如何在 rstudio 服务器中使用 mamba/conda 包

Question

我正在尝试写入 dockerfile 以从 yaml 安装 conda 环境并从我的项目目录中运行 rstudio 服务器。我知道rstudio-server-conda，但希望能够使用单个 dockerfile 来创建图像。

Dockerfile

FROM rocker/rstudio-stable:devel

# Set working directory
WORKDIR ${HOME}

# Copy directory files to image 
COPY . ${HOME}

# Copy repo into ${HOME}, make user own $HOME
USER root

# Install base utilities
RUN apt-get update && \
    apt-get install -y wget && \
    apt-get clean && \
    rm -rf /var/lib/apt/lists/*

# Install miniconda
ENV CONDA_DIR /opt/conda
RUN wget --quiet https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh \
    -O ~/miniconda.sh && \
     /bin/bash ~/miniconda.sh -b -p /opt/conda && \
     

# Put conda in path so we can use conda activate
ENV PATH=$CONDA_DIR/bin:$PATH

# Install umamba
RUN conda install -y micromamba -c conda-forge

# Create a conda environment from the environment yml
COPY --chown=$MAMBA_USER:$MAMBA_USER environment.yml /tmp/environment.yml
RUN micromamba create --yes --file /tmp/environment.yml && \
    micromamba  clean --all --yes

# Activate the conda environment
ARG MAMBA_DOCKERFILE_ACTIVATE=1 

RUN chown -R ${NB_USER} . ${HOME}
USER ${NB_USER}

# Settings required for conda+rstudio
ENV RSTUDIO_WHICH_R=${CONDAENV}/bin/R
ENV RETICULATE_PYTHON=${CONDAENV}/bin/python

RUN echo rsession-which-r=${RSTUDIO_WHICH_R} > /etc/rstudio/rserver.conf && \
    echo rsession-ld-library-path=${CONDAENV}/lib >> /etc/rstudio/rserver.conf && \
    echo "R_LIBS_USER=${CONDAENV}/lib/R/library" > /home/rstudio/.Renviron

## Run an install.R script, if it exists.
#RUN if [ -f /R/install.R ]; then R --quiet -f /R/install.R; fi

环境.yml

channels:
  - conda-forge
dependencies:
  - r-devtools=2.4.3=r41hc72bb7e_0
  - r-tidyverse=1.3.1=r41hc72bb7e_0

dockerfile 安装了 rstudio-server 和 micromamba，但是当我尝试读取环境 yaml 中导入的包时，找不到它们。

# build image
docker build --tag umamba-rstudio -f Dockerfile .

docker run --rm \
    -e ENV_NAME=environment \
    --mount type=bind,source="$(pwd)",destination=/home/rstudio \
    -p 127.0.0.1:8787:8787 \
    -e DISABLE_AUTH=true \
    umamba-rstudio

Answer 1

我将介绍两种方法：

1. 对现有文件进行最少的更改
2. 导致改进的 Dockerfile 和映像的较大更改

这里是改动很小的 Dockerfile，其中 3 行更改后附加了 # DIFF: ... cmets：

FROM rocker/rstudio-stable:devel

# Set working directory
WORKDIR ${HOME}

# Copy directory files to image 
COPY . ${HOME}

# Copy repo into ${HOME}, make user own $HOME
USER root

# Install base utilities
RUN apt-get update && \
    apt-get install -y wget && \
    apt-get clean && \
    rm -rf /var/lib/apt/lists/*

# Install miniconda
ENV CONDA_DIR /opt/conda
RUN wget --quiet https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh \
    -O ~/miniconda.sh && \
     /bin/bash ~/miniconda.sh -b -p /opt/conda # DIFF: removed " && \" from end of line
     

# Put conda in path so we can use conda activate
ENV PATH=$CONDA_DIR/bin:$PATH

# Install umamba
RUN conda install -y micromamba -c conda-forge

# Create a conda environment from the environment yml
ENV MAMBA_ROOT_PREFIX="/opt/conda" # DIFF: added this line
COPY --chown=$MAMBA_USER:$MAMBA_USER environment.yml /tmp/environment.yml
RUN micromamba install --prefix "$MAMBA_ROOT_PREFIX" --yes --file /tmp/environment.yml && \ # DIFF: added --prefix flag and argument
    micromamba  clean --all --yes

# Activate the conda environment
ARG MAMBA_DOCKERFILE_ACTIVATE=1 

RUN chown -R ${NB_USER} . ${HOME}
USER ${NB_USER}

# Settings required for conda+rstudio
ENV RSTUDIO_WHICH_R=${CONDAENV}/bin/R
ENV RETICULATE_PYTHON=${CONDAENV}/bin/python

RUN echo rsession-which-r=${RSTUDIO_WHICH_R} > /etc/rstudio/rserver.conf && \
    echo rsession-ld-library-path=${CONDAENV}/lib >> /etc/rstudio/rserver.conf && \
    echo "R_LIBS_USER=${CONDAENV}/lib/R/library" > /home/rstudio/.Renviron

## Run an install.R script, if it exists.
#RUN if [ -f /R/install.R ]; then R --quiet -f /R/install.R; fi

但我建议对 Dockerfile 进行更多实质性更改，以使 Dockerfile 更易于维护、减少构建时间并创建更小的镜像。此 Dockerfile 主要基于来自 mamba-org/micromamba-docker 对 adding micromamba to an existing Docker image 的指令（披露：我维护 mamba-org/micromamba-docker）。

# bring in the micromamba image so we can copy files from it
FROM mambaorg/micromamba:0.24.0 as micromamba

# This is the image we are going add micromaba to:
FROM rocker/rstudio-stable:devel

ARG MAMBA_USER=root
ARG MAMBA_USER_ID=0
ARG MAMBA_USER_GID=0
ENV MAMBA_USER=$MAMBA_USER
ENV MAMBA_ROOT_PREFIX="/opt/conda"
ENV MAMBA_EXE="/bin/micromamba"

COPY --from=micromamba "$MAMBA_EXE" "$MAMBA_EXE"
COPY --from=micromamba /usr/local/bin/_activate_current_env.sh /usr/local/bin/_activate_current_env.sh
COPY --from=micromamba /usr/local/bin/_dockerfile_shell.sh /usr/local/bin/_dockerfile_shell.sh
COPY --from=micromamba /usr/local/bin/_entrypoint.sh /usr/local/bin/_entrypoint.sh
COPY --from=micromamba /usr/local/bin/_activate_current_env.sh /usr/local/bin/_activate_current_env.sh
COPY --from=micromamba /usr/local/bin/_dockerfile_initialize_user_accounts.sh /usr/local/bin/_dockerfile_initialize_user_accounts.sh
COPY --from=micromamba /usr/local/bin/_dockerfile_setup_root_prefix.sh /usr/local/bin/_dockerfile_setup_root_prefix.sh

RUN /usr/local/bin/_dockerfile_initialize_user_accounts.sh && \
    /usr/local/bin/_dockerfile_setup_root_prefix.sh && \
    echo rsession-which-r=${RSTUDIO_WHICH_R} > /etc/rstudio/rserver.conf && \
    echo rsession-ld-library-path=${CONDAENV}/lib >> /etc/rstudio/rserver.conf && \
    echo "R_LIBS_USER=${CONDAENV}/lib/R/library" > /home/rstudio/.Renviron

SHELL ["/usr/local/bin/_dockerfile_shell.sh"]

ENTRYPOINT ["/usr/local/bin/_entrypoint.sh"]

# populate the "base" conda environment:
USER $MAMBA_USER
COPY --chown=$MAMBA_USER:$MAMBA_USER environment.yml /tmp/environment.yml
RUN micromamba install --yes --file /tmp/environment.yml && \
    micromamba  clean --all --yes

WORKDIR ${HOME}

# Copy directory files to image 
COPY --chown=$MAMBA_USER_ID:$MAMBA_USER_GID . ${HOME}

# Settings required for conda+rstudio
ENV RSTUDIO_WHICH_R=${CONDAENV}/bin/R
ENV RETICULATE_PYTHON=${CONDAENV}/bin/python

CMD ["/init"]