Add new cuda version (#102)

* compact Dockerfile
* optimize install script
* update cuda pkgs
* fix sys py3

.github/workflows/build-docker.yml

@@ -232,6 +232,30 @@ jobs:
           alias_image cuda_11.8 latest cuda latest
           push_image cuda
 
+  qpod_cuda_123:
+    name: 'cuda_12.3'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - run: |
+          source ./tool.sh
+          build_image tmp latest docker_atom/Dockerfile --build-arg "BASE_IMG=nvidia/cuda:12.3.2-cudnn9-devel-ubuntu22.04" && clear_images nvidia/cuda
+          build_image tmp latest docker_base/Dockerfile --build-arg "BASE_IMG=tmp"
+          build_image cuda_12.3 latest docker_cuda/Dockerfile --build-arg "BASE_IMG=tmp"
+          push_image cuda
+
+  qpod_cuda_120:
+    name: 'cuda_12.0'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - run: |
+          source ./tool.sh
+          build_image tmp latest docker_atom/Dockerfile --build-arg "BASE_IMG=nvidia/cuda:12.0.1-cudnn8-devel-ubuntu22.04" && clear_images nvidia/cuda
+          build_image tmp latest docker_base/Dockerfile --build-arg "BASE_IMG=tmp"
+          build_image cuda_12.0 latest docker_cuda/Dockerfile --build-arg "BASE_IMG=tmp"
+          push_image cuda
+
   qpod_cuda_112:
     name: 'cuda_11.2'
     runs-on: ubuntu-latest
@@ -245,15 +269,15 @@ jobs:
           push_image cuda
 
   qpod_tf2:
-    name: 'tf2,tf2-cuda112'
-    needs: qpod_cuda_112
+    name: 'tf2,tf2-cuda123'
+    needs: qpod_cuda_123
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
       - run: |
           source ./tool.sh
-          build_image tf2-cuda112 latest docker_core/Dockerfile --build-arg "BASE_IMG=cuda_11.2"  --build-arg "ARG_PROFILE_PYTHON=tf2"
-          alias_image tf2-cuda112 latest tf2 latest
+          build_image tf2-cuda123 latest docker_core/Dockerfile --build-arg "BASE_IMG=cuda_12.3"  --build-arg "ARG_PROFILE_PYTHON=tf2"
+          alias_image tf2-cuda123 latest tf2 latest
           push_image
 
   qpod_torch_cuda118:
@@ -269,26 +293,26 @@ jobs:
           push_image
 
   qpod_torch_cuda112:
-    name: 'torch-cuda112'
-    needs: qpod_cuda_112
+    name: 'torch-cuda123'
+    needs: qpod_cuda_123
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
       - run: |
           source ./tool.sh
-          build_image torch-cuda112 latest docker_core/Dockerfile --build-arg "BASE_IMG=cuda_11.2" --build-arg "ARG_PROFILE_PYTHON=torch"
+          build_image torch-cuda123 latest docker_core/Dockerfile --build-arg "BASE_IMG=cuda_12.3" --build-arg "ARG_PROFILE_PYTHON=torch"
           push_image
 
-  qpod_paddle_cuda118:
-    name: 'paddle-cuda,paddle-cuda118'
-    needs: qpod_cuda_118
+  qpod_paddle_cuda120:
+    name: 'paddle-cuda,paddle-cuda120'
+    needs: qpod_cuda_120
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
       - run: |
           source ./tool.sh
-          build_image paddle-cuda118 latest docker_core/Dockerfile --build-arg "BASE_IMG=cuda_11.8" --build-arg "ARG_PROFILE_PYTHON=paddle,mkl"
-          alias_image paddle-cuda118 latest paddle-cuda latest
+          build_image paddle-cuda120 latest docker_core/Dockerfile --build-arg "BASE_IMG=cuda_12.0" --build-arg "ARG_PROFILE_PYTHON=paddle,mkl"
+          alias_image paddle-cuda120 latest paddle-cuda latest
           push_image
 
   qpod_paddle_cuda112:
@@ -379,9 +403,10 @@ jobs:
       - uses: actions/checkout@v4
       - run: |
           source ./tool.sh && build_image docker-kit latest docker_docker_kit/Dockerfile
-          printenv > /tmp/docker.env
+          printenv | grep -v 'PATH' > /tmp/docker.env
+          docker run --rm ${IMG_PREFIX}/docker-kit printenv | sort
           docker run --rm --env-file /tmp/docker.env -v $(pwd):/tmp -w /tmp \
-            ${IMG_PREFIX}/docker-kit /opt/conda/bin/python /opt/utils/image-syncer/run_sync.py ${IMG_NAMESPACE}/base
+            ${IMG_PREFIX}/docker-kit python /opt/utils/image-syncer/run_sync.py ${IMG_NAMESPACE}/base
           push_image docker-kit
 
   ## Sync all images in this build (listed by "names") to mirror registry.
@@ -392,6 +417,6 @@ jobs:
       - uses: actions/checkout@v4
       - run: |
           source ./tool.sh
-          printenv > /tmp/docker.env
+          printenv | grep -v 'PATH' > /tmp/docker.env
           docker run --rm --env-file /tmp/docker.env -v $(pwd):/tmp -w /tmp \
-            ${IMG_NAMESPACE:-qpod}/docker-kit /opt/conda/bin/python /opt/utils/image-syncer/run_jobs.py
+            ${IMG_NAMESPACE:-qpod}/docker-kit python /opt/utils/image-syncer/run_jobs.py

docker_atom/Dockerfile

@@ -26,7 +26,7 @@ ENV SHELL=/bin/bash \
 SHELL ["/bin/bash", "--login", "-c"]
 
 # --> Install OS libraries and setup some configurations
-RUN set -x && cd /tmp \
+RUN set -eux && cd /tmp \
  && apt-get -qq update --fix-missing && apt-get -y -qq upgrade \
  && apt-get -qq install -y --no-install-recommends \
      apt-utils apt-transport-https ca-certificates gnupg2 dirmngr locales sudo lsb-release curl \

docker_atom/work/script-setup.sh

@@ -10,8 +10,9 @@ setup_mamba() {
   && ln -sf /opt/mamba/mamba /usr/bin/ \
   && touch /etc/conda/.condarc \
   && printf "channels:\n"       >> /etc/conda/.condarc \
-  && printf "  - conda-forge\n" >> /etc/conda/.condarc \
-  && echo "@ Version of mamba:" && mamba info
+  && printf "  - conda-forge\n" >> /etc/conda/.condarc
+
+     echo "@ Version of mamba:" && mamba info
 }
 
 
@@ -62,21 +63,21 @@ setup_conda_with_mamba() {
 
 setup_conda_download() {
   # https://docs.conda.io/projects/miniconda/en/latest/index.html
-  mkdir -pv "${CONDA_PREFIX}"
-  wget -qO- "https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-$(arch).sh" -O /tmp/conda.sh
-  bash /tmp/conda.sh -f -b -p "${CONDA_PREFIX}/"
-  rm -rf /tmp/conda.sh
-  setup_conda_postprocess
+     mkdir -pv "${CONDA_PREFIX}" \
+  && wget -qO- "https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-$(arch).sh" -O /tmp/conda.sh \
+  && bash /tmp/conda.sh -f -b -p "${CONDA_PREFIX}/" \
+  && rm -rf /tmp/conda.sh \
+  && setup_conda_postprocess
 }
 
 setup_nvtop() {
   # The compiliation requries CMake 3.18 or higher. default version in CUDA 11.2 images is 3.16.3
-  curl -sL https://apt.kitware.com/keys/kitware-archive-latest.asc | sudo tee /etc/apt/trusted.gpg.d/kitware.asc \
+     curl -sL https://apt.kitware.com/keys/kitware-archive-latest.asc | sudo tee /etc/apt/trusted.gpg.d/kitware.asc \
   && echo "deb https://apt.kitware.com/ubuntu/ $(lsb_release -cs) main" > /etc/apt/sources.list.d/kitware.list \
   && apt-get -qq update -yq --fix-missing && apt-get -qq install -yq --no-install-recommends cmake
 
   # Install Utilities "nvtop" from source: libdrm-dev libsystemd-dev used by AMD/Intel GPU support, libudev-dev used by ubuntu18.04
-  LIB_PATH=$(find / -name "libnvidia-ml*" 2>/dev/null) \
+     LIB_PATH=$(find / -name "libnvidia-ml*" 2>/dev/null) \
   && DIRECTORY=$(pwd) && cd /tmp \
   && sudo apt-get -qq update --fix-missing \
   && sudo apt-get -qq install -y --no-install-recommends libncurses5-dev libdrm-dev libsystemd-dev libudev-dev \
@@ -85,8 +86,9 @@ setup_nvtop() {
   && cmake .. -DCMAKE_LIBRARY_PATH="$(dirname ${LIB_PATH})" -DNVIDIA_SUPPORT=ON -DAMDGPU_SUPPORT=ON -DINTEL_SUPPORT=ON \
   && make && sudo make install \
   && cd "${DIRECTORY}" && rm -rf /tmp/nvtop \
-  && sudo apt-get -qq remove -y libncurses5-dev libdrm-dev libsystemd-dev libudev-dev \
-  && nvtop --version
+  && sudo apt-get -qq remove -y libncurses5-dev libdrm-dev libsystemd-dev libudev-dev
+
+     nvtop --version
 }
 
 
@@ -123,8 +125,9 @@ setup_java_maven() {
      VERSION_MAVEN=$(curl -sL https://maven.apache.org/download.cgi | grep 'latest' | head -1 | grep -Po '\d[\d.]+') \
   && install_zip "http://archive.apache.org/dist/maven/maven-3/${VERSION_MAVEN}/binaries/apache-maven-${VERSION_MAVEN}-bin.zip" \
   && mv "/opt/apache-maven-${VERSION_MAVEN}" /opt/maven \
-  && ln -sf /opt/maven/bin/mvn* /usr/bin/ \
-  && echo "@ Version of Maven: $(mvn --version)"
+  && ln -sf /opt/maven/bin/mvn* /usr/bin/
+
+     echo "@ Version of Maven: $(mvn --version)"
 }
 
 
@@ -139,8 +142,9 @@ setup_node() {
   && ln -sf /opt/node/bin/n* /usr/bin/ \
   && echo 'export PATH=${PATH}:/opt/node/bin' >> /etc/profile.d/path-node.sh \
   && npm install -g npm \
-  && echo "@ Version of Node and npm: $(node -v) $(npm -v)" \
-  && cd /tmp && corepack enable && yarn set version stable \
+  && cd /tmp && corepack enable && yarn set version stable
+
+     echo "@ Version of Node and npm: $(node -v) $(npm -v)" \
   && echo "@ Version of Yarn: $(yarn -v)"
 }
 
@@ -153,8 +157,9 @@ setup_GO() {
   && echo 'export GOROOT="/opt/go"'		      >> /etc/profile.d/path-go.sh \
   && echo 'export  GOBIN="$GOROOT/bin"'		  >> /etc/profile.d/path-go.sh \
   && echo 'export GOPATH="$GOROOT/path"'		>> /etc/profile.d/path-go.sh \
-  && echo 'export PATH=$PATH:$GOROOT/bin:$GOPATH/bin'	  >> /etc/profile.d/path-go.sh \
-  && echo "@ Version of golang: $(go version)"
+  && echo 'export PATH=$PATH:$GOROOT/bin:$GOPATH/bin'	  >> /etc/profile.d/path-go.sh
+
+     echo "@ Version of golang: $(go version)"
 }
 
 
@@ -165,8 +170,9 @@ setup_rust() {
   && curl -sSf https://sh.rustup.rs | sh -s -- -y --no-modify-path --profile minimal --default-toolchain stable \
   && echo 'export CARGO_HOME="/opt/cargo"'		>> /etc/profile.d/path-rust.sh \
   && echo 'export RUSTUP_HOME="/opt/rust"'		>> /etc/profile.d/path-rust.sh \
-  && echo 'export PATH=$PATH:/opt/cargo/bin'	>> /etc/profile.d/path-rust.sh \
-  && echo "@ Version of rustup: $(rustup --version)" \
+  && echo 'export PATH=$PATH:/opt/cargo/bin'	>> /etc/profile.d/path-rust.sh
+
+     echo "@ Version of rustup: $(rustup --version)" \
   && echo "@ Version of rustc:  $(rustc --version)"
 }
 
@@ -178,8 +184,9 @@ setup_julia() {
   && ln -fs /opt/julia/bin/julia /usr/bin/julia \
   && mkdir -p /opt/julia/pkg \
   && echo "import Libdl; push!(Libdl.DL_LOAD_PATH, \"/opt/conda/lib\")" >> /opt/julia/etc/julia/startup.jl \
-  && echo "DEPOT_PATH[1]=\"/opt/julia/pkg\""                            >> /opt/julia/etc/julia/startup.jl \
-  && echo "@ Version of Julia: $(julia --version)"
+  && echo "DEPOT_PATH[1]=\"/opt/julia/pkg\""                            >> /opt/julia/etc/julia/startup.jl
+
+     echo "@ Version of Julia: $(julia --version)"
 }
 
 
@@ -191,8 +198,9 @@ setup_lua_base() {
  && mv /opt/lua-* /tmp/lua && cd /tmp/lua \
  && make linux test && make install INSTALL_TOP=${LUA_HOME:-"/opt/lua"} \
  && ln -sf ${LUA_HOME:-"/opt/lua"}/bin/lua* /usr/bin/ \
- && rm -rf /tmp/lua \
- && echo "@ Version of LUA installed: $(lua -v)"
+ && rm -rf /tmp/lua
+
+    echo "@ Version of LUA installed: $(lua -v)"
 }
 
 setup_lua_rocks() {
@@ -204,32 +212,36 @@ setup_lua_rocks() {
  && mv /opt/luarocks-* /tmp/luarocks && cd /tmp/luarocks \
  && ./configure --prefix=${LUA_HOME:-"/opt/lua"} --with-lua-include=${LUA_HOME:-"/opt/lua"}/include && make install \
  && ln -sf /opt/lua/bin/lua* /usr/bin/ \
- && rm -rf /tmp/luarocks \
- && echo "@ Version of luarocks: $(luarocks --version)"
+ && rm -rf /tmp/luarocks
+
+    echo "@ Version of luarocks: $(luarocks --version)"
 }
 
 setup_traefik() {
      TRAEFIK_VERSION=$(curl -sL https://github.com/traefik/traefik/releases.atom | grep 'releases/tag' | head -1 | grep -Po '\d[\d.]+') \
   && TRAEFIK_URL="https://github.com/traefik/traefik/releases/download/v${TRAEFIK_VERSION}/traefik_v${TRAEFIK_VERSION}_linux_$(dpkg --print-architecture).tar.gz" \
   && install_tar_gz "${TRAEFIK_URL}" traefik \
-  && ln -sf /opt/traefik /usr/bin/ \
-  && echo "@ Version of traefik: $(traefik version)"
+  && ln -sf /opt/traefik /usr/bin/
+
+    echo "@ Version of traefik: $(traefik version)"
 }
 
 
 setup_bazel() {
      BAZEL_VERSION=$(curl -sL https://github.com/bazelbuild/bazel/releases.atom | grep 'releases/tag' | head -1 | grep -Po '\d[\d.]+' ) \
   && BAZEL_URL="https://github.com/bazelbuild/bazel/releases/download/${BAZEL_VERSION}/bazel-${BAZEL_VERSION}-installer-linux-x86_64.sh" \
   && curl -o /tmp/bazel.sh -sL "${BAZEL_URL}" && chmod +x /tmp/bazel.sh \
-  && /tmp/bazel.sh && rm /tmp/bazel.sh \
-  && echo "@ Version of bazel: $(bazel --version)"
+  && /tmp/bazel.sh && rm /tmp/bazel.sh
+
+    echo "@ Version of bazel: $(bazel --version)"
 }
 
 
 setup_gradle() {
      GRADLE_VERSION=$(curl -sL https://github.com/gradle/gradle/releases.atom | grep 'releases/tag' | grep -v 'M' | head -1 | grep -Po '\d[\d.]+' ) \
   && install_zip "https://downloads.gradle.org/distributions/gradle-${GRADLE_VERSION}-bin.zip" \
   && mv /opt/gradle* /opt/gradle \
-  && ln -sf /opt/gradle/bin/gradle /usr/bin \
-  && echo "@ Version of gradle: $(gradle --version)"
+  && ln -sf /opt/gradle/bin/gradle /usr/bin
+
+     echo "@ Version of gradle: $(gradle --version)"
 }

docker_atom/work/script-utils.sh

@@ -29,16 +29,16 @@ install_octave()  { cat $1 | cut -d "%" -f 1 | sed '/^$/d' | xargs -r -n1 -I {}
 install_zip()     { curl -o /tmp/TMP.zip -sL $1 && unzip -q -d /opt/ /tmp/TMP.zip && rm /tmp/TMP.zip ; }
 
 # function to download a .tar.gz file and unzip it to /opt/, add a second argument to extract only those file
-install_tar_gz()  { curl -o /tmp/TMP.tgz -sL $1 && tar -C /opt/ -xzf /tmp/TMP.tgz $2 && rm /tmp/TMP.tgz ; }
+install_tar_gz()  { curl -o /tmp/TMP.tgz -sL $1 && tar -C /opt/ -xzf /tmp/TMP.tgz ${2:-} && rm /tmp/TMP.tgz ; }
 
 # function to download a .tar.bz file and unzip it to /opt/, add a second argument to extract only those file
-install_tar_bz()  { curl -o /tmp/TMP.tbz -sL $1 && tar -C /opt/ -xjf /tmp/TMP.tbz $2 && rm /tmp/TMP.tbz ; }
+install_tar_bz()  { curl -o /tmp/TMP.tbz -sL $1 && tar -C /opt/ -xjf /tmp/TMP.tbz ${2:-} && rm /tmp/TMP.tbz ; }
 
 # function to download a .tar.xz file and unzip it to /opt/, add a second argument to extract only those file
-install_tar_xz()  { curl -o /tmp/TMP.txz -sL $1 && tar -C /opt/ -xJf /tmp/TMP.txz $2 && rm /tmp/TMP.txz ; }
+install_tar_xz()  { curl -o /tmp/TMP.txz -sL $1 && tar -C /opt/ -xJf /tmp/TMP.txz ${2:-} && rm /tmp/TMP.txz ; }
 
 # function to install java packages from a text file which lists JAR file maven full names (add comments with % char)
-install_mvn() { cat $1 | cut -d "%" -f 1 | xargs -r -n1 -I {} mvn dependency:copy -DlocalRepositoryDirectory="/tmp/m2repo" -Djavax.net.ssl.trustStorePassword=changeit -Dartifact="{}" -DoutputDirectory="$2" ; }
+install_mvn() { cat $1 | cut -d "%" -f 1 | xargs -r -n1 -I {} mvn dependency:copy -DlocalRepositoryDirectory="/tmp/m2repo" -Djavax.net.ssl.trustStorePassword=changeit -Dartifact="{}" -DoutputDirectory="${2:-}" ; }
 
 # function to clean up
 install__clean(){

docker_base/Dockerfile

@@ -9,30 +9,32 @@ ARG PYTHON_VERSION="3.12"
 
 ENV CONDA_PREFIX=/opt/conda
 
-RUN set -x && source /opt/utils/script-setup.sh && source /opt/utils/script-setup-sys.sh \
+RUN set -eux && source /opt/utils/script-setup.sh && source /opt/utils/script-setup-sys.sh \
  && export PATH=$PATH:${CONDA_PREFIX}/bin \
  && install_apt /opt/utils/install_list_base.apt \
  && echo "Install tini:" && setup_tini \
- && export HAS_SYS_PY3=$( [ -x "$(command -v python3)" ] && echo 'true' || echo 'false' ) && echo "@ Detect OS Py3 installed: $HAS_SYS_PY3" \
+ && export HAS_SYS_PY3=$( [ -x "$(command -v python3)" ] && echo 'true' || echo 'false' ) && echo "@ Detect OS Py3 installed: ${HAS_SYS_PY3:-error}" \
  && echo "Install Mamba:" && setup_mamba \
  && echo "Install Python ${PYTHON_VERSION} and conda:" && setup_conda_with_mamba ${PYTHON_VERSION} \
  && echo "Backup the conda version of platform.py as it's different from the system version." \
  && cp "${CONDA_PREFIX}/lib/python${PYTHON_VERSION}"/platform.py "${CONDA_PREFIX}/lib/python${PYTHON_VERSION}"/platform.py.bak \
- && $($HAS_SYS_PY3) && ( \
-       echo "Replace system Python3 with conda Python - note that /bin and /sbin are symlinks of /usr/bin in docker image ubuntu" \
+ && if $(${HAS_SYS_PY3:-false}) ; then \
+       py3versions -d \
     && PYTHON_VERSION_DEFAULT=$(py3versions -v -i) \
-    && sed -i "s/${PYTHON_VERSION_DEFAULT}/${PYTHON_VERSION}/g" /usr/share/python3/debian_defaults \
     && PYTHON_PTH_FILE=$("${CONDA_PREFIX}"/bin/python3 -c 'import sys;print(sys.path[-1]+"/usr_share.pth")') \
+    && echo "Found default system python3 version: ${PYTHON_VERSION_DEFAULT}" \
+    && echo "Replace system Python3 with conda Python - note that /bin and /sbin are symlinks of /usr/bin in docker image ubuntu" \
+    && sed -i "s/${PYTHON_VERSION_DEFAULT}/${PYTHON_VERSION}/g" /usr/share/python3/debian_defaults \
     && echo "/usr/share/pyshared/" >>  "${PYTHON_PTH_FILE}" \
     && echo "/usr/share/python3/"  >>  "${PYTHON_PTH_FILE}" \
     && rm -rf $(/usr/bin/python3 -c 'import sys; print(" ".join(i for i in sys.path if "python" in i))') /usr/bin/python3* /usr/lib/python${PYTHON_VERSION} \
- ) || true \
- && ln -sf "${CONDA_PREFIX}"/lib/python${PYTHON_VERSION} /usr/lib/ \
+    && rm -rf /usr/lib/python${PYTHON_VERSION} && ln -sf "${CONDA_PREFIX}"/lib/python${PYTHON_VERSION} /usr/lib/ ; \
+ else \
+     echo "No default python found in OS." ; \
+ fi \
  && ln -sf "${CONDA_PREFIX}"/bin/python3.* /usr/bin/ \
  && ln -sf "/usr/bin/python${PYTHON_VERSION}" /usr/bin/python  \
  && ln -sf "/usr/bin/python${PYTHON_VERSION}" /usr/bin/python3 \
- && ls -alh /usr/bin/python* \
- && $($HAS_SYS_PY3) && ( py3versions -d ) || true \
  && lsb_release -a \
- && which python && python --version \
+ && which python && python --version && ls -alh /usr/bin/python* \
  && install__clean && fix_permission 0 /opt