cuda_arch_list="${CUDA_ARCH_X86}" \
--build-arg INSTALL_KV_CONNECTORS=true \
--build-arg MOONCAKE_WHEEL_AARCH64="${MOONCAKE_WHEEL_AARCH64_2_39}" \
--build-arg MOONCAKE_WHEEL_X86_64="${MOONCAKE_WHEEL_X86_64}" \
--build-arg BUILD_BASE_IMAGE=nvidia/cuda:13.0.2-devel-ubuntu24.04 \
--target vllm-openai \
--progress plain \
@@ -261,6 +274,8 @@ steps:
--build-arg GDRCOPY_OS_VERSION=Ubuntu24_04 \
--build-arg torch_cuda_arch_list="${CUDA_ARCH_AARCH64}" \
--build-arg INSTALL_KV_CONNECTORS=true \
--build-arg MOONCAKE_WHEEL_AARCH64="${MOONCAKE_WHEEL_AARCH64_2_39}" \
--build-arg MOONCAKE_WHEEL_X86_64="${MOONCAKE_WHEEL_X86_64}" \
--build-arg BUILD_BASE_IMAGE=nvidia/cuda:13.0.2-devel-ubuntu24.04 \
--target vllm-openai \
--progress plain \
@@ -286,6 +301,8 @@ steps:
--build-arg GDRCOPY_OS_VERSION=Ubuntu24_04 \
--build-arg torch_cuda_arch_list="${CUDA_ARCH_X86_CU129}" \
--build-arg INSTALL_KV_CONNECTORS=true \
--build-arg MOONCAKE_WHEEL_AARCH64="${MOONCAKE_WHEEL_AARCH64_2_39}" \
--build-arg MOONCAKE_WHEEL_X86_64="${MOONCAKE_WHEEL_X86_64}" \
--target vllm-openai \
--progress plain \
-f docker/Dockerfile .
@@ -312,6 +329,8 @@ steps:
--build-arg GDRCOPY_OS_VERSION=Ubuntu24_04 \
--build-arg torch_cuda_arch_list="${CUDA_ARCH_AARCH64_CU129}" \
--build-arg INSTALL_KV_CONNECTORS=true \
--build-arg MOONCAKE_WHEEL_AARCH64="${MOONCAKE_WHEEL_AARCH64_2_39}" \
--build-arg MOONCAKE_WHEEL_X86_64="${MOONCAKE_WHEEL_X86_64}" \
--target vllm-openai \
--progress plain \
-f docker/Dockerfile .
+22
View File
@@ -858,6 +858,28 @@ RUN --mount=type=cache,target=/root/.cache/uv \
uv pip install --system --force-reinstall --no-deps nixl-cu${CUDA_MAJOR}; \
fi
# Optional override: install mooncake-transfer-engine from a URL instead of the
# PyPI release pulled in above. Use this for wheels built with non-default CMake
# flags (e.g. `STORE_USE_ETCD=ON` for master HA). The URL's manylinux glibc
# floor must be <= the FINAL_BASE_IMAGE's glibc.
ARG MOONCAKE_WHEEL_AARCH64
ARG MOONCAKE_WHEEL_X86_64
RUN if [ "$INSTALL_KV_CONNECTORS" = "true" ]; then \
if [ "$TARGETPLATFORM" = "linux/arm64" ]; then \
WHEEL="${MOONCAKE_WHEEL_AARCH64}"; \
else \
WHEEL="${MOONCAKE_WHEEL_X86_64}"; \
fi && \
if [ -n "${WHEEL}" ]; then \
uv pip install --system "${WHEEL}" && \
CUDA_MAJOR="${CUDA_VERSION%%.*}" && \
if [ ! -f /usr/local/cuda/lib64/libcudart.so ] && \
[ -f "/usr/local/cuda/lib64/libcudart.so.${CUDA_MAJOR}" ]; then \
ln -s "libcudart.so.${CUDA_MAJOR}" /usr/local/cuda/lib64/libcudart.so; \
fi; \
fi; \
fi
ENV VLLM_USAGE_SOURCE production-docker-image
ENV VLLM_BUILD_COMMIT=${VLLM_BUILD_COMMIT:-unknown} \
VLLM_BUILD_PIPELINE=${VLLM_BUILD_PIPELINE:-local} \