2025-10-01 13:01:35 UTC

/bin/sh -c #(nop) ARG RELEASE

2025-10-01 13:01:35 UTC

/bin/sh -c #(nop) ARG LAUNCHPAD_BUILD_ARCH

2025-10-01 13:01:35 UTC

/bin/sh -c #(nop) LABEL org.opencontainers.image.ref.name=ubuntu

2025-10-01 13:01:35 UTC

/bin/sh -c #(nop) LABEL org.opencontainers.image.version=24.04

2025-10-01 13:01:37 UTC

/bin/sh -c #(nop) ADD file:249778a1782b02a1c2bcf9f292f5778d81442a53c3de1958d712f10baf7e0b60 in /

2025-10-01 13:01:37 UTC

/bin/sh -c #(nop) CMD ["/bin/bash"]

2025-10-10 16:48:09 UTC (buildkit.dockerfile.v0)

ENV NVARCH=x86_64

2025-10-10 16:48:09 UTC (buildkit.dockerfile.v0)

ENV NVIDIA_REQUIRE_CUDA=cuda>=13.0 brand=unknown,driver>=535,driver<536 brand=grid,driver>=535,driver<536 brand=tesla,driver>=535,driver<536 brand=nvidia,driver>=535,driver<536 brand=quadro,driver>=535,driver<536 brand=quadrortx,driver>=535,driver<536 brand=nvidiartx,driver>=535,driver<536 brand=vapps,driver>=535,driver<536 brand=vpc,driver>=535,driver<536 brand=vcs,driver>=535,driver<536 brand=vws,driver>=535,driver<536 brand=cloudgaming,driver>=535,driver<536 brand=unknown,driver>=550,driver<551 brand=grid,driver>=550,driver<551 brand=tesla,driver>=550,driver<551 brand=nvidia,driver>=550,driver<551 brand=quadro,driver>=550,driver<551 brand=quadrortx,driver>=550,driver<551 brand=nvidiartx,driver>=550,driver<551 brand=vapps,driver>=550,driver<551 brand=vpc,driver>=550,driver<551 brand=vcs,driver>=550,driver<551 brand=vws,driver>=550,driver<551 brand=cloudgaming,driver>=550,driver<551 brand=unknown,driver>=565,driver<566 brand=grid,driver>=565,driver<566 brand=tesla,driver>=565,driver<566 brand=nvidia,driver>=565,driver<566 brand=quadro,driver>=565,driver<566 brand=quadrortx,driver>=565,driver<566 brand=nvidiartx,driver>=565,driver<566 brand=vapps,driver>=565,driver<566 brand=vpc,driver>=565,driver<566 brand=vcs,driver>=565,driver<566 brand=vws,driver>=565,driver<566 brand=cloudgaming,driver>=565,driver<566 brand=unknown,driver>=570,driver<571 brand=grid,driver>=570,driver<571 brand=tesla,driver>=570,driver<571 brand=nvidia,driver>=570,driver<571 brand=quadro,driver>=570,driver<571 brand=quadrortx,driver>=570,driver<571 brand=nvidiartx,driver>=570,driver<571 brand=vapps,driver>=570,driver<571 brand=vpc,driver>=570,driver<571 brand=vcs,driver>=570,driver<571 brand=vws,driver>=570,driver<571 brand=cloudgaming,driver>=570,driver<571 brand=unknown,driver>=575,driver<576 brand=grid,driver>=575,driver<576 brand=tesla,driver>=575,driver<576 brand=nvidia,driver>=575,driver<576 brand=quadro,driver>=575,driver<576 brand=quadrortx,driver>=575,driver<576 brand=nvidiartx,driver>=575,driver<576 brand=vapps,driver>=575,driver<576 brand=vpc,driver>=575,driver<576 brand=vcs,driver>=575,driver<576 brand=vws,driver>=575,driver<576 brand=cloudgaming,driver>=575,driver<576

2025-10-10 16:48:09 UTC (buildkit.dockerfile.v0)

ENV NV_CUDA_CUDART_VERSION=13.0.96-1

2025-10-10 16:48:09 UTC (buildkit.dockerfile.v0)

ARG TARGETARCH

2025-10-10 16:48:09 UTC (buildkit.dockerfile.v0)

LABEL maintainer=NVIDIA CORPORATION <cudatools@nvidia.com>

2025-10-10 16:48:09 UTC (buildkit.dockerfile.v0)

RUN |1 TARGETARCH=amd64 /bin/sh -c apt-get update && apt-get install -y --no-install-recommends gnupg2 curl ca-certificates && curl -fsSL https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/${NVARCH}/3bf863cc.pub | apt-key add - && echo "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/${NVARCH} /" > /etc/apt/sources.list.d/cuda.list && apt-get purge --autoremove -y curl && rm -rf /var/lib/apt/lists/* # buildkit

2025-10-10 16:48:09 UTC (buildkit.dockerfile.v0)

ENV CUDA_VERSION=13.0.2

2025-10-10 16:48:26 UTC (buildkit.dockerfile.v0)

RUN |1 TARGETARCH=amd64 /bin/sh -c apt-get update && apt-get install -y --no-install-recommends cuda-cudart-13-0=${NV_CUDA_CUDART_VERSION} cuda-compat-13-0 && rm -rf /var/lib/apt/lists/* # buildkit

2025-10-10 16:48:26 UTC (buildkit.dockerfile.v0)

RUN |1 TARGETARCH=amd64 /bin/sh -c echo "/usr/local/cuda/lib64" >> /etc/ld.so.conf.d/nvidia.conf # buildkit

2025-10-10 16:48:26 UTC (buildkit.dockerfile.v0)

ENV PATH=/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin

2025-10-10 16:48:26 UTC (buildkit.dockerfile.v0)

ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda/lib64

2025-10-10 16:48:26 UTC (buildkit.dockerfile.v0)

COPY NGC-DL-CONTAINER-LICENSE / # buildkit

2025-10-10 16:48:26 UTC (buildkit.dockerfile.v0)

ENV NVIDIA_VISIBLE_DEVICES=all

2025-10-10 16:48:26 UTC (buildkit.dockerfile.v0)

ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility

2026-09-09 01:50:44 UTC (buildkit.dockerfile.v0)

ARG INSTALL_RUBIN_PRERELEASE

2026-09-09 01:50:44 UTC (buildkit.dockerfile.v0)

ARG CUDA_VERSION

2026-09-09 01:50:44 UTC (buildkit.dockerfile.v0)

ARG PYTHON_VERSION

2026-09-09 01:50:44 UTC (buildkit.dockerfile.v0)

ARG NCCL_VERSION

2026-09-09 01:50:44 UTC (buildkit.dockerfile.v0)

ARG DEADSNAKES_MIRROR_URL

2026-09-09 01:50:44 UTC (buildkit.dockerfile.v0)

ARG DEADSNAKES_GPGKEY_URL

2026-09-09 01:50:44 UTC (buildkit.dockerfile.v0)

ARG GET_PIP_URL

2026-09-09 01:50:44 UTC (buildkit.dockerfile.v0)

ENV CUDA_VERSION=13.0.2

2026-09-09 01:50:44 UTC (buildkit.dockerfile.v0)

ENV DEBIAN_FRONTEND=noninteractive

2026-09-09 01:50:44 UTC (buildkit.dockerfile.v0)

WORKDIR /vllm-workspace

2026-09-09 01:50:44 UTC (buildkit.dockerfile.v0)

RUN |7 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py /bin/sh -c PYTHON_VERSION_STR=$(echo ${PYTHON_VERSION} | sed 's/\.//g') && echo "export PYTHON_VERSION_STR=${PYTHON_VERSION_STR}" >> /etc/environment # buildkit

2026-09-09 01:51:56 UTC (buildkit.dockerfile.v0)

RUN |7 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py /bin/sh -c apt-get update -y && apt-get install -y --no-install-recommends software-properties-common curl sudo ffmpeg libsm6 libxext6 libgl1 && if [ ! -z ${DEADSNAKES_MIRROR_URL} ] ; then if [ ! -z "${DEADSNAKES_GPGKEY_URL}" ] ; then mkdir -p -m 0755 /etc/apt/keyrings ; curl -L ${DEADSNAKES_GPGKEY_URL} | gpg --dearmor > /etc/apt/keyrings/deadsnakes.gpg ; sudo chmod 644 /etc/apt/keyrings/deadsnakes.gpg ; echo "deb [signed-by=/etc/apt/keyrings/deadsnakes.gpg] ${DEADSNAKES_MIRROR_URL} $(lsb_release -cs) main" > /etc/apt/sources.list.d/deadsnakes.list ; fi ; else for i in 1 2 3; do add-apt-repository -y ppa:deadsnakes/ppa && break || { echo "Attempt $i failed, retrying in 5s..."; sleep 5; }; done ; fi && apt-get update -y && apt-get install -y --no-install-recommends python${PYTHON_VERSION} python${PYTHON_VERSION}-dev python${PYTHON_VERSION}-venv libibverbs-dev && rm -rf /var/lib/apt/lists/* && update-alternatives --install /usr/bin/python3 python3 /usr/bin/python${PYTHON_VERSION} 1 && update-alternatives --set python3 /usr/bin/python${PYTHON_VERSION} && ln -sf /usr/bin/python${PYTHON_VERSION}-config /usr/bin/python3-config && rm -f /usr/lib/python${PYTHON_VERSION}/EXTERNALLY-MANAGED && curl -sS ${GET_PIP_URL} | python${PYTHON_VERSION} && python3 --version && python3 -m pip --version # buildkit

2026-09-09 01:52:34 UTC (buildkit.dockerfile.v0)

RUN |7 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py /bin/sh -c CUDA_VERSION_DASH=$(echo $CUDA_VERSION | cut -d. -f1,2 | tr '.' '-') && apt-get update -y && apt-get install -y --no-install-recommends --allow-change-held-packages cuda-nvcc-${CUDA_VERSION_DASH} cuda-cudart-${CUDA_VERSION_DASH} cuda-nvrtc-${CUDA_VERSION_DASH} cuda-cuobjdump-${CUDA_VERSION_DASH} libcurand-dev-${CUDA_VERSION_DASH} libcublas-dev-${CUDA_VERSION_DASH} libnuma-dev numactl && rm -rf /var/lib/apt/lists/* # buildkit

2026-09-09 01:52:36 UTC (buildkit.dockerfile.v0)

RUN |7 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py /bin/sh -c if [ "${INSTALL_RUBIN_PRERELEASE}" = "true" ]; then CUDA_MAJOR_MINOR=$(echo "${CUDA_VERSION}" | cut -d. -f1,2) && if [ "${CUDA_MAJOR_MINOR}" = "13.4" ]; then CURAND_LIBRARY=$(find -H /usr/local/cuda -name libcurand.so.10 -print -quit) && test -n "${CURAND_LIBRARY}" && ln -sf "${CURAND_LIBRARY}" /usr/local/cuda/lib64/libcurand.so.11; elif [ "${CUDA_MAJOR_MINOR}" = "13.5" ]; then CURAND_LIBRARY=$(find -H /usr/local/cuda -name libcurand.so.11 -print -quit) && test -n "${CURAND_LIBRARY}" && ! readlink -f "${CURAND_LIBRARY}" | grep -q 'libcurand\.so\.10'; else echo "Unsupported Rubin CUDA version: ${CUDA_MAJOR_MINOR}" >&2; exit 1; fi && NCCL_LIBRARY=$(find /usr/lib -name libnccl.so.2 -print -quit) && test -n "${NCCL_LIBRARY}" && ln -sf "${NCCL_LIBRARY}" /usr/lib/libnccl.so.2 && ln -sf libnccl.so.2 /usr/lib/libnccl.so && ldconfig; fi # buildkit

2026-09-09 01:52:37 UTC (buildkit.dockerfile.v0)

RUN |7 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py /bin/sh -c python3 -m pip install uv # buildkit

2026-09-09 01:52:37 UTC (buildkit.dockerfile.v0)

ENV UV_HTTP_TIMEOUT=500

2026-09-09 01:52:37 UTC (buildkit.dockerfile.v0)

ENV UV_INDEX_STRATEGY=unsafe-best-match

2026-09-09 01:52:37 UTC (buildkit.dockerfile.v0)

ENV UV_LINK_MODE=copy

2026-09-09 01:52:37 UTC (buildkit.dockerfile.v0)

ENV UV_PYTHON_INSTALL_DIR=/opt/uv/python

2026-09-09 01:52:37 UTC (buildkit.dockerfile.v0)

ENV UV_CACHE_DIR=/opt/uv/cache

2026-09-09 01:52:38 UTC (buildkit.dockerfile.v0)

RUN |7 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py /bin/sh -c mkdir -p "${UV_PYTHON_INSTALL_DIR}" "${UV_CACHE_DIR}" && chgrp -R 0 /opt/uv && chmod -R g+rwX,a+rX /opt/uv # buildkit

2026-09-09 01:52:39 UTC (buildkit.dockerfile.v0)

RUN |7 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py /bin/sh -c CUDA_MAJOR=$(echo $CUDA_VERSION | cut -d. -f1) && if [ -n "$NCCL_VERSION" ]; then echo "nvidia-nccl-cu${CUDA_MAJOR}==${NCCL_VERSION}" > /etc/uv-overrides.txt; else touch /etc/uv-overrides.txt; fi # buildkit

2026-09-09 01:52:39 UTC (buildkit.dockerfile.v0)

ENV UV_OVERRIDE=/etc/uv-overrides.txt

2026-09-09 01:52:40 UTC (buildkit.dockerfile.v0)

RUN |7 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py /bin/sh -c useradd --uid 2000 --gid 0 --create-home --home-dir /home/vllm --shell /bin/bash vllm && mkdir -p /home/vllm/.cache /home/vllm/.config && chown -R 2000:0 /home/vllm && chmod -R g+rwX /home/vllm && chgrp 0 /etc/passwd /etc/group && chmod g=u /etc/passwd /etc/group # buildkit

2026-09-09 01:52:40 UTC (buildkit.dockerfile.v0)

COPY docker/entrypoints/vllm-nonroot-entrypoint.sh /usr/local/bin/vllm-nonroot-entrypoint.sh # buildkit

2026-09-09 01:52:41 UTC (buildkit.dockerfile.v0)

RUN |7 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py /bin/sh -c chmod 0755 /usr/local/bin/vllm-nonroot-entrypoint.sh # buildkit

2026-09-09 01:52:41 UTC (buildkit.dockerfile.v0)

ENV VLLM_ENABLE_CUDA_COMPATIBILITY=0

2026-09-09 01:52:41 UTC (buildkit.dockerfile.v0)

ARG PYTORCH_CUDA_INDEX_BASE_URL

2026-09-09 01:52:41 UTC (buildkit.dockerfile.v0)

COPY requirements/common.txt /tmp/common.txt # buildkit

2026-09-09 01:52:41 UTC (buildkit.dockerfile.v0)

COPY requirements/cuda.txt /tmp/requirements-cuda.txt # buildkit

2026-09-09 01:54:03 UTC (buildkit.dockerfile.v0)

RUN |8 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py PYTORCH_CUDA_INDEX_BASE_URL=https://download.pytorch.org/whl /bin/sh -c if [ "${INSTALL_RUBIN_PRERELEASE}" = "true" ] && [ ! -f /tmp/vllm-requirements/rubin-prerelease.txt ]; then echo "INSTALL_RUBIN_PRERELEASE=true requires requirements/rubin-prerelease.txt" >&2; exit 1; fi && if [ "$(echo $CUDA_VERSION | cut -d. -f1)" = "12" ]; then sed -i 's/^nvidia-cutlass-dsl\[cu13\]/nvidia-cutlass-dsl/' /tmp/requirements-cuda.txt; sed -i 's/^humming-kernels\[cu13\]/humming-kernels[cu12]/' /tmp/requirements-cuda.txt; fi && if [ "${INSTALL_RUBIN_PRERELEASE}" = "true" ]; then uv pip install --system -r /tmp/vllm-requirements/rubin-prerelease.txt && sed -Ei '/^[[:space:]]*(torch|torchvision|torchaudio)[[:space:]]*==/Id' /tmp/requirements-cuda.txt && uv pip install --system -r /tmp/requirements-cuda.txt && uv pip install --system -r /tmp/vllm-requirements/rubin-prerelease.txt && printf '%s\n' "/usr/local/lib/python${PYTHON_VERSION}/dist-packages/nvpl/lib" > /etc/ld.so.conf.d/vllm-rubin-python.conf && ldconfig; else uv pip install --system -r /tmp/requirements-cuda.txt --extra-index-url ${PYTORCH_CUDA_INDEX_BASE_URL}/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.'); fi && rm /tmp/requirements-cuda.txt /tmp/common.txt # buildkit

2026-09-09 01:54:03 UTC (buildkit.dockerfile.v0)

ARG FLASHINFER_VERSION=0.6.18

2026-09-09 01:54:26 UTC (buildkit.dockerfile.v0)

RUN |9 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py PYTORCH_CUDA_INDEX_BASE_URL=https://download.pytorch.org/whl FLASHINFER_VERSION=0.6.18 /bin/sh -c if [ "${INSTALL_RUBIN_PRERELEASE}" != "true" ]; then uv pip install --system flashinfer-jit-cache==${FLASHINFER_VERSION} --index-url https://flashinfer.ai/whl/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.'); fi # buildkit

2026-09-09 01:54:26 UTC (buildkit.dockerfile.v0)

ARG GDRCOPY_CUDA_VERSION=12.8

2026-09-09 01:54:26 UTC (buildkit.dockerfile.v0)

ARG GDRCOPY_OS_VERSION=Ubuntu24_04

2026-09-09 01:54:26 UTC (buildkit.dockerfile.v0)

ARG TARGETPLATFORM

2026-09-09 01:54:26 UTC (buildkit.dockerfile.v0)

COPY tools/install_gdrcopy.sh /tmp/install_gdrcopy.sh # buildkit

2026-09-09 01:54:31 UTC (buildkit.dockerfile.v0)

RUN |12 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py PYTORCH_CUDA_INDEX_BASE_URL=https://download.pytorch.org/whl FLASHINFER_VERSION=0.6.18 GDRCOPY_CUDA_VERSION=12.8 GDRCOPY_OS_VERSION=Ubuntu24_04 TARGETPLATFORM=linux/amd64 /bin/sh -c set -eux; if [ "${INSTALL_RUBIN_PRERELEASE}" = "true" ]; then ldconfig -p | grep -q 'libgdrapi\.so'; else case "${TARGETPLATFORM}" in linux/arm64) UUARCH="aarch64" ;; linux/amd64) UUARCH="x64" ;; *) echo "Unsupported TARGETPLATFORM: ${TARGETPLATFORM}" >&2; exit 1 ;; esac; /tmp/install_gdrcopy.sh "${GDRCOPY_OS_VERSION}" "${GDRCOPY_CUDA_VERSION}" "${UUARCH}"; fi; rm /tmp/install_gdrcopy.sh # buildkit

2026-09-09 01:54:31 UTC (buildkit.dockerfile.v0)

ARG TIMM_VERSION=>=1.0.17

2026-09-09 01:54:31 UTC (buildkit.dockerfile.v0)

ARG RUNAI_MODEL_STREAMER_VERSION=>=0.15.7

2026-09-09 01:54:34 UTC (buildkit.dockerfile.v0)

RUN |14 INSTALL_RUBIN_PRERELEASE=false CUDA_VERSION=13.0.2 PYTHON_VERSION=3.12 NCCL_VERSION=2.30.7 DEADSNAKES_MIRROR_URL= DEADSNAKES_GPGKEY_URL= GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py PYTORCH_CUDA_INDEX_BASE_URL=https://download.pytorch.org/whl FLASHINFER_VERSION=0.6.18 GDRCOPY_CUDA_VERSION=12.8 GDRCOPY_OS_VERSION=Ubuntu24_04 TARGETPLATFORM=linux/amd64 TIMM_VERSION=>=1.0.17 RUNAI_MODEL_STREAMER_VERSION=>=0.15.7 /bin/sh -c uv pip install --system accelerate 'modelscope<1.38' "timm${TIMM_VERSION}" "runai-model-streamer[s3,gcs,azure]${RUNAI_MODEL_STREAMER_VERSION}" # buildkit

2026-09-09 01:54:34 UTC (buildkit.dockerfile.v0)

ARG PIP_INDEX_URL UV_INDEX_URL

2026-09-09 01:54:34 UTC (buildkit.dockerfile.v0)

ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL

2026-09-09 01:54:34 UTC (buildkit.dockerfile.v0)

ARG PYTORCH_CUDA_INDEX_BASE_URL

2026-09-09 01:54:34 UTC (buildkit.dockerfile.v0)

ARG PIP_KEYRING_PROVIDER UV_KEYRING_PROVIDER

2026-09-09 01:54:34 UTC (buildkit.dockerfile.v0)

ARG PYTORCH_NIGHTLY

2026-09-09 01:55:27 UTC (buildkit.dockerfile.v0)

COPY /workspace/torch_lib_versions.txt torch_lib_versions.txt # buildkit

2026-09-09 02:08:46 UTC (buildkit.dockerfile.v0)

COPY /workspace/dist/wheel.sha256 /tmp/vllm-wheel.sha256 # buildkit

2026-09-09 02:08:51 UTC (buildkit.dockerfile.v0)

RUN |8 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL= UV_EXTRA_INDEX_URL= PYTORCH_CUDA_INDEX_BASE_URL=https://download.pytorch.org/whl PIP_KEYRING_PROVIDER=disabled UV_KEYRING_PROVIDER=disabled PYTORCH_NIGHTLY= /bin/sh -c if [ "${PYTORCH_NIGHTLY}" = "1" ]; then echo "Installing torch nightly..." && uv pip install --system $(cat torch_lib_versions.txt | xargs) --pre --index-url ${PYTORCH_CUDA_INDEX_BASE_URL}/nightly/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.') && echo "Installing vLLM..." && uv pip install --system dist/*.whl --verbose --extra-index-url ${PYTORCH_CUDA_INDEX_BASE_URL}/nightly/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.'); else echo "Installing vLLM..." && uv pip install --system dist/*.whl --verbose --extra-index-url ${PYTORCH_CUDA_INDEX_BASE_URL}/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.'); fi # buildkit

2026-09-09 02:08:52 UTC (buildkit.dockerfile.v0)

RUN |8 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL= UV_EXTRA_INDEX_URL= PYTORCH_CUDA_INDEX_BASE_URL=https://download.pytorch.org/whl PIP_KEYRING_PROVIDER=disabled UV_KEYRING_PROVIDER=disabled PYTORCH_NIGHTLY= /bin/sh -c . /etc/environment && uv pip list # buildkit

2026-09-09 02:08:52 UTC (buildkit.dockerfile.v0)

ENV LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda/lib64

2026-09-09 02:08:52 UTC (buildkit.dockerfile.v0)

COPY /tmp/ep_kernels_workspace/dist/wheels.sha256 /tmp/ep-kernels-wheels.sha256 # buildkit

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

RUN |8 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL= UV_EXTRA_INDEX_URL= PYTORCH_CUDA_INDEX_BASE_URL=https://download.pytorch.org/whl PIP_KEYRING_PROVIDER=disabled UV_KEYRING_PROVIDER=disabled PYTORCH_NIGHTLY= /bin/sh -c uv pip install --system ep_kernels/dist/*.whl --verbose --extra-index-url ${PYTORCH_CUDA_INDEX_BASE_URL}/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.') # buildkit

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib64:/usr/local/cuda/lib64:/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda/lib64

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

COPY examples examples # buildkit

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

COPY benchmarks benchmarks # buildkit

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

COPY ./vllm/collect_env.py . # buildkit

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ARG TARGETPLATFORM

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ARG INSTALL_KV_CONNECTORS=false

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ARG INSTALL_RUBIN_PRERELEASE

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ARG TRITON_INSTALL_FROM_SOURCE_REPO

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ARG CUDA_VERSION

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ARG VLLM_BUILD_COMMIT

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ARG VLLM_BUILD_PIPELINE

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ARG VLLM_BUILD_URL

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ARG VLLM_IMAGE_TAG

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ARG PIP_INDEX_URL UV_INDEX_URL

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ENV UV_HTTP_TIMEOUT=500

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ARG torch_cuda_arch_list=7.5 8.0 8.6 8.9 9.0 10.0 11.0 12.0

2026-09-09 02:08:53 UTC (buildkit.dockerfile.v0)

ENV TORCH_CUDA_ARCH_LIST=7.5 8.0 8.6 8.9 9.0 10.0 12.0

2026-09-09 02:09:01 UTC (buildkit.dockerfile.v0)

RUN |14 TARGETPLATFORM=linux/amd64 INSTALL_KV_CONNECTORS=true INSTALL_RUBIN_PRERELEASE=false TRITON_INSTALL_FROM_SOURCE_REPO= CUDA_VERSION=13.0.2 VLLM_BUILD_COMMIT=98dff2a81d747d1dba01a47f939f48c3526d4206 VLLM_BUILD_PIPELINE=019d130e-464e-4ff7-b84b-492992c0c06b VLLM_BUILD_URL=https://buildkite.com/vllm/release-v2/builds/6292 VLLM_IMAGE_TAG=vllm/vllm-openai:v0.29.0-ubuntu2404 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL= UV_EXTRA_INDEX_URL= torch_cuda_arch_list=7.5 8.0 8.6 8.9 9.0 10.0 12.0 /bin/sh -c CUDA_MAJOR="${CUDA_VERSION%%.*}"; CUDA_VERSION_DASH=$(echo $CUDA_VERSION | cut -d. -f1,2 | tr '.' '-'); CUDA_HOME=/usr/local/cuda; BUILD_PKGS="libcusparse-dev-${CUDA_VERSION_DASH} libcublas-dev-${CUDA_VERSION_DASH} libcusolver-dev-${CUDA_VERSION_DASH}"; if [ "$INSTALL_KV_CONNECTORS" = "true" ]; then uv pip install --system -r /tmp/kv_connectors.txt --no-build || ( apt-get update -y && apt-get install -y --no-install-recommends --allow-change-held-packages ${BUILD_PKGS} && uv pip install --system -r /tmp/kv_connectors.txt --no-build-isolation && apt-get purge -y ${BUILD_PKGS} && rm -rf /var/lib/apt/lists/* ); MOONCAKE_VERSION=$(python3 -c "import importlib.metadata as m; print(m.version('mooncake-transfer-engine'))" 2>/dev/null || true) && if [ "${CUDA_MAJOR}" = "13" ] && [ -n "${MOONCAKE_VERSION}" ]; then uv pip uninstall --system mooncake-transfer-engine 2>/dev/null || true; uv pip install --system "mooncake-transfer-engine-cuda13==${MOONCAKE_VERSION}"; fi; fi # buildkit

2026-09-09 02:09:01 UTC (buildkit.dockerfile.v0)

RUN |14 TARGETPLATFORM=linux/amd64 INSTALL_KV_CONNECTORS=true INSTALL_RUBIN_PRERELEASE=false TRITON_INSTALL_FROM_SOURCE_REPO= CUDA_VERSION=13.0.2 VLLM_BUILD_COMMIT=98dff2a81d747d1dba01a47f939f48c3526d4206 VLLM_BUILD_PIPELINE=019d130e-464e-4ff7-b84b-492992c0c06b VLLM_BUILD_URL=https://buildkite.com/vllm/release-v2/builds/6292 VLLM_IMAGE_TAG=vllm/vllm-openai:v0.29.0-ubuntu2404 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL= UV_EXTRA_INDEX_URL= torch_cuda_arch_list=7.5 8.0 8.6 8.9 9.0 10.0 12.0 /bin/sh -c if [ "${INSTALL_RUBIN_PRERELEASE}" = "true" ]; then echo "Installing Rubin prerelease requirements..." && uv pip install --system -r /tmp/vllm-requirements/rubin-prerelease.txt; fi # buildkit

2026-09-09 02:09:02 UTC (buildkit.dockerfile.v0)

COPY /tmp/triton_from_source_workspace/dist/wheel.sha256 /tmp/triton-source-wheel.sha256 # buildkit

2026-09-09 02:09:02 UTC (buildkit.dockerfile.v0)

COPY tools/install_triton_from_source.sh /tmp/install_triton_from_source.sh # buildkit

2026-09-09 02:09:02 UTC (buildkit.dockerfile.v0)

RUN |14 TARGETPLATFORM=linux/amd64 INSTALL_KV_CONNECTORS=true INSTALL_RUBIN_PRERELEASE=false TRITON_INSTALL_FROM_SOURCE_REPO= CUDA_VERSION=13.0.2 VLLM_BUILD_COMMIT=98dff2a81d747d1dba01a47f939f48c3526d4206 VLLM_BUILD_PIPELINE=019d130e-464e-4ff7-b84b-492992c0c06b VLLM_BUILD_URL=https://buildkite.com/vllm/release-v2/builds/6292 VLLM_IMAGE_TAG=vllm/vllm-openai:v0.29.0-ubuntu2404 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL= UV_EXTRA_INDEX_URL= torch_cuda_arch_list=7.5 8.0 8.6 8.9 9.0 10.0 12.0 /bin/sh -c if [ -n "${TRITON_INSTALL_FROM_SOURCE_REPO}" ]; then CUDA_MAJOR_MINOR=$(echo "${CUDA_VERSION}" | cut -d. -f1,2); if [ "${INSTALL_RUBIN_PRERELEASE}" = "true" ]; then case "${CUDA_MAJOR_MINOR}" in 13.4|13.5) TRITON_SYSTEM_PTXAS_PATH=/usr/local/cuda/bin/ptxas TRITON_EXPECTED_PTXAS_VERSION="${CUDA_MAJOR_MINOR}" bash /tmp/install_triton_from_source.sh /tmp/triton-source-dist /tmp/triton-source-wheel.sha256 ;; *) echo "Unsupported Rubin CUDA version: ${CUDA_MAJOR_MINOR}" >&2; exit 1 ;; esac; else bash /tmp/install_triton_from_source.sh /tmp/triton-source-dist /tmp/triton-source-wheel.sha256; fi; fi && rm -f /tmp/install_triton_from_source.sh /tmp/triton-source-wheel.sha256 # buildkit

2026-09-09 02:09:02 UTC (buildkit.dockerfile.v0)

ENV VLLM_USAGE_SOURCE=production-docker-image

2026-09-09 02:09:02 UTC (buildkit.dockerfile.v0)

ENV VLLM_BUILD_COMMIT=98dff2a81d747d1dba01a47f939f48c3526d4206 VLLM_BUILD_PIPELINE=019d130e-464e-4ff7-b84b-492992c0c06b VLLM_BUILD_URL=https://buildkite.com/vllm/release-v2/builds/6292 VLLM_IMAGE_TAG=vllm/vllm-openai:v0.29.0-ubuntu2404

2026-09-09 02:09:02 UTC (buildkit.dockerfile.v0)

LABEL org.opencontainers.image.source=https://github.com/vllm-project/vllm org.opencontainers.image.revision=98dff2a81d747d1dba01a47f939f48c3526d4206 org.opencontainers.image.version=vllm/vllm-openai:v0.29.0-ubuntu2404 org.opencontainers.image.url=https://buildkite.com/vllm/release-v2/builds/6292 ai.vllm.build.commit=98dff2a81d747d1dba01a47f939f48c3526d4206 ai.vllm.build.pipeline=019d130e-464e-4ff7-b84b-492992c0c06b ai.vllm.build.url=https://buildkite.com/vllm/release-v2/builds/6292 ai.vllm.image.tag=vllm/vllm-openai:v0.29.0-ubuntu2404

2026-09-09 02:09:02 UTC (buildkit.dockerfile.v0)

ENTRYPOINT ["vllm" "serve"]

2026-09-14 09:05:34 UTC (buildkit.dockerfile.v0)

SHELL [/bin/bash -eo pipefail -c]

2026-09-14 09:05:34 UTC (buildkit.dockerfile.v0)

ARG TARGETPLATFORM=linux/amd64

2026-09-14 09:05:34 UTC (buildkit.dockerfile.v0)

ARG TARGETOS=linux

2026-09-14 09:05:34 UTC (buildkit.dockerfile.v0)

ARG TARGETARCH=amd64

2026-09-14 09:05:34 UTC (buildkit.dockerfile.v0)

ENV DEBIAN_FRONTEND=noninteractive LANG=en_US.UTF-8 LANGUAGE=en_US:en LC_ALL=en_US.UTF-8

2026-09-14 09:05:34 UTC (buildkit.dockerfile.v0)

RUN |3 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 /bin/bash -eo pipefail -c # Tools # Refresh apt-get update -y && apt-get install -y --no-install-recommends \ software-properties-common apt-transport-https \ ca-certificates gnupg2 lsb-release gnupg-agent \ && apt-get update -y # Install apt-get install -y --no-install-recommends \ ca-certificates build-essential binutils bash openssl \ curl wget aria2 \ git git-lfs \ unzip xz-utils \ tzdata locales \ iproute2 iputils-ping ifstat net-tools dnsutils pciutils ipmitool \ rdma-core rdmacm-utils infiniband-diags \ procps sysstat htop \ tini vim jq bc tree # Update locale localedef -i en_US -c -f UTF-8 -A /usr/share/locale/locale.alias en_US.UTF-8 # Update timezone rm -f /etc/localtime \ && ln -sf /usr/share/zoneinfo/Asia/Shanghai /etc/localtime \ && echo "Asia/Shanghai" > /etc/timezone \ && dpkg-reconfigure --frontend noninteractive tzdata # Cleanup rm -rf /var/tmp/* \ && rm -rf /tmp/* \ && rm -rf /var/cache/apt # buildkit

2026-09-14 09:05:34 UTC (buildkit.dockerfile.v0)

ARG PYTHON_VERSION=3.12

2026-09-14 09:05:34 UTC (buildkit.dockerfile.v0)

ENV PYTHON_VERSION=3.12

2026-09-14 09:05:34 UTC (buildkit.dockerfile.v0)

RUN |4 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 PYTHON_VERSION=3.12 /bin/bash -eo pipefail -c # Python # Update alternatives if ! command -v python >/dev/null 2>&1 \ && command -v python${PYTHON_VERSION} >/dev/null 2>&1; then update-alternatives --install /usr/bin/python python $(which python${PYTHON_VERSION}) 1 fi # buildkit

2026-09-14 09:05:34 UTC (buildkit.dockerfile.v0)

ENV PIP_NO_CACHE_DIR=1 PIP_DISABLE_PIP_VERSION_CHECK=1 PIP_ROOT_USER_ACTION=ignore

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

RUN |4 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 PYTHON_VERSION=3.12 /bin/bash -eo pipefail -c # Buildkit cat <<EOT >/tmp/requirements.txt build cmake<4 wheel pybind11<3 Cython EOT pip install -r /tmp/requirements.txt # Cleanup rm -rf /var/tmp/* \ && rm -rf /tmp/* # buildkit

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

ARG CUDA_VERSION=13.0.1

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

ARG CUDA_ARCHS

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

ENV CUDA_HOME=/usr/local/cuda CUDA_VERSION=13.0.1 CUDA_ARCHS=

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

ENV UV_SYSTEM_PYTHON=1 UV_PRERELEASE=allow UV_SKIP_WHEEL_FILENAME_CHECK=1 UV_NO_CACHE=1 UV_HTTP_TIMEOUT=500 UV_INDEX_STRATEGY=unsafe-best-match

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

ARG VLLM_VERSION=0.29.0

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

ARG VLLM_TORCH_VERSION=2.13.0

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

ARG VLLM_TORCH_CUDA_VERSION=13.0.1

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

ENV VLLM_VERSION=0.29.0 VLLM_TORCH_VERSION=2.13.0 VLLM_TORCH_CUDA_VERSION=13.0.1

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

SHELL [/bin/bash -eo pipefail -c]

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

ARG TARGETPLATFORM=linux/amd64

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

ARG TARGETOS=linux

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

ARG TARGETARCH=amd64

2026-09-14 09:05:37 UTC (buildkit.dockerfile.v0)

ARG VLLM_ROUTER_VERSION=0.1.15

2026-09-15 05:05:50 UTC (buildkit.dockerfile.v0)

RUN |10 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 PYTHON_VERSION=3.12 CUDA_VERSION=13.0.1 CUDA_ARCHS= VLLM_VERSION=0.29.0 VLLM_TORCH_VERSION=2.13.0 VLLM_TORCH_CUDA_VERSION=13.0.1 VLLM_ROUTER_VERSION=0.1.15 /bin/bash -eo pipefail -c # vLLM Router # # The PD router GPUStack renders for a disaggregated group. Without it in # the image there is no router to run, and the group either falls back to # an engine example script (no metrics, no circuit breaker, no health # check) or does not start at all. uv pip install \ vllm-router==${VLLM_ROUTER_VERSION} # Fail the build rather than ship an image whose router is missing: the # alternative surfaces as a container that exits at deploy time, one # layer away from anything that explains it. command -v vllm-router >/dev/null # Cleanup rm -rf /var/tmp/* \ && rm -rf /tmp/* # buildkit

2026-09-15 05:05:53 UTC (buildkit.dockerfile.v0)

RUN |10 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 PYTHON_VERSION=3.12 CUDA_VERSION=13.0.1 CUDA_ARCHS= VLLM_VERSION=0.29.0 VLLM_TORCH_VERSION=2.13.0 VLLM_TORCH_CUDA_VERSION=13.0.1 VLLM_ROUTER_VERSION=0.1.15 /bin/bash -eo pipefail -c # Ray # Install Ray Client and Default. # Pinned to 2.54.0: ray 2.55 (ray-project/ray#61029) breaks distributed # ray.init when the driver has no co-located raylet (gpustack split containers). RAY_VERSION=2.54.0 cat <<EOT >/tmp/requirements.txt ray[cgraph]==${RAY_VERSION} ray[client]==${RAY_VERSION} ray[default]==${RAY_VERSION} EOT uv pip install \ -r /tmp/requirements.txt # Cleanup rm -rf /var/tmp/* \ && rm -rf /tmp/* # buildkit

2026-09-15 05:06:20 UTC (buildkit.dockerfile.v0)

RUN |10 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 PYTHON_VERSION=3.12 CUDA_VERSION=13.0.1 CUDA_ARCHS= VLLM_VERSION=0.29.0 VLLM_TORCH_VERSION=2.13.0 VLLM_TORCH_CUDA_VERSION=13.0.1 VLLM_ROUTER_VERSION=0.1.15 /bin/bash -eo pipefail -c # Omni if [[ ! -d /omni/workspace ]]; then echo "Skipping vLLM Omni installation for ${TARGETARCH}..." exit 0 fi # Install uv pip install --no-build-isolation \ /omni/workspace/*.whl # Dependencies cat <<EOT >/tmp/requirements.txt onnxruntime sox EOT uv pip install \ -r /tmp/requirements.txt # Cleanup rm -rf /var/tmp/* \ && rm -rf /tmp/* # buildkit

2026-09-15 05:06:20 UTC (buildkit.dockerfile.v0)

ARG VLLM_LMCACHE_VERSION=0.5.4

2026-09-15 05:06:20 UTC (buildkit.dockerfile.v0)

ENV VLLM_LMCACHE_VERSION=0.5.4

2026-09-15 05:06:25 UTC (buildkit.dockerfile.v0)

RUN |11 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 PYTHON_VERSION=3.12 CUDA_VERSION=13.0.1 CUDA_ARCHS= VLLM_VERSION=0.29.0 VLLM_TORCH_VERSION=2.13.0 VLLM_TORCH_CUDA_VERSION=13.0.1 VLLM_ROUTER_VERSION=0.1.15 VLLM_LMCACHE_VERSION=0.5.4 /bin/bash -eo pipefail -c # LMCache # Install ## Replaces the base's own lmcache —— see the note on VLLM_LMCACHE_VERSION. Must run before ## the CuPy dedup below, since lmcache declares a CUDA-major-specific cupy variant. uv pip install --no-build-isolation \ /lmcache/workspace/*.whl # Review uv pip show lmcache ## Fail here on a torch ABI mismatch: lmcache itself only WARNs and falls back to its slow ## torch baseline. RTLD_NOW resolves all symbols without module init, so this needs no GPU. cat <<EOT >/tmp/check_lmcache.py import ctypes, glob, os, sysconfig, torch platlib = sysconfig.get_paths()["platlib"] sos = glob.glob(platlib + "/lmcache/*.so") assert sos, "no lmcache extension found under " + platlib for so in sos: ctypes.CDLL(so, mode=os.RTLD_NOW) print("lmcache resolved against torch " + torch.__version__ + ": " + ", ".join(sos)) EOT python /tmp/check_lmcache.py # Cleanup rm -rf /var/tmp/* \ && rm -rf /tmp/* # buildkit

2026-09-15 05:06:27 UTC (buildkit.dockerfile.v0)

RUN |11 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 PYTHON_VERSION=3.12 CUDA_VERSION=13.0.1 CUDA_ARCHS= VLLM_VERSION=0.29.0 VLLM_TORCH_VERSION=2.13.0 VLLM_TORCH_CUDA_VERSION=13.0.1 VLLM_ROUTER_VERSION=0.1.15 VLLM_LMCACHE_VERSION=0.5.4 /bin/bash -eo pipefail -c # CuPy # ray[cgraph] pulls cupy-cuda12x while the base already ships cupy-cuda13x; both share # the same cupy/ tree, so uninstalling one breaks the other. Purge both, then reinstall # the CUDA-matching variant at the version vLLM pinned in the base (its cupy-cuda13x), # so cu129 also gets that pinned version instead of ray's unconstrained cupy-cuda12x. # Our own lmcache is a third source but agrees with this reinstall: its build stage sets # LMCACHE_CUDA_MAJOR from the same CUDA major used here. IFS="." read -r CUDA_MAJOR CUDA_MINOR CUDA_PATCH <<< "${CUDA_VERSION}" PLATLIB="$(python -c 'import sysconfig; print(sysconfig.get_paths()["platlib"])')" CUPY="cupy-cuda${CUDA_MAJOR}x" CUPY_VER="$(uv pip show cupy-cuda13x 2>/dev/null | awk '/^Version:/{print $2}' || true)" uv pip uninstall cupy-cuda12x cupy-cuda13x || true rm -rf "${PLATLIB}"/cupy "${PLATLIB}"/cupyx "${PLATLIB}"/cupy_backends # cupy-cuda12x and cupy-cuda13x share the same version number, so reuse the base's # pinned version and install only the variant matching the CUDA major we build for. # Fail loudly if the base no longer ships cupy, rather than installing an unpinned latest. if [[ -z "${CUPY_VER}" ]]; then echo "cupy-cuda13x not found in base image; refusing to install unpinned cupy" >&2 exit 1 fi uv pip install "${CUPY}==${CUPY_VER}" # Review uv pip show "${CUPY}" # Cleanup rm -rf /var/tmp/* \ && rm -rf /tmp/* # buildkit

2026-09-15 05:06:27 UTC (buildkit.dockerfile.v0)

RUN |11 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 PYTHON_VERSION=3.12 CUDA_VERSION=13.0.1 CUDA_ARCHS= VLLM_VERSION=0.29.0 VLLM_TORCH_VERSION=2.13.0 VLLM_TORCH_CUDA_VERSION=13.0.1 VLLM_ROUTER_VERSION=0.1.15 VLLM_LMCACHE_VERSION=0.5.4 /bin/bash -eo pipefail -c # Postprocess # Review uv pip tree \ --package vllm \ --package vllm-omni \ --package flashinfer-python \ --package torch \ --package triton \ --package deep-gemm \ --package deep-ep \ --package lmcache \ --package mooncake-transfer-engine \ --package ray # buildkit

2026-09-18 08:01:31 UTC (buildkit.dockerfile.v0)

RUN |11 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 PYTHON_VERSION=3.12 CUDA_VERSION=13.0.1 CUDA_ARCHS= VLLM_VERSION=0.29.0 VLLM_TORCH_VERSION=2.13.0 VLLM_TORCH_CUDA_VERSION=13.0.1 VLLM_ROUTER_VERSION=0.1.15 VLLM_LMCACHE_VERSION=0.5.4 /bin/bash -eo pipefail -c # Patch tree -hs /workspace/patches pushd $(pip show vllm | grep Location: | cut -d" " -f 2) \ && for f in /workspace/patches/vllm/*.patch; do patch -p1 < "$f" || exit 1; done if pip show vllm_omni > /dev/null 2>&1; then \ pushd $(pip show vllm_omni | grep Location: | cut -d" " -f 2) \ && patch -p1 < /workspace/patches/vllm_omni/*.patch; \ fi # buildkit

2026-09-18 08:01:33 UTC (buildkit.dockerfile.v0)

RUN |11 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 PYTHON_VERSION=3.12 CUDA_VERSION=13.0.1 CUDA_ARCHS= VLLM_VERSION=0.29.0 VLLM_TORCH_VERSION=2.13.0 VLLM_TORCH_CUDA_VERSION=13.0.1 VLLM_ROUTER_VERSION=0.1.15 VLLM_LMCACHE_VERSION=0.5.4 /bin/bash -eo pipefail -c # Dependencies # Install cat <<EOT >/tmp/requirements.txt # audio extras librosa soundfile mistral_common[audio] # tokenizer extras fastokens==0.2.0 EOT uv pip install \ -r /tmp/requirements.txt # Cleanup rm -rf /var/tmp/* \ && rm -rf /tmp/* # buildkit

2026-09-18 08:01:33 UTC (buildkit.dockerfile.v0)

ARG DEPENDENCY_PACKAGES=diffusers flash-attn flashinfer-python lmcache lmcache-ascend mooncake-transfer-engine-npu mooncake-transfer-engine-rocm mooncake-transfer-engine ray runai-model-streamer sgl-kernel-npu sglang-kernel sglang-router torch torch-musa torch-npu transformers vllm-ascend vllm-metax vllm-musa vllm-omni vllm-router

2026-09-18 08:01:33 UTC (buildkit.dockerfile.v0)

RUN |12 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 PYTHON_VERSION=3.12 CUDA_VERSION=13.0.1 CUDA_ARCHS= VLLM_VERSION=0.29.0 VLLM_TORCH_VERSION=2.13.0 VLLM_TORCH_CUDA_VERSION=13.0.1 VLLM_ROUTER_VERSION=0.1.15 VLLM_LMCACHE_VERSION=0.5.4 DEPENDENCY_PACKAGES=diffusers flash-attn flashinfer-python lmcache lmcache-ascend mooncake-transfer-engine-npu mooncake-transfer-engine-rocm mooncake-transfer-engine ray runai-model-streamer sgl-kernel-npu sglang-kernel sglang-router torch torch-musa torch-npu transformers vllm-ascend vllm-metax vllm-musa vllm-omni vllm-router /bin/bash -eo pipefail -c DEPENDENCY_PACKAGES="${DEPENDENCY_PACKAGES}" bash /tmp/probe_dependencies.sh # buildkit

2026-09-18 08:01:33 UTC (buildkit.dockerfile.v0)

ENV RAY_EXPERIMENTAL_NOSET_CUDA_VISIBLE_DEVICES=1 SAFETENSORS_FAST_GPU=1 RUNAI_STREAMER_MEMORY_LIMIT=0 RUNAI_STREAMER_LOG_TO_STDERR=1 RUNAI_STREAMER_LOG_LEVEL=INFO

2026-09-18 08:01:33 UTC (buildkit.dockerfile.v0)

WORKDIR /

2026-09-18 08:01:33 UTC (buildkit.dockerfile.v0)

ENTRYPOINT ["tini" "--"]

2026-09-24 01:04:29 UTC (buildkit.dockerfile.v0)

SHELL [/bin/bash -eo pipefail -c]

2026-09-24 01:04:29 UTC (buildkit.dockerfile.v0)

ARG TARGETPLATFORM=linux/amd64

2026-09-24 01:04:29 UTC (buildkit.dockerfile.v0)

ARG TARGETOS=linux

2026-09-24 01:04:29 UTC (buildkit.dockerfile.v0)

ARG TARGETARCH=amd64

2026-09-24 01:04:29 UTC (buildkit.dockerfile.v0)

RUN |3 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 /bin/bash -eo pipefail -c # Stop MooncakeStoreConnector from saving a load MultiConnector rejected # Locate vLLM without importing it: an import runs vLLM's logging setup, which # writes to stdout and would end up inside the captured path. VLLM_SITE="$(python3 -c 'import importlib.util, os; print(os.path.dirname(os.path.dirname(importlib.util.find_spec("vllm").origin)))')" SCHEDULER="${VLLM_SITE}/vllm/distributed/kv_transfer/kv_connector/v1/mooncake/store/scheduler.py" # Refuse to guess at a layout this patch was not written against. test -f "${SCHEDULER}" # Idempotent: an image that already carries the fix is left alone. if grep -q "if load_spec is None or not load_spec.can_load:" "${SCHEDULER}"; then echo "[info] The pending-load branch already skips rejected load specs; nothing to patch." exit 0 fi pushd "${VLLM_SITE}" patch -p1 <<'PATCHEOF' diff --git a/vllm/distributed/kv_transfer/kv_connector/v1/mooncake/store/scheduler.py b/vllm/distributed/kv_transfer/kv_connector/v1/mooncake/store/scheduler.py --- a/vllm/distributed/kv_transfer/kv_connector/v1/mooncake/store/scheduler.py +++ b/vllm/distributed/kv_transfer/kv_connector/v1/mooncake/store/scheduler.py @@ -371,7 +371,12 @@ ) in self._unfinished_requests.items(): if request_id not in request_ids and request_id not in cached_reqs.req_ids: load_spec = self.load_specs.pop(request_id, None) - if not load_spec: + # A load spec may have been proposed by this connector's + # lookup but rejected by MultiConnector in favor of another + # connector. Only the chosen connector may issue the pending + # load; the normal store path gets its blocks later from + # SchedulerOutput once the request is actually scheduled. + if load_spec is None or not load_spec.can_load: continue num_tokens_to_compute = load_spec.kvpool_cached_tokens request_tracker = RequestTracker( PATCHEOF popd # Assert the patch landed. The grep is POSITIVE: a negated grep is exempt from # set -e, so the inverted form would pass over a failure. grep -q "if load_spec is None or not load_spec.can_load:" "${SCHEDULER}" # Import it for real: the grep proves the text is there, this proves it parses. python3 -c " from vllm.distributed.kv_transfer.kv_connector.v1.mooncake.store.scheduler import MooncakeStoreScheduler print('[info] MooncakeStoreScheduler imports.') " # Cleanup rm -rf /var/tmp/* \ && rm -rf /tmp/* # buildkit

2026-09-24 01:04:29 UTC (buildkit.dockerfile.v0)

ARG DEPENDENCY_PACKAGES=diffusers flash-attn flashinfer-python lmcache lmcache-ascend mooncake-transfer-engine-npu mooncake-transfer-engine-rocm mooncake-transfer-engine ray runai-model-streamer sgl-kernel-npu sglang-kernel sglang-router torch torch-musa torch-npu transformers vllm-ascend vllm-metax vllm-musa vllm-omni vllm-router

2026-09-24 01:04:29 UTC (buildkit.dockerfile.v0)

RUN |4 TARGETPLATFORM=linux/amd64 TARGETOS=linux TARGETARCH=amd64 DEPENDENCY_PACKAGES=diffusers flash-attn flashinfer-python lmcache lmcache-ascend mooncake-transfer-engine-npu mooncake-transfer-engine-rocm mooncake-transfer-engine ray runai-model-streamer sgl-kernel-npu sglang-kernel sglang-router torch torch-musa torch-npu transformers vllm-ascend vllm-metax vllm-musa vllm-omni vllm-router /bin/bash -eo pipefail -c DEPENDENCY_PACKAGES="${DEPENDENCY_PACKAGES}" bash /tmp/probe_dependencies.sh # buildkit

2026-09-24 01:04:29 UTC (buildkit.dockerfile.v0)

WORKDIR /

2026-09-24 01:04:29 UTC (buildkit.dockerfile.v0)

ENTRYPOINT ["tini" "--"]