maverick/docbrown
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
docbrown/Containerfile
Josh Klar e4d33b316c
devxp: Base the application container images off new Ruby base image. (#19633) 
This "rebases" our application images off of the new Ruby base image
added in #19632, and fixes numerous problems and quirks with how the
images were built along the way. Notably:

- Issues where layers attempted to delete files in prior layers have
  been resolved (this caused build failures on some Docker filesystem
  drivers, notably overlay2).

- Bundler is no longer allowed to deviate from or modify the lockfile
  (`BUNDLE_FROZEN` is now `true`).

- `git(1)` is no longer required to live inside the container and `.git`
  is no longer required to be copied into the Docker build context, as
  these were only used to calculate `FOREM_BUILD_SHA`, which is now
  passed in as a Build Argument to the container build context.

- The entire source tree is no longer `chmod` in one giant swing, which
  ran so long on my system (as just one example) that I gave up after
  15-20 minutes and issued it a `SIGTERM`. Instead, `COPY --chown` is
  used more heavily and ensures the `APP_USER` will have access to the
  requisite files.

This new container image appears to build successfully for
`linux/arm64`, which refs (but does not complete) #19626. Currently,
such builds aren't automated , and must be built on a developer
workstation. For example:

```sh
docker buildx build --platform linux/amd64,linux/arm64 -f Containerfile . -t ghcr.io/forem/forem:klardotsh-test --push --build-arg VCS_REF=$(git rev-parse --short HEAD)
```

In the meantime, the existing `linux/amd64`-only BuildKite scripts have
been updated to allow this PR to merge as a separate unit, and CI
refactors to enable the multiarch builds of `linux/arm64,linux/amd64`
can come later when more time is available.

This is one of several blockers on the path to getting #19603 merged.
The next step in that chronology will be rebasing that work on top of
this work, which *should* be, on the containerization side, as
straightforward as bumping `Containerfile.base` to reference the new
upstream image, rebuilding the base container, and then bumping the
reference in `Containerfile`.
2023-06-29 10:50:26 -07:00

165 lines
6.2 KiB
Docker
Raw Blame History

FROM ghcr.io/forem/ruby:3.0.2@sha256:04c945460e5999c0483b119074597dba62b863f5f5ea3c98bff4169e0d2f990b as base
FROM base as builder
# This is provided by BuildKit
ARG TARGETARCH
USER root
# pkg-config,
# libpixman-1-dev,
# libcairo2-dev,
# libpango1.0-dev
#
# are needed only on arm64: some nodejs dependency doesn't provide
# pre-built binaries for that arch, and so falls back to building
# from source, which then requires a few extra packages installed.
#
# Since we wipe out node_modules as part of this image after calling
# the bundler, we don't need these headers (or their sofile counterparts)
# in any of the other build stages.
RUN apt update && \
apt install -y \
build-essential \
libcurl4-openssl-dev \
libffi-dev \
libxml2-dev \
libxslt-dev \
libpcre3-dev \
libpq-dev \
pkg-config \
libpixman-1-dev \
libcairo2-dev \
libpango1.0-dev \
&& \
apt clean
ENV BUNDLER_VERSION=2.2.22 \
BUNDLE_SILENCE_ROOT_WARNING=true \
BUNDLE_SILENCE_DEPRECATIONS=true
RUN gem install -N bundler:"${BUNDLER_VERSION}"
ENV APP_USER=forem APP_UID=1000 APP_GID=1000 APP_HOME=/opt/apps/forem \
LD_PRELOAD=libjemalloc.so.2
RUN mkdir -p ${APP_HOME} && chown "${APP_UID}":"${APP_GID}" "${APP_HOME}" && \
groupadd -g "${APP_GID}" "${APP_USER}" && \
adduser --uid "${APP_UID}" --gid "${APP_GID}" --home "${APP_HOME}" "${APP_USER}"
ENV DOCKERIZE_VERSION=v0.7.0
RUN curl -fsSLO https://github.com/jwilder/dockerize/releases/download/"${DOCKERIZE_VERSION}"/dockerize-linux-${TARGETARCH}-"${DOCKERIZE_VERSION}".tar.gz \
&& tar -C /usr/local/bin -xzvf dockerize-linux-${TARGETARCH}-"${DOCKERIZE_VERSION}".tar.gz \
&& rm dockerize-linux-${TARGETARCH}-"${DOCKERIZE_VERSION}".tar.gz \
&& chown root:root /usr/local/bin/dockerize
USER "${APP_USER}"
WORKDIR "${APP_HOME}"
COPY --chown=${APP_UID}:${APP_GID} ./.ruby-version "${APP_HOME}"/
COPY --chown=${APP_UID}:${APP_GID} ./Gemfile ./Gemfile.lock "${APP_HOME}"/
COPY --chown=${APP_UID}:${APP_GID} ./vendor/cache "${APP_HOME}"/vendor/cache
# Have to reset APP_CONFIG, which appears to be set by upstream images, to
# avoid permission errors in the development/test images (which run bundle
# as a user and require write access to the config file for setting things
# like BUNDLE_WITHOUT (a value that is cached by root here in this builder
# layer, see https://michaelheap.com/bundler-ignoring-bundle-without/))
ENV BUNDLE_APP_CONFIG="${APP_HOME}/.bundle"
RUN mkdir -p "${BUNDLE_APP_CONFIG}" && \
touch "${BUNDLE_APP_CONFIG}/config" && \
chown -R "${APP_UID}:${APP_GID}" "${BUNDLE_APP_CONFIG}" && \
bundle config --local build.sassc --disable-march-tune-native && \
bundle config --local without development:test && \
BUNDLE_FROZEN=true bundle install --deployment --jobs 4 --retry 5 && \
find "${APP_HOME}"/vendor/bundle -name "*.c" -delete && \
find "${APP_HOME}"/vendor/bundle -name "*.o" -delete
COPY --chown=${APP_UID}:${APP_GID} . "${APP_HOME}"
RUN mkdir -p "${APP_HOME}"/public/{assets,images,packs,podcasts,uploads}
# While it's relatively rare for bare metal builds to hit the default
# timeout, QEMU-based ones (as is the case with Docker BuildX for
# cross-compiling) quite often can. This increased timeout should help
# reduce false-negatives when building multiarch images.
RUN echo 'network-timeout 300000' >> ~/.yarnrc
# This is one giant step now because previously, removing node_modules to save
# layer space was done in a later step, which is invalid in at least some
# Docker storage drivers (resulting in Directory Not Empty errors).
RUN NODE_ENV=production yarn install && \
RAILS_ENV=production NODE_ENV=production bundle exec rake assets:precompile && \
rm -rf node_modules
# This used to be calculated within the container build, but we then tried
# to rm -rf the .git that was copied in, which isn't valid (removing
# directories created in lower layers of an image isn't a thing (at least
# with the overlayfs drivers). Instead, we'll pass this in over CLI when
# building images (eg. in CI), but leave a default value for callers who don't
# override (perhaps docker-compose). This isn't perfect, but it'll do for now.
ARG VCS_REF=unspecified
RUN echo $(date -u +'%Y-%m-%dT%H:%M:%SZ') >> "${APP_HOME}"/FOREM_BUILD_DATE && \
echo "${VCS_REF}" >> "${APP_HOME}"/FOREM_BUILD_SHA
## Production
FROM base as production
USER root
ENV BUNDLER_VERSION=2.2.22 BUNDLE_SILENCE_ROOT_WARNING=1
RUN gem install -N bundler:"${BUNDLER_VERSION}"
ENV APP_USER=forem APP_UID=1000 APP_GID=1000 APP_HOME=/opt/apps/forem \
LD_PRELOAD=libjemalloc.so.2
RUN mkdir -p ${APP_HOME} && chown "${APP_UID}":"${APP_GID}" "${APP_HOME}" && \
groupadd -g "${APP_GID}" "${APP_USER}" && \
adduser --uid "${APP_UID}" --gid "${APP_GID}" --home "${APP_HOME}" "${APP_USER}"
COPY --from=builder --chown="${APP_USER}":"${APP_USER}" ${APP_HOME} ${APP_HOME}
USER "${APP_USER}"
WORKDIR "${APP_HOME}"
VOLUME "${APP_HOME}"/public/
ENTRYPOINT ["./scripts/entrypoint.sh"]
CMD ["bundle", "exec", "rails", "server", "-b", "0.0.0.0", "-p", "3000"]
## Testing
FROM builder AS testing
USER "${APP_USER}"
COPY --chown="${APP_USER}":"${APP_USER}" ./spec "${APP_HOME}"/spec
COPY --from=builder /usr/local/bin/dockerize /usr/local/bin/dockerize
RUN bundle config --local build.sassc --disable-march-tune-native && \
bundle config --delete without && \
BUNDLE_FROZEN=true bundle install --deployment --jobs 4 --retry 5 && \
find "${APP_HOME}"/vendor/bundle -name "*.c" -delete && \
find "${APP_HOME}"/vendor/bundle -name "*.o" -delete
ENTRYPOINT ["./scripts/entrypoint.sh"]
CMD ["bundle", "exec", "rails", "server", "-b", "0.0.0.0", "-p", "3000"]
## Development
FROM builder AS development
USER "${APP_USER}"
COPY --chown="${APP_USER}":"${APP_USER}" ./spec "${APP_HOME}"/spec
COPY --from=builder /usr/local/bin/dockerize /usr/local/bin/dockerize
RUN bundle config --local build.sassc --disable-march-tune-native && \
bundle config --delete without && \
BUNDLE_FROZEN=true bundle install --deployment --jobs 4 --retry 5 && \
find "${APP_HOME}"/vendor/bundle -name "*.c" -delete && \
find "${APP_HOME}"/vendor/bundle -name "*.o" -delete
ENTRYPOINT ["./scripts/entrypoint.sh"]
CMD ["bundle", "exec", "rails", "server", "-b", "0.0.0.0", "-p", "3000"]
Reference in a new issue View git blame Copy permalink