I'm running some issues with docker/docker-compose and bundler. After building my image I can run the rails server correctly without any issue, but, when I try to run a console with rails console I constantly get:
Could not find i18n-0.7.0 in any of the sources
Run `bundle install` to install missing gems.
If I try to run bundle install in the container there is no problem, all seems to be correctly installed.
docker-compose run web bundle install
...
Using spring 1.3.6
Using therubyracer 0.12.2
Using turbolinks 2.5.3
Using uglifier 2.7.2
Using web-console 2.2.1
Updating files in vendor/cache
Bundle complete! 24 Gemfile dependencies, 119 gems now installed.
Bundled gems are installed into /usr/local/bundle.
My docker-compose.yml looks like this:
db:
image: postgres
web:
build: .
command: rails s -p 3000 -b 0.0.0.0
ports:
- "3000:3000"
volumes:
- .:/app
- ./github_rsa:/root/.ssh/id_rsa
links:
- db
I need to mount a volume with the ssh key because there are some gems that need to be pulled from private repositories.
My Dockerfile looks like this:
FROM ruby:2.2.0
RUN apt-get update -qq && apt-get install -y build-essential \
libpq-dev \
libxml2-dev libxslt1-dev \
nodejs
ENV HOME /root
ENV APP_HOME /app
RUN mkdir $APP_HOME
RUN mkdir $APP_HOME/tmp
RUN mkdir $APP_HOME/log
# Copy the Gemfile and Gemfile.lock into the image.
# Temporarily set the working directory to where they are.
WORKDIR /tmp
ADD Gemfile Gemfile
ADD Gemfile.lock Gemfile.lock
# Copy the github key for pulling gems
COPY github_rsa /root/.ssh/id_rsa
RUN \
chown -R root:root /root/.ssh && \
chmod 700 $HOME/.ssh && \
chmod 600 $HOME/.ssh/id_rsa
RUN echo "Host github.com\n\tStrictHostKeyChecking no\n" >> /root/.ssh/config
# Start ssh agent and add keys
RUN eval "$(ssh-agent)" && \
ssh-add && \
ssh-add -l
# Install bundler
RUN gem install bundler -v '~> 1.10'
# Install ruby dependencies
RUN bundle install
# Remove the ssh key now, we don't want to ship secrets on our images
RUN rm /root/.ssh/id_rsa
# Add app to container
ADD . $APP_HOME
# Add container working directory
WORKDIR $APP_HOME
# Expose puma port
EXPOSE 3000
You need to take this image down immediately. The following does not work:
RUN rm /root/.ssh/id_rsa
You cannot remove files like this; they will still exist in previous layers and be accessible to anyone who has the image. At the moment, you are shipping your secrets.
Regarding the actual question, I suspect it's just to do with the working directory and paths. Try moving RUN bundle install to after the WORKDIR instruction.
Related
I am having issues deploying a Ruby on Rails App to ECS Fargate. When I build the image locally (the same way it is done in the pipeline). I can easily start the web service with this command ["bundle", "exec", "unicorn", "-c", "config/unicorn.rb"]. However, running this same image in Fargate returns this: 2022-07-27 15:46:33bundler: failed to load command: unicorn (/usr/local/bundle/bin/unicorn)
Here is my Docker file:
FROM ruby:2.6.7
RUN apt-get update -qq && apt-get install -y nodejs postgresql-client sudo
WORKDIR /app
ENV RAILS_ENV="staging"
ENV NODE_ENV="staging"
ENV LANG="en_US.UTF-8"
ENV RACK_ENV ="staging"
ENV BUNDLE_WITHOUT='development:test'
ARG GITHUB_TOKEN
RUN gem install bundler -v '2.2.28'
RUN bundle config https://github.com/somename/somerepo someuser:"${GITHUB_TOKEN}"
COPY . /app
RUN rm -rf /app/tmp
RUN mkdir -p /app/tmp
RUN bundle install
RUN bundle exec rails assets:precompile
EXPOSE 3000
The CMD is missing from Docker because its'a added in the task definition. Has anyone run into this issue? I've tried a number of different approaches, but am unsure of what is being changed locally running and running in Fargate.
Update
Looking into this issue further and found some more information that will need to be updated here.
I am using CircleCI to build and push this image to ECR. The issue seems to be that when created in CircleCI, any artifacts created by the bundle install become unreachable on run time and Docker is unable to run any gems because the GEM path is not even accessible to the root user. I pulled the image created by CircleCI Docker, locally, and confirmed the same errors. EXECing into the container and running chmod 755 -R /usr/local/bundle/bin/ and then executing the bundle exec to start the service, properly starts unicorn.
Next Steps
As a result, I attempted to add those changes into the Dockerfile on build and the same behavior still persists.
RUN bundle install
RUN chmod 755 -R /usr/local/bundle/bin/
Then I tried changing permissions in an entrypoint script and the container won't start at all.
Finally figured this out a few days ago. The answer is to add VOLUME arguments at the end of your Dockerfile. This will maintain persistence with any changes you have made. My final Dockerfile:
FROM ruby:2.6.7
ARG NPM_TOKEN
RUN apt-get update -qq && apt-get install -y \
build-essential \
curl \
postgresql-client \
software-properties-common \
sudo
RUN curl -fsSL https://deb.nodesource.com/setup_9.x | sudo -E bash - && \
sudo apt-get install -y nodejs
RUN apt-get install -y \
npm \
yarn
ENV BUNDLE_WITHOUT='development:test'
ENV RAILS_ENV="staging"
ENV RACK_ENV="staging"
ENV NPM_TOKEN="${NPM_TOKEN}"
RUN mkdir -p /dashboard
WORKDIR /dashboard
RUN mkdir -p /dashboard/tmp/pids
COPY Gemfile* ./
COPY gems/rails_admin_history_rollback /dashboard/gems/rails_admin_history_rollback
ARG GITHUB_TOKEN
RUN sudo gem install bundler -v '2.2.28' && \
bundle config https://github.com/some-company/some-repo some-name:"${GITHUB_TOKEN}"
RUN bundle install
COPY . /dashboard
RUN bundle exec rails assets:precompile
RUN chmod +x bin/entrypoint.sh
VOLUME /dashboard/
VOLUME /usr/local/bundle
EXPOSE 3000
I've been trying to build my Docker image:
docker build -t <tag> -f Dockerfile.production .
However, this hangs while bundling.
I have tried bundling with:
DEBUG_RESOLVER=true bundle install --verbose
Running bundle on my host machine works fine - only the Docker image has this problem.
Attached is my Dockerfile:
FROM cimg/ruby:2.7.4-node
LABEL maintainer=budgeneration#gmail.com
SHELL ["/bin/bash", "-c"]
USER root
RUN sudo apt-get update && \
apt-get install -y nodejs npm libvips-tools libsodium-dev \
apt-transport-https ca-certificates curl software-properties-common \
librocksdb-dev \
libsnappy-dev \
python3-distutils \
rsyslog --no-install-recommends
# Other tools not related to building by still required.
RUN sudo apt-get install -y ffmpeg gifsicle
USER circleci
# Install all gems first.
# This hits the warm cache if unchanged so bundling is faster.
COPY Gemfile* /tmp/
WORKDIR /tmp
RUN gem install sassc-rails -v 2.1.2
RUN gem install bulma-rails -v 0.9.1
RUN bundle config set without 'development test' \
&& bundle install --verbose \
&& bundle binstubs railties
# Remove yarn (the other yarn)
RUN apt-get purge cmdtest
RUN yarn global add mjml
WORKDIR /sapco
# First we copy just Yarn files, to run yarn install
COPY package.json /sapco
COPY yarn.lock /sapco
RUN yarn install
WORKDIR /sapco
# Now copy everything
COPY . /sapco
EXPOSE 3000
Any tips to try to debug this further?
I have managed to resolve this issue but not sure why:
I changed my base image file to FROM circleci/ruby:2.7.4-buster-node and the bundling step continues fine.
scenario1
So the problem I am facing like after building the docker images. If I am going inside the docker container and doing GEM_PATH=$GEM_HOME gem list it is not showing all the gems installed at the time of building the image. it is showing only a few gems.
scenario2
If I am not sourcing the RVM path inside the Dockerfile. Then If I am going inside the docker container and doing GEM_PATH=$GEM_HOME gem list then it is showing all the gems.
Can anyone please explain to me why this is happening and how I should install and source RVM? So I can see all the gems inside the container. Basically, I want the scenario1 should work. Thanks
Below is my Dockerfile
FROM ruby:2.6.5
RUN apt-get update -qq && apt-get install -y sudo && apt-get install -y build-essential && apt-get install -y apt-utils && apt-get install -y git
RUN gpg --keyserver hkp://pool.sks-keyservers.net --recv-keys 409B6B1796C275462A1703113804BB82D39DC0E3 7D2BAF1CF37B13E2069D6956105BD0E739499BDB && \curl -sSL https://get.rvm.io | sudo bash -s stable --ruby && echo 'source /usr/local/rvm/scripts/rvm' >> /etc/bash.bashrc
RUN /bin/bash -l -c 'gem install bundler -v 1.17.3'
RUN /bin/bash -l -c 'gem install rails --version=5.2.4.1'
WORKDIR /app
RUN bundle config build.nokogiri --use-system-libraries
RUN bundle config set path 'vendor/bundle'
COPY Gemfile Gemfile.lock ./
ARG SSH_KEY
# Make ssh dir
RUN mkdir /root/.ssh/
# Copy ssh
RUN echo "$SSH_KEY" > /root/.ssh/id_rsa && \
chmod 0600 /root/.ssh/id_rsa
RUN touch /root/.ssh/known_hosts
# Add bitbuckets key
RUN ssh-keyscan bitbucket.org >> /root/.ssh/known_hosts
RUN bundle install
COPY . ./
EXPOSE 80:80
CMD rails s -p 80 -b '0.0.0.0' -e qa
TL;DR: don't use rvm\rbenv within a container
the whole thing when using containers, is to bundle\package all the dependencies inside a container. for instance, if you need ruby 2.9 then use a docker image which has ruby 2.9 installed rather than using rvm[rbenv](https://github.com/rbenv/rbenv).
since you are using bundler, i will advise you to add rails gem to it, and let bundler manage the dependencies.
i assume that you care about the directory where rubygems are being installed for caching propose, and i see that you instruct bundler exactly what is the rubygem home directory
RUN bundle config set path 'vendor/bundle'
I have a very old project I'm giving support. I'm using docker to make my builds and everything was working until this last one try.
This is my Dockerfile:
FROM ruby:2.1
# Install apt based dependencies required to run Rails as
# well as RubyGems. As the Ruby image itself is based on a
# Debian image, we use apt-get to install those.
RUN apt-get update && apt-get install -y \
git \
ghostscript \
libgs-dev \
build-essential \
libmysqlclient-dev \
locales \
nodejs
# Use en_US.UTF-8 as our locale
RUN locale-gen en_US.UTF-8
ENV LANG en_US.UTF-8
ENV LANGUAGE en_US:en
ENV LC_ALL en_US.UTF-8
# Configure the main working directory. This is the base
# directory used in any further RUN, COPY, and ENTRYPOINT
# commands.
RUN mkdir -p /app
WORKDIR /app
# Copy the Gemfile as well as the Gemfile.lock and install
# the RubyGems. This is a separate step so the dependencies
# will be cached unless changes to one of those two files
# are made.
COPY Gemfile Gemfile.lock ./
COPY vendor/oohlalog_gem ./vendor/oohlalog_gem
RUN gem install bundler && bundle install
# Copy the main application.
COPY . ./
# Expose port 3000 to the Docker host, so we can access it
# from the outside.
EXPOSE 3000
# Configure an entry point, so we don't need to specify
# "bundle exec" for each of our commands.
ENTRYPOINT ["bundle", "exec"]
# The main command to run when the container starts. Also
# tell the Rails dev server to bind to all interfaces by
# default.
CMD ["rails", "server", "-b", "0.0.0.0"]
When I try to run:
docker build . -t my_image
I'm getting:
Step 11 : RUN gem install bundler && bundle install
---> Running in 7f8e7442b423 ERROR: Error installing bundler:
bundler requires Ruby version >= 2.3.0.
I don't know how to fix this.
This is a result of a release of a new Bundler version 2.0.0 that happened on January 03, 2019. According to GitHub issue this is a desired behaviour.
As this is an old and not maintained anymore project the best option for you would be using older bundler.
You can modify the line from your Dockerfile:
RUN gem install bundler && bundle install
to the following:
RUN gem install bundler -v '$OLD_BUNDLER_VERSION' && bundle install
where $OLD_BUNDLER_VERSION should be set to last known working version, i.e. 1.16.1
Try a newer version of ruby image as your base. You are using 2.1 ( FROM ruby:2.1 ) but you need ruby version 2.3.0
My way is to use the correct bundler version and run it like this:
ENV BUNDLER_VERSION 2.2.26
RUN gem update --system \
&& gem install bundler -v $BUNDLER_VERSION \
&& bundle install -j 4
I am trying to use fig and docker to set up a local development environment for an existing rails app.
In the build process, I clearly see bundler installing the app's gems, but when I try to start the container with fig up or even reopen it with the /bin/bash command, the gems are not visible.
Here is the Dockerfile:
FROM ubuntu:14.04
# REPOS
RUN apt-get -qq update
RUN apt-get install -y software-properties-common
RUN add-apt-repository -y "deb http://archive.ubuntu.com/ubuntu $(lsb_release -sc) universe"
RUN add-apt-repository -y ppa:chris-lea/node.js
RUN apt-get -y update
#INSTALL
RUN apt-get install -y -q build-essential openssl libreadline6 libreadline6-dev curl git-core zlib1g zlib1g-dev libssl-dev libyaml-dev libsqlite3-dev sqlite3 libxml2-dev libxslt-dev autoconf libc6-dev ncurses-dev automake libtool bison pkg-config libpq-dev make wget unzip git vim nano nodejs gawk libgdbm-dev libffi-dev
#RUBY
RUN mkdir -p /download
WORKDIR download
RUN wget http://cache.ruby-lang.org/pub/ruby/2.1/ruby-2.1.2.tar.gz
RUN tar xvfz ruby-2.1.2.tar.gz
WORKDIR /download/ruby-2.1.2
RUN ./configure
RUN make
RUN make install
RUN gem update --system
RUN gem install bundler
RUN mkdir /rent
WORKDIR /rent
ADD Gemfile /rent/Gemfile
ADD Gemfile.lock /rent/Gemfile.lock
RUN bundle install --deployment
And here is the fig.yml file:
web:
build: .
command: bundle exec rails s -p 3000
volumes:
- .:/rent
ports:
- "3000:3000"
Running fig build clearly shows the app's gems being installed.
Running fig up fails with the message
bundler: command not found: rails
If I run fig run web /bin/bash and check the contents of
/local/lib/ruby/gems/2.1.0/gems
it only has bundler, rdoc, rake and a few others installed.
If I navigate to the app's directory and run the bundle command, it will install the app's gems and I can see that they are installed in the directory above. I can even start the app with rails server.
Why aren't the bundled gems being persisted in the image(container?).
I ran the rails tutuorial from the fig website and didn't have this problem.
Thanks
It seems that the problem was caused by using the --deployment flag with bundle install.
This flag's main effect is to deploy the gems to the vendor/bundle/ directory instead of the normal gem location. I checked and the gems were there, so I'm not sure why ruby couldn't find them.
Anyways, removing --deployment fixed the problem.
If you're using fig and mounted volumes, I found a solution which allows updates to the mounted files in development (like Gemfile.lock when you do fig run web bundle install) while still keeping container caching behavior.
See this for the full thing: https://gist.github.com/fotinakis/04077671bec4edf77c08
It's slightly convoluted, but basically you always install bundler and the gems as the non-root application user:
# Add 'web' user which will run the application.
RUN adduser web --home /home/web --shell /bin/bash --disabled-password --gecos ""
# Add directory where all application code will live and own it by the web user.
RUN mkdir /app
RUN chown -R web:web /app
# Install gems separately here to take advantage of container caching of `bundle install`.
# Also, ensure that gems are installed as the web user and not system-wide so that we can run
# `fig web bundle install` and the web user will have permissions to update the shared Gemfile.lock.
ADD Gemfile /app/
ADD Gemfile.lock /app/
RUN chown -R web:web /app
USER web
ENV HOME /home/web
ENV PATH $PATH:/home/web/.gem/ruby/2.1.0/bin
ENV GEM_HOME /home/web/.gem/ruby/2.1.0
ENV GEM_PATH $GEM_HOME
RUN gem install --user-install bundler
WORKDIR /app/
RUN bundle install
USER root
# Add the whole application source to the image and own it all by web:web.
# Note: this is overwritten in development because fig mounts a shared volume at /app.
ADD . /app/
RUN chown -R web:web /app
Running this will now work and update your local Gemfile.lock:
$ fig run web bundle install