Run selenium in the host with docker and Capybara - selenium

Question

I have tests with capybara in a docker container. I use this to setup selenium :
Capybara.register_driver :selenium do |app|
require 'selenium/webdriver'
Selenium::WebDriver::Firefox::Binary.path = ENV['FIREFOX_BINARY_PATH'] || Selenium::WebDriver::Firefox::Binary.path
Capybara::Selenium::Driver.new(app, :browser => :firefox)
end
It work when we run tests with xvfb but I want to see the real browser when tests are running so I'm looking for a way to use the browser in the host.
I think it's possible to launch geckodriver on the host and share the port 4444 but I didn't succeeded yet. Capybara launch a new instance of geckodriver, on the container, each time.
What can I do?
Edit 1 : Add more info
I all config I have for capybara :
#<Capybara::SessionConfig:0x0055ce67731a00
#always_include_port=false,
#app_host="http://domain-test.engagement.lvh.me:1300",
#automatic_label_click=false,
#automatic_reload=true,
#default_host="http://www.example.com",
#default_max_wait_time=5,
#default_selector=:css,
#enable_aria_label=false,
#exact=false,
#exact_text=false,
#ignore_hidden_elements=true,
#match=:smart,
#raise_server_errors=true,
#run_server=true,
#save_path=#<Pathname:/app/tmp/capybara>,
#server_errors=[StandardError],
#server_host=nil,
#server_port=1300,
#visible_text_only=false,
#wait_on_first_by_default=false>
Here is my docker-compose file :
version: '3'
services:
web:
build: .
command: rails s -b 0.0.0.0
working_dir: /app
volumes:
- .:/app
- ./tmp/bundle:/usr/local/bundle
- $SSH_AUTH_SOCK:/ssh-agent
environment:
- BUNDLE_JOBS=4
- SSH_AUTH_SOCK=/ssh-agent
- MONGO_HOST=mongo
- REDIS_HOST=redis
- MEMCACHE_HOST=memcache
ports:
- "80:3000"
- "1300:1300"
links:
- mongo
- redis
- memcache
mongo:
image: mongo:3.4.9
volumes:
- ~/data/mongo/db:/data/db
redis:
image: redis:2.8.17
volumes:
- ~/data/redis:/data
memcache:
image: memcached:1.5-alpine
And finally my Dockerfile :
FROM ruby:2.3.1
RUN apt-get update && apt-get install -y build-essential qt5-default \
libqt5webkit5-dev gstreamer1.0-plugins-base gstreamer1.0-tools gstreamer1.0-x \
xvfb rsync
ARG GECKODRIVER_VERSION=0.19.0
RUN wget --no-verbose -O /tmp/geckodriver.tar.gz https://github.com/mozilla/geckodriver/releases/download/v$GECKODRIVER_VERSION/geckodriver-v$GECKODRIVER_VERSION-linux64.tar.gz \
&& rm -rf /opt/geckodriver \
&& tar -C /opt -zxf /tmp/geckodriver.tar.gz \
&& rm /tmp/geckodriver.tar.gz \
&& mv /opt/geckodriver /opt/geckodriver-$GECKODRIVER_VERSION \
&& chmod 755 /opt/geckodriver-$GECKODRIVER_VERSION \
&& ln -fs /opt/geckodriver-$GECKODRIVER_VERSION /usr/bin/geckodriver
RUN apt-get install -y libgtk-3-dev \
&& wget --no-verbose https://ftp.mozilla.org/pub/firefox/releases/56.0/linux-x86_64/en-US/firefox-56.0.tar.bz2 \
&& tar -xjf firefox-56.0.tar.bz2 \
&& mv firefox /opt/firefox56 \
&& ln -s /opt/firefox56/firefox /usr/bin/firefox
ENV TZ Europe/Paris
RUN echo $TZ > /etc/timezone && \
apt-get update && apt-get install -y tzdata && \
dpkg-reconfigure -f noninteractive tzdata && \
apt-get clean
RUN apt-key adv --keyserver hkp://keyserver.ubuntu.com:80 --recv 0C49F3730359A14518585931BC711F9BA15703C6 && \
echo "deb http://repo.mongodb.org/apt/debian jessie/mongodb-org/3.4 main" | tee /etc/apt/sources.list.d/mongodb-org-3.4.list && \
apt-get update && \
apt-get install -y mongodb-org
RUN gem install bundler
RUN mkdir /app
WORKDIR /app

Answer 1

In order to get Selenium to use a remote geckodriver instance you need to provide the url option to it.
Capybara.register_driver :selenium do |app|
require 'selenium/webdriver'
Capybara::Selenium::Driver.new(app, :browser => :firefox, url: 'http://<your ip as reachable from docker>:<port geckodriver is available on>')
end
This will then require you to run geckodriver on the machine your want firefox to run on, possibly using the --binary option to specify where firefox is located. It will also probably require setting Capybara.app_host (and possibly Capybara.always_include_port depending on your exact configuration) so the browser requests are routed back to the app under test running on the docker instance.
Another thing to consider is that the AUT will need to be bound to an interface on the docker instance which is reachable from the host. By default Capybara binds to the 127.0.0.1 interface which probably isn't reachable, so you can set Capybara.server = '0.0.0.0' to bind to all available interfaces, or specify the specific external interface.