xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-06-01 19:27:08 +08:00
Code Issues Packages Projects Releases Wiki Activity

refactor

Browse Source
This commit is contained in:
atalman 2025-12-23 14:05:27 -08:00
parent 69b9a83985
commit 06df43c720
1 changed files with 22 additions and 61 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

83
docker/Dockerfile
View File

@ -54,12 +54,13 @@ ARG PIP_EXTRA_INDEX_URL
ARG UV_INDEX_URL=${PIP_INDEX_URL} ARG UV_INDEX_URL=${PIP_INDEX_URL}
ARG UV_EXTRA_INDEX_URL=${PIP_EXTRA_INDEX_URL} ARG UV_EXTRA_INDEX_URL=${PIP_EXTRA_INDEX_URL}
# PyTorch provides its own indexes for standard and nightly builds
ARG PYTORCH_CUDA_INDEX_BASE_URL=https://download.pytorch.org/whl
# Flag to install PyTorch nightly instead of stable # Flag to install PyTorch nightly instead of stable
ARG USE_TORCH_NIGHTLY=false ARG USE_TORCH_NIGHTLY=false
# PyTorch provides its own indexes for standard and nightly builds
# This will be set conditionally based on USE_TORCH_NIGHTLY in the base stage
ARG PYTORCH_CUDA_INDEX_BASE_URL=https://download.pytorch.org/whl
# PIP supports multiple authentication schemes, including keyring # PIP supports multiple authentication schemes, including keyring
# By parameterizing the PIP_KEYRING_PROVIDER variable and setting it to # By parameterizing the PIP_KEYRING_PROVIDER variable and setting it to
# disabled by default, we allow third-party to use keyring authentication for # disabled by default, we allow third-party to use keyring authentication for
@ -133,14 +134,20 @@ RUN ldconfig /usr/local/cuda-$(echo $CUDA_VERSION | cut -d. -f1,2)/compat/
ARG PYTORCH_CUDA_INDEX_BASE_URL ARG PYTORCH_CUDA_INDEX_BASE_URL
ARG USE_TORCH_NIGHTLY ARG USE_TORCH_NIGHTLY
# Set PyTorch index URL and prerelease flag based on USE_TORCH_NIGHTLY # Set PyTorch index URL based on USE_TORCH_NIGHTLY
# We compute the index URL once and reuse it across all stages
RUN if [ "$USE_TORCH_NIGHTLY" = "true" ]; then \ RUN if [ "$USE_TORCH_NIGHTLY" = "true" ]; then \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}/nightly/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment; \ PYTORCH_SUFFIX="/nightly"; \
else \
PYTORCH_SUFFIX=""; \
fi && \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}${PYTORCH_SUFFIX}/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment && \
if [ "$USE_TORCH_NIGHTLY" = "true" ]; then \
echo "PRERELEASE_FLAG=--prerelease=allow" >> /etc/environment; \ echo "PRERELEASE_FLAG=--prerelease=allow" >> /etc/environment; \
else \ else \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment; \
echo "PRERELEASE_FLAG=" >> /etc/environment; \ echo "PRERELEASE_FLAG=" >> /etc/environment; \
fi fi
ENV PYTORCH_INDEX="" ENV PYTORCH_INDEX=""
ENV PRERELEASE_FLAG="" ENV PRERELEASE_FLAG=""
RUN . /etc/environment && echo "PYTORCH_INDEX=${PYTORCH_INDEX}" && echo "PRERELEASE_FLAG=${PRERELEASE_FLAG}" RUN . /etc/environment && echo "PYTORCH_INDEX=${PYTORCH_INDEX}" && echo "PRERELEASE_FLAG=${PRERELEASE_FLAG}"
@ -168,8 +175,6 @@ ARG TARGETPLATFORM
ARG PIP_INDEX_URL UV_INDEX_URL ARG PIP_INDEX_URL UV_INDEX_URL
ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL
ARG PYTORCH_CUDA_INDEX_BASE_URL
ARG USE_TORCH_NIGHTLY
# install build dependencies # install build dependencies
COPY requirements/build.txt requirements/build.txt COPY requirements/build.txt requirements/build.txt
@ -181,17 +186,6 @@ ENV UV_INDEX_STRATEGY="unsafe-best-match"
# Use copy mode to avoid hardlink failures with Docker cache mounts # Use copy mode to avoid hardlink failures with Docker cache mounts
ENV UV_LINK_MODE=copy ENV UV_LINK_MODE=copy
# Set PyTorch index URL and prerelease flag based on USE_TORCH_NIGHTLY
RUN if [ "$USE_TORCH_NIGHTLY" = "true" ]; then \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}/nightly/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment; \
echo "PRERELEASE_FLAG=--prerelease=allow" >> /etc/environment; \
else \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment; \
echo "PRERELEASE_FLAG=" >> /etc/environment; \
fi
ENV PYTORCH_INDEX=""
ENV PRERELEASE_FLAG=""
RUN --mount=type=cache,target=/root/.cache/uv \ RUN --mount=type=cache,target=/root/.cache/uv \
. /etc/environment && \ . /etc/environment && \
uv pip install --python /opt/venv/bin/python3 -r requirements/build.txt \ uv pip install --python /opt/venv/bin/python3 -r requirements/build.txt \
@ -314,8 +308,6 @@ ARG TARGETPLATFORM
ARG PIP_INDEX_URL UV_INDEX_URL ARG PIP_INDEX_URL UV_INDEX_URL
ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL
ARG PYTORCH_CUDA_INDEX_BASE_URL
ARG USE_TORCH_NIGHTLY
# install build dependencies # install build dependencies
COPY requirements/build.txt requirements/build.txt COPY requirements/build.txt requirements/build.txt
@ -327,17 +319,6 @@ ENV UV_INDEX_STRATEGY="unsafe-best-match"
# Use copy mode to avoid hardlink failures with Docker cache mounts # Use copy mode to avoid hardlink failures with Docker cache mounts
ENV UV_LINK_MODE=copy ENV UV_LINK_MODE=copy
# Set PyTorch index URL and prerelease flag based on USE_TORCH_NIGHTLY
RUN if [ "$USE_TORCH_NIGHTLY" = "true" ]; then \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}/nightly/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment; \
echo "PRERELEASE_FLAG=--prerelease=allow" >> /etc/environment; \
else \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment; \
echo "PRERELEASE_FLAG=" >> /etc/environment; \
fi
ENV PYTORCH_INDEX=""
ENV PRERELEASE_FLAG=""
RUN --mount=type=cache,target=/root/.cache/uv \ RUN --mount=type=cache,target=/root/.cache/uv \
. /etc/environment && \ . /etc/environment && \
uv pip install --python /opt/venv/bin/python3 -r requirements/build.txt \ uv pip install --python /opt/venv/bin/python3 -r requirements/build.txt \
@ -389,8 +370,6 @@ FROM base AS dev
ARG PIP_INDEX_URL UV_INDEX_URL ARG PIP_INDEX_URL UV_INDEX_URL
ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL
ARG PYTORCH_CUDA_INDEX_BASE_URL
ARG USE_TORCH_NIGHTLY
# This timeout (in seconds) is necessary when installing some dependencies via uv since it's likely to time out # This timeout (in seconds) is necessary when installing some dependencies via uv since it's likely to time out
# Reference: https://github.com/astral-sh/uv/pull/1694 # Reference: https://github.com/astral-sh/uv/pull/1694
@ -399,17 +378,6 @@ ENV UV_INDEX_STRATEGY="unsafe-best-match"
# Use copy mode to avoid hardlink failures with Docker cache mounts # Use copy mode to avoid hardlink failures with Docker cache mounts
ENV UV_LINK_MODE=copy ENV UV_LINK_MODE=copy
# Set PyTorch index URL and prerelease flag based on USE_TORCH_NIGHTLY
RUN if [ "$USE_TORCH_NIGHTLY" = "true" ]; then \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}/nightly/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment; \
echo "PRERELEASE_FLAG=--prerelease=allow" >> /etc/environment; \
else \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment; \
echo "PRERELEASE_FLAG=" >> /etc/environment; \
fi
ENV PYTORCH_INDEX=""
ENV PRERELEASE_FLAG=""
# Install libnuma-dev, required by fastsafetensors (fixes #20384) # Install libnuma-dev, required by fastsafetensors (fixes #20384)
RUN apt-get update && apt-get install -y --no-install-recommends libnuma-dev && rm -rf /var/lib/apt/lists/* RUN apt-get update && apt-get install -y --no-install-recommends libnuma-dev && rm -rf /var/lib/apt/lists/*
COPY requirements/lint.txt requirements/lint.txt COPY requirements/lint.txt requirements/lint.txt
@ -514,14 +482,20 @@ RUN ldconfig /usr/local/cuda-$(echo $CUDA_VERSION | cut -d. -f1,2)/compat/
ARG PYTORCH_CUDA_INDEX_BASE_URL ARG PYTORCH_CUDA_INDEX_BASE_URL
ARG USE_TORCH_NIGHTLY ARG USE_TORCH_NIGHTLY
# Set PyTorch index URL and prerelease flag based on USE_TORCH_NIGHTLY # Set PyTorch index URL based on USE_TORCH_NIGHTLY
# We compute the index URL once and reuse it across all stages
RUN if [ "$USE_TORCH_NIGHTLY" = "true" ]; then \ RUN if [ "$USE_TORCH_NIGHTLY" = "true" ]; then \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}/nightly/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment; \ PYTORCH_SUFFIX="/nightly"; \
else \
PYTORCH_SUFFIX=""; \
fi && \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}${PYTORCH_SUFFIX}/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment && \
if [ "$USE_TORCH_NIGHTLY" = "true" ]; then \
echo "PRERELEASE_FLAG=--prerelease=allow" >> /etc/environment; \ echo "PRERELEASE_FLAG=--prerelease=allow" >> /etc/environment; \
else \ else \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment; \
echo "PRERELEASE_FLAG=" >> /etc/environment; \ echo "PRERELEASE_FLAG=" >> /etc/environment; \
fi fi
ENV PYTORCH_INDEX="" ENV PYTORCH_INDEX=""
ENV PRERELEASE_FLAG="" ENV PRERELEASE_FLAG=""
@ -636,8 +610,6 @@ ARG PYTHON_VERSION
ARG PIP_INDEX_URL UV_INDEX_URL ARG PIP_INDEX_URL UV_INDEX_URL
ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL
ARG PYTORCH_CUDA_INDEX_BASE_URL
ARG USE_TORCH_NIGHTLY
# This timeout (in seconds) is necessary when installing some dependencies via uv since it's likely to time out # This timeout (in seconds) is necessary when installing some dependencies via uv since it's likely to time out
# Reference: https://github.com/astral-sh/uv/pull/1694 # Reference: https://github.com/astral-sh/uv/pull/1694
@ -646,17 +618,6 @@ ENV UV_INDEX_STRATEGY="unsafe-best-match"
# Use copy mode to avoid hardlink failures with Docker cache mounts # Use copy mode to avoid hardlink failures with Docker cache mounts
ENV UV_LINK_MODE=copy ENV UV_LINK_MODE=copy
# Set PyTorch index URL and prerelease flag based on USE_TORCH_NIGHTLY
RUN if [ "$USE_TORCH_NIGHTLY" = "true" ]; then \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}/nightly/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment; \
echo "PRERELEASE_FLAG=--prerelease=allow" >> /etc/environment; \
else \
echo "PYTORCH_INDEX=${PYTORCH_CUDA_INDEX_BASE_URL}/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')" >> /etc/environment; \
echo "PRERELEASE_FLAG=" >> /etc/environment; \
fi
ENV PYTORCH_INDEX=""
ENV PRERELEASE_FLAG=""
RUN echo 'tzdata tzdata/Areas select America' | debconf-set-selections \ RUN echo 'tzdata tzdata/Areas select America' | debconf-set-selections \
&& echo 'tzdata tzdata/Zones/America select Los_Angeles' | debconf-set-selections \ && echo 'tzdata tzdata/Zones/America select Los_Angeles' | debconf-set-selections \
&& apt-get update -y \ && apt-get update -y \