[Image][Build] Cherry pick #4062 from main (#4506)

### What this PR does / why we need it?
This patch aims to integrate the mooncake
[v0.3.7.2.post2](https://github.com/kvcache-ai/Mooncake/releases/tag/v0.3.7.post2)
to vllm-ascend images

Signed-off-by: wangli <[email protected]>

Author: Potabk

Dockerfile

@@ -19,20 +19,29 @@ FROM quay.io/ascend/cann:8.3.rc2-910b-ubuntu22.04-py3.11
 
 ARG PIP_INDEX_URL="https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple"
 ARG COMPILE_CUSTOM_KERNELS=1
+ARG MOONCAKE_TAG="v0.3.7.post2"
 
 # Define environments
 ENV DEBIAN_FRONTEND=noninteractive
 ENV COMPILE_CUSTOM_KERNELS=${COMPILE_CUSTOM_KERNELS}
 
-RUN apt-get update -y && \
-    apt-get install -y python3-pip git vim wget net-tools gcc g++ cmake libnuma-dev && \
-    rm -rf /var/cache/apt/* && \
-    rm -rf /var/lib/apt/lists/*
-
 WORKDIR /workspace
 
 COPY . /vllm-workspace/vllm-ascend/
 
+# Install Mooncake dependencies
+RUN apt-get update -y && \
+    apt-get install -y git vim wget net-tools gcc g++ cmake libnuma-dev && \
+    git clone --depth 1 --branch ${MOONCAKE_TAG} https://github.com/kvcache-ai/Mooncake /vllm-workspace/Mooncake && \
+    cp /vllm-workspace/vllm-ascend/tools/mooncake_installer.sh /vllm-workspace/Mooncake/ && \
+    cd /vllm-workspace/Mooncake && bash mooncake_installer.sh -y && \
+    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/lib64 && \
+    mkdir -p build && cd build && cmake .. -DUSE_ASCEND_DIRECT=ON && \
+    make -j$(nproc) && make install && \
+    rm -fr /vllm-workspace/Mooncake/build && \
+    rm -rf /var/cache/apt/* && \
+    rm -rf /var/lib/apt/lists/*
+
 RUN pip config set global.index-url ${PIP_INDEX_URL}
 
 # Install vLLM
@@ -54,7 +63,7 @@ RUN export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi
     python3 -m pip cache purge
 
 # Install modelscope (for fast download) and ray (for multinode)
-RUN python3 -m pip install modelscope 'ray>=2.47.1' 'protobuf>3.20.0' && \
+RUN python3 -m pip install modelscope 'ray>=2.47.1,<=2.48.0' 'protobuf>3.20.0' && \
     python3 -m pip cache purge
 
 CMD ["/bin/bash"]

Dockerfile.a3

@@ -19,21 +19,29 @@ FROM quay.io/ascend/cann:8.3.rc2-a3-ubuntu22.04-py3.11
 
 ARG PIP_INDEX_URL="https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple"
 ARG COMPILE_CUSTOM_KERNELS=1
+ARG MOONCAKE_TAG=v0.3.7.post2
 
+COPY . /vllm-workspace/vllm-ascend/
 # Define environments
 ENV DEBIAN_FRONTEND=noninteractive
 ENV COMPILE_CUSTOM_KERNELS=${COMPILE_CUSTOM_KERNELS}
 
-RUN apt-get update -y && \
-    apt-get install -y python3-pip git vim wget net-tools gcc g++ cmake libnuma-dev && \
-    rm -rf /var/cache/apt/* && \
-    rm -rf /var/lib/apt/lists/*
+RUN pip config set global.index-url ${PIP_INDEX_URL}
 
 WORKDIR /workspace
 
-COPY . /vllm-workspace/vllm-ascend/
-
-RUN pip config set global.index-url ${PIP_INDEX_URL}
+# Install Mooncake dependencies
+RUN apt-get update -y && \
+    apt-get install -y git vim wget net-tools gcc g++ cmake libnuma-dev && \
+    git clone --depth 1 --branch ${MOONCAKE_TAG} https://github.com/kvcache-ai/Mooncake /vllm-workspace/Mooncake && \
+    cp /vllm-workspace/vllm-ascend/tools/mooncake_installer.sh /vllm-workspace/Mooncake/ && \
+    cd /vllm-workspace/Mooncake && bash mooncake_installer.sh -y && \
+    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/lib64 && \
+    mkdir -p build && cd build && cmake .. -DUSE_ASCEND_DIRECT=ON && \
+    make -j$(nproc) && make install && \
+    rm -fr /vllm-workspace/Mooncake/build && \
+    rm -rf /var/cache/apt/* && \
+    rm -rf /var/lib/apt/lists/*
 
 # Install vLLM
 ARG VLLM_REPO=https://github.com/vllm-project/vllm.git
@@ -54,7 +62,7 @@ RUN export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi
     python3 -m pip cache purge
 
 # Install modelscope (for fast download) and ray (for multinode)
-RUN python3 -m pip install modelscope 'ray>=2.47.1' 'protobuf>3.20.0' && \
+RUN python3 -m pip install modelscope 'ray>=2.47.1,<=2.48.0' 'protobuf>3.20.0' && \
     python3 -m pip cache purge
 
 CMD ["/bin/bash"]

Dockerfile.a3.openEuler

@@ -19,23 +19,36 @@ FROM quay.io/ascend/cann:8.3.rc2-a3-openeuler24.03-py3.11
 
 ARG PIP_INDEX_URL="https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple"
 ARG COMPILE_CUSTOM_KERNELS=1
+ARG MOONCAKE_TAG="v0.3.7.post2"
 
 ENV COMPILE_CUSTOM_KERNELS=${COMPILE_CUSTOM_KERNELS}
 
-RUN yum update -y && \
-    yum install -y python3-pip git vim wget net-tools gcc gcc-c++ make cmake numactl-devel && \
-    rm -rf /var/cache/yum
-
 RUN pip config set global.index-url ${PIP_INDEX_URL}
 
 WORKDIR /workspace
 
 COPY . /vllm-workspace/vllm-ascend/
 
+SHELL ["/bin/bash", "-c"]
+
+RUN yum update -y && \
+    yum install -y git vim wget net-tools gcc gcc-c++ make cmake numactl-devel && \
+    git clone --depth 1 --branch ${MOONCAKE_TAG} https://github.com/kvcache-ai/Mooncake /vllm-workspace/Mooncake && \
+    cp /vllm-workspace/vllm-ascend/tools/mooncake_installer.sh /vllm-workspace/Mooncake/ && \
+    ARCH=$(uname -m) && \
+    source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
+    export LD_LIBRARY_PATH=/usr/local/Ascend/ascend-toolkit/latest/${ARCH}-linux/devlib:/usr/local/Ascend/ascend-toolkit/latest/${ARCH}-linux/lib64:$LD_LIBRARY_PATH && \
+    export CPLUS_INCLUDE_PATH=$CPLUS_INCLUDE_PATH:/usr/include/c++/12:/usr/include/c++/12/${ARCH}-openEuler-linux && \
+    cd /vllm-workspace/Mooncake && \
+    bash mooncake_installer.sh -y && \
+    mkdir -p build && cd build && cmake .. -DUSE_ASCEND_DIRECT=ON && \
+    make -j$(nproc) && make install && \
+    rm -fr /vllm-workspace/Mooncake/build && \
+    rm -rf /var/cache/yum/*
+
 # Install vLLM
 ARG VLLM_REPO=https://github.com/vllm-project/vllm.git
 ARG VLLM_TAG=v0.11.0
-
 RUN git clone --depth 1 $VLLM_REPO --branch $VLLM_TAG /vllm-workspace/vllm
 # In x86, triton will be installed by vllm. But in Ascend, triton doesn't work correctly. we need to uninstall it.
 RUN VLLM_TARGET_DEVICE="empty" python3 -m pip install -e /vllm-workspace/vllm/[audio] --extra-index https://download.pytorch.org/whl/cpu/ && \
@@ -52,7 +65,7 @@ RUN export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi
     python3 -m pip cache purge
 
 # Install modelscope (for fast download) and ray (for multinode)
-RUN python3 -m pip install modelscope 'ray>=2.47.1' 'protobuf>3.20.0' && \
+RUN python3 -m pip install modelscope 'ray>=2.47.1,<=2.48.0' 'protobuf>3.20.0' && \
     python3 -m pip cache purge
 
 CMD ["/bin/bash"]

Dockerfile.openEuler

@@ -19,23 +19,37 @@ FROM quay.io/ascend/cann:8.3.rc2-910b-openeuler24.03-py3.11
 
 ARG PIP_INDEX_URL="https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple"
 ARG COMPILE_CUSTOM_KERNELS=1
+ARG MOONCAKE_TAG="v0.3.7.post2"
 
 ENV COMPILE_CUSTOM_KERNELS=${COMPILE_CUSTOM_KERNELS}
 
-RUN yum update -y && \
-    yum install -y python3-pip git vim wget net-tools gcc gcc-c++ make cmake numactl-devel && \
-    rm -rf /var/cache/yum
 
 RUN pip config set global.index-url ${PIP_INDEX_URL}
 
 WORKDIR /workspace
 
 COPY . /vllm-workspace/vllm-ascend/
 
+SHELL ["/bin/bash", "-c"]
+
+RUN yum update -y && \
+    yum install -y git vim wget net-tools gcc gcc-c++ make cmake numactl-devel && \
+    git clone --depth 1 --branch ${MOONCAKE_TAG} https://github.com/kvcache-ai/Mooncake /vllm-workspace/Mooncake && \
+    cp /vllm-workspace/vllm-ascend/tools/mooncake_installer.sh /vllm-workspace/Mooncake/ && \
+    ARCH=$(uname -m) && \
+    source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
+    export LD_LIBRARY_PATH=/usr/local/Ascend/ascend-toolkit/latest/${ARCH}-linux/devlib:/usr/local/Ascend/ascend-toolkit/latest/${ARCH}-linux/lib64:$LD_LIBRARY_PATH && \
+    export CPLUS_INCLUDE_PATH=$CPLUS_INCLUDE_PATH:/usr/include/c++/12:/usr/include/c++/12/${ARCH}-openEuler-linux && \
+    cd /vllm-workspace/Mooncake && \
+    bash mooncake_installer.sh -y && \
+    mkdir -p build && cd build && cmake .. -DUSE_ASCEND_DIRECT=ON && \
+    make -j$(nproc) && make install && \
+    rm -fr /vllm-workspace/Mooncake/build && \
+    rm -rf /var/cache/yum/*
+
 # Install vLLM
 ARG VLLM_REPO=https://github.com/vllm-project/vllm.git
-ARG VLLM_TAG=v0.11.0
-
+ARG VLLM_TAG=v0.11.2
 RUN git clone --depth 1 $VLLM_REPO --branch $VLLM_TAG /vllm-workspace/vllm
 # In x86, triton will be installed by vllm. But in Ascend, triton doesn't work correctly. we need to uninstall it.
 RUN VLLM_TARGET_DEVICE="empty" python3 -m pip install -e /vllm-workspace/vllm/[audio] --extra-index https://download.pytorch.org/whl/cpu/ && \
@@ -52,7 +66,7 @@ RUN export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi
     python3 -m pip cache purge
 
 # Install modelscope (for fast download) and ray (for multinode)
-RUN python3 -m pip install modelscope 'ray>=2.47.1' 'protobuf>3.20.0' && \
+RUN python3 -m pip install modelscope 'ray>=2.47.1,<=2.48.0' 'protobuf>3.20.0' && \
     python3 -m pip cache purge
 
 CMD ["/bin/bash"]