notebooks/components/example-notebook-servers/rstudio/Dockerfile

169 lines
6.6 KiB
Docker

#
# NOTE: Use the Makefiles to build this image correctly.
#
ARG BASE_IMG=<base>
FROM $BASE_IMG
ARG TARGETARCH
# args - software versions (python)
ARG MINIFORGE_VERSION=24.11.3-0
ARG PIP_VERSION=24.3.1
ARG PYTHON_VERSION=3.11.11
# args - software versions (R)
# - rstudio: https://dailies.rstudio.com/release/
# - r-base: https://cran.r-project.org/doc/manuals/r-release/NEWS.html
# - r-png: https://www.rforge.net/png/news.html
# - r-reticulate: https://github.com/rstudio/reticulate/releases
# - r-shiny: https://github.com/rstudio/shiny/releases
ARG RSTUDIO_VERSION=2024.12.1-563
ARG R_BASE_VERSION=4.4.2
ARG R_PNG_VERSION=0.1_8
ARG R_RETICULATE_VERSION=1.40.0
ARG R_SHINY_VERSION=1.10.0
USER root
# install - binary rstudio dependencies
RUN apt-get -yq update \
&& apt-get -yq install --no-install-recommends \
dpkg-sig \
libapparmor1 \
libc6 \
libclang-dev \
libedit2 \
libpq5 \
libssl-dev \
psmisc \
rrdtool \
sudo \
&& apt-get clean \
&& rm -rf /var/lib/apt/lists/*
# setup environment for conda
ENV CONDA_DIR /opt/conda
ENV PATH "${CONDA_DIR}/bin:${PATH}"
RUN mkdir -pv ${CONDA_DIR} \
&& chmod 2775 ${CONDA_DIR} \
&& echo ". /opt/conda/etc/profile.d/conda.sh" >> ${HOME}/.bashrc \
&& echo ". /opt/conda/etc/profile.d/conda.sh" >> /etc/profile \
&& echo "conda activate base" >> ${HOME}/.bashrc \
&& echo "conda activate base" >> /etc/profile \
&& chown -R ${NB_USER}:${NB_GID} ${CONDA_DIR} \
&& chown -R ${NB_USER}:${USERS_GID} ${HOME}
# setup environment for R
ENV R_HOME ${CONDA_DIR}/lib/R
# rstudio logs to a file by default, but we need to see them in the container logs
# NOTE: we leave log-level as the default (WARN) because it also affects the user-facing console
# and there are some confusing messages at INFO level that are not useful to the user
ENV RS_LOG_LEVEL WARN
ENV RS_LOGGER_TYPE stderr
# rstudio must run as a named user.
# we change the UID of 'jovyan' at runtime to match the UID of the container.
# this is important for environments like OpenShift where the container has a random UID.
RUN chmod g+w /etc/passwd
USER $NB_UID
# install - conda, pip, python
RUN case "${TARGETARCH}" in \
amd64) MINIFORGE_ARCH="x86_64" ;; \
arm64) MINIFORGE_ARCH="aarch64" ;; \
ppc64le) MINIFORGE_ARCH="ppc64le" ;; \
*) echo "Unsupported architecture: ${TARGETARCH}"; exit 1 ;; \
esac \
&& curl -fsSL "https://github.com/conda-forge/miniforge/releases/download/${MINIFORGE_VERSION}/Miniforge3-${MINIFORGE_VERSION}-Linux-${MINIFORGE_ARCH}.sh" -o /tmp/Miniforge3.sh \
&& curl -fsSL "https://github.com/conda-forge/miniforge/releases/download/${MINIFORGE_VERSION}/Miniforge3-${MINIFORGE_VERSION}-Linux-${MINIFORGE_ARCH}.sh.sha256" -o /tmp/Miniforge3.sh.sha256 \
&& echo "$(cat /tmp/Miniforge3.sh.sha256 | awk '{ print $1; }') /tmp/Miniforge3.sh" | sha256sum -c - \
&& rm /tmp/Miniforge3.sh.sha256 \
&& /bin/bash /tmp/Miniforge3.sh -b -f -p ${CONDA_DIR} \
&& rm /tmp/Miniforge3.sh \
&& conda config --system --set auto_update_conda false \
&& conda config --system --set show_channel_urls true \
&& echo "python ==${PYTHON_VERSION}" >> ${CONDA_DIR}/conda-meta/pinned \
&& conda install -y -q \
python=${PYTHON_VERSION} \
pip=${PIP_VERSION} \
&& conda update -y -q --all \
&& conda clean -a -f -y
# install - r packages
RUN echo "r-base ==${R_BASE_VERSION}" >> ${CONDA_DIR}/conda-meta/pinned \
&& conda install -y -q \
r-base=${R_BASE_VERSION} \
r-png=${R_PNG_VERSION} \
r-reticulate=${R_RETICULATE_VERSION} \
r-shiny=${R_SHINY_VERSION} \
&& conda clean -a -f -y
# set default CRAN repo to RSPM (it has pre-compiled R packages, increasing user install speed)
RUN echo 'options(repos=c(CRAN="https://packagemanager.posit.co/all/__linux__/jammy/latest"))' >> ${R_HOME}/etc/Rprofile.site \
&& echo 'options(HTTPUserAgent=sprintf("R/%s R (%s)", getRversion(), paste(getRversion(), R.version["platform"], R.version["arch"], R.version["os"])))' >> ${R_HOME}/etc/Rprofile.site
# R needs TZ set
ENV TZ Etc/UTC
RUN echo "TZ=${TZ}" >> ${R_HOME}/etc/Renviron.site
USER root
# install - rstudio-server
RUN case "${TARGETARCH}" in \
amd64) RSTUDIO_DEB_URL="https://s3.amazonaws.com/rstudio-ide-build/server/jammy/amd64/rstudio-server-${RSTUDIO_VERSION}-amd64.deb" ;; \
arm64) RSTUDIO_DEB_URL="https://s3.amazonaws.com/rstudio-ide-build/server/jammy/arm64/rstudio-server-${RSTUDIO_VERSION}-arm64.deb" ;; \
*) echo "Unsupported architecture: ${TARGETARCH}"; exit 1 ;; \
esac \
&& curl -fsSL "${RSTUDIO_DEB_URL}" -o /tmp/rstudio-server.deb \
# validate the build signature
&& export GNUPGHOME="$(mktemp -d)" \
&& gpg --keyserver keys.gnupg.net --keyserver pgp.surfnet.nl --recv-keys 3F32EE77E331692F \
&& gpg --keyserver keys.openpgp.org --recv-keys 51C0B5BB19F92D60 \
&& dpkg-sig --verify /tmp/rstudio-server.deb \
&& rm -rf "${GNUPGHOME}" \
# install rstudio-server
&& dpkg -i /tmp/rstudio-server.deb \
&& rm -f /tmp/rstudio-server.deb \
# the default DB path '/var/lib/rstudio-server' causes permission issues when running as a random UID (like in OpenShift)
&& echo "provider=sqlite" > /etc/rstudio/database.conf \
&& echo "directory=/tmp/rstudio_db" >> /etc/rstudio/database.conf \
# use advisory file-locks to improve PVC support
&& echo "lock-type=advisory" > /etc/rstudio/file-locks \
# allow kubeflow to display rstudio in an iframe
&& echo "www-frame-origin=same" >> /etc/rstudio/rserver.conf \
# allows the non-root NB_USER to run rstudio
&& chown -R ${NB_USER}:${NB_GID} ${R_HOME} \
&& chown -R ${NB_USER}:${NB_GID} /etc/rstudio \
&& chown -R ${NB_USER}:${NB_GID} /run/rstudio-server* \
&& chown -R ${NB_USER}:${NB_GID} /usr/lib/rstudio-server \
&& chown -R ${NB_USER}:${NB_GID} /var/lib/rstudio-server \
&& chown -R ${NB_USER}:${NB_GID} /var/log/rstudio \
# give group same access as user (needed for OpenShift)
&& chmod -R g=u ${R_HOME} \
&& chmod -R g=u /etc/rstudio \
&& chmod -R g=u /run/rstudio-server* \
&& chmod -R g=u /usr/lib/rstudio-server \
&& chmod -R g=u /var/lib/rstudio-server \
&& chmod -R g=u /var/log/rstudio
# tell rstudio to use conda python by setting `RETICULATE_PYTHON` with `--rsession-path=/opt/rsession.sh`
COPY --chown=${NB_USER}:${NB_GID} --chmod=755 rsession.sh /opt
RUN chmod +x /opt/rsession.sh
# s6 - copy scripts
COPY --chown=${NB_USER}:${NB_GID} --chmod=755 s6/ /etc
# s6 - 01-copy-tmp-home
# NOTE: the contents of $HOME_TMP are copied to $HOME at runtime
# this is a workaround because a PVC will be mounted at $HOME
# and the contents of $HOME will be hidden
RUN cp -p -r -T "${HOME}" "${HOME_TMP}" \
# give group same access as user (needed for OpenShift)
&& chmod -R g=u "${HOME_TMP}"
USER $NB_UID
EXPOSE 8888