Skip to content

Latest commit

 

History

History
471 lines (346 loc) · 17.3 KB

README.md

File metadata and controls

471 lines (346 loc) · 17.3 KB

Field Papers

Quick links

Development

Setting environment variables

The following is required to run Field Papers, whether locally or via Docker / docker-compose:

cp sample.env .env
# provide some AWS credentials, etc.
open -t .env

In the opened text editor, add variables per Environment Variables. Contact another Field Papers contributor for any required values not present in sample.env.

Using docker-compose

compose is a Docker-based tool for orchestrating development environments. Rather than using foreman to manage multiple processes locally, compose runs each component process in a separate container, built up from local Dockerfiles or from remote repositories.

Prerequisites

Starting

This will fetch and build images as appropriate, logging STDOUT from all containers.

docker-compose build

If this is the first time you're running this, you'll need to load the schema into MySQL:

docker-compose run web rake db:schema:load

If you have pending migrations, run:

docker-compose run web rake db:migrate

To start the stack, logging STDOUT from all containers:

docker-compose up

The app will now be running on port 3000 on the Docker host. If you're lucky, it will be available at docker.local:3000, otherwise you'll need to determine the IP of your Docker host (localhost on Linux) and use that in place of docker.local.

If docker.local doesn't work, you'll need to update docker-compose.yml to set TILE_BASE_URL (in the environment section of web) to reflect your Docker host's IP. In my case, it's 192.168.64.6. You should be able to determine appropriate values using:

docker-compose port web 8080

Note, adding system dependencies to the Dockerfile requires you to docker-compose build in order to recreate the base web image. If you've just made changes to Gemfile, run docker-compose run web bundle.

Some helpful docker and docker-compose commands to know:

  1. To get a list of docker images ( and versions ) that the containers are running:

    $ docker images
    REPOSITORY                  TAG                 IMAGE ID            CREATED             VIRTUAL SIZE
    fpweb_web                   latest              e2cafa474299        58 minutes ago      940.4 MB
    quay.io/fieldpapers/tiler   v0.2.0              b2683b4d606d        3 days ago          855.9 MB
    mysql                       latest              c607d9b50dfa        13 days ago         374.1 MB
    ruby                        2.2.4               9168c99105ac        2 weeks ago         719.3 MB
    quay.io/fieldpapers/tasks   v0.10.2             f637d9257755        8 weeks ago         843 MB
  2. To see a list of the containers and their state that are running under docker-compose:

    $ docker-compose ps
        Name                   Command               State                       Ports
    -------------------------------------------------------------------------------------------------------
    fpweb_db_1      docker-entrypoint.sh mysqld      Up      3306/tcp
    fpweb_tasks_1   /bin/sh -c npm start             Up
    fpweb_tiler_1   /bin/sh -c npm start             Up
    fpweb_web_1     /bin/sh -c rm -f tmp/pids/ ...   Up      0.0.0.0:3000->3000/tcp, 0.0.0.0:8080->8080/tcp

Running Locally

Given the potential complexity of the above, or the need to make changes to the peripheral services, it may make more sense to run the application locally (you can still use docker-compose to run supplementary services like MySQL, etc.). If not using docker-compose, be sure MySQL is installed ($``brew install mysql if not) and running ($``mysql.server start if not).

On OS X, you'll want to use rbenv (and ruby-build) in order to isolate the version of Ruby used here (and to prevent it from conflicting with other projects). bundler is similarly used to localize gem dependencies.

direnv is a handy way to set project-specific environment variables (such as PATH or DATABASE_URL). A default .envrc has been provided that adds bin/ to your PATH ($(pwd)/bin, technically, to prevent abuse) so that bundler binstubs can be used. It's opt-in, so you'll need to enable it with direnv allow ..

Ghostscript is used to merge atlas pages together into a single PDF, so you'll need that (and boot2docker generate individual pages) to generate atlases.

OS X

brew install rbenv ruby-build direnv ghostscript boot2docker

boot2docker init         # create the Docker host if necessary
boot2docker up           # start the Docker host
$(boot2docker shellinit) # set the necessary Docker environment vars

eval "$(rbenv init -)"     # initialize rbenv
eval "$(direnv hook bash)" # initialize direnv
rbenv install $(< .ruby-version) # install the desired ruby version

gem install bundler        # install bundler using rbenv-installed ruby

xcode-select --install     # install Xcode command line utilities

# on 10.11, openssl headers aren't easily findable
bundle config build.eventmachine --with-opt-dir=/usr/local/opt/openssl

bundle install -j4 --path vendor/bundle # install dependencies

direnv allow .             # whitelist the local .envrc

echo $DATABASE_URL         # ensure that your environment is prepared

rake db:create             # create a database if one doesn't already exist
rake db:schema:load        # initialize your database

rails server -b 0.0.0.0 # start the app, listening on all interfaces

Ubuntu

Install Docker.

sudo apt-get install ghostscript git-core curl zlib1g-dev \
  build-essential libssl-dev libreadline-dev libyaml-dev \
  libsqlite3-dev sqlite3 libxml2-dev libxslt1-dev libcurl4-openssl-dev \
  python-software-properties libffi-dev

# install rbenv + ruby-build
git clone git://github.com/sstephenson/rbenv.git ~/.rbenv
echo 'export PATH="$HOME/.rbenv/bin:$PATH"' >> ~/.bash_profile
echo 'eval "$(rbenv init -)"' >> ~/.bash_profile

git clone git://github.com/sstephenson/ruby-build.git ~/.rbenv/plugins/ruby-build
echo 'export PATH="$HOME/.rbenv/plugins/ruby-build/bin:$PATH"' >> ~/.bash_profile
source ~/.bash_profile

eval "$(direnv hook bash)" # initialize direnv
rbenv install $(< .ruby-version) # install the desired ruby version

gem install bundler        # install bundler using rbenv-installed ruby

bundle install -j4 --path vendor/bundle # install dependencies

cp sample.env .env
sensible-editor .env

bundle exec foreman run echo $DATABASE_URL # ensure that your environment is prepared

rake db:create             # create a database if one doesn't already exist
rake db:schema:load        # initialize your database

bundle exec foreman run rails server -b 0.0.0.0 # start the app, listening on all interfaces

The app will now be running on localhost:3000 and will also be available as <you>.local (which is what should be used for BASE_URL).

You'll probably want to add the following to the end of your .bash_profile (or equivalent):

if which rbenv > /dev/null; then eval "$(rbenv init -)"; fi`
eval "$(direnv hook bash)"

If you choose not to use direnv, you'll need to ensure that the contents of .env are exported in your environment.

foreman is an alternative, in which case you'll prefix all commands with foreman run <cmd> in order to expose environment variables to them.

Barring that, export <VAR>=<VAL> for each pair in each shell instance you're using.

When updating, the following should be sufficient to sync your working copy:

bundle
rake db:migrate RAILS_ENV=development

There are probably additional Homebrew dependencies I'm missing because they were already installed.

NOTE: If you later decide to use fig, you'll need to delete vendor/bundle first.

Environment Variables

If using direnv or foreman, add these to .env. Otherwise, ensure that they are available to the environment in which Rails is running.

  • DATABASE_URL - development database URL. Probably similar to mysql2://root@localhost/fieldpapers_development
  • TEST_DATABASE_URL - test database URL.
  • RDS_DB_NAME - production database name.
  • RDS_HOSTNAME - production database hostname.
  • RDS_PASSWORD - production database password.
  • RDS_PORT - production database port.
  • RDS_USERNAME - production database username.
  • MAIL_ORIGIN - From address to use for automated system emails.
  • MAIL_SOURCE_ARN - AWS SES mail source identity. (Associated credentials must be granted access to send from this)
  • BASE_URL - Site base URL (Network-accessible, i.e. from a Docker container).
  • S3_BUCKET_NAME - S3 bucket for file storage. Required. Some endpoints might be dev.files.fieldpapers.org (development), test.files.fieldpapers.org (test), and files.fieldpapers.org (production).
  • AWS_ACCESS_KEY_ID - AWS key with read/write access to the configured S3 bucket(s).
  • AWS_SECRET_ACCESS_KEY - Corresponding secret.
  • AWS_REGION - AWS region to use for services.
  • BASE_URL - Base URL, e.g. http://fieldpapers.org.
  • TASK_BASE_URL - Base URL for the task server (probably an instance of fp-tasks).
  • TILE_BASE_URL - Base URL for the snapshot tiler (probably an instance of fp-tiler).
  • SENTRY_DSN - Sentry DSN for exception logging. Optional.
  • MAPZEN_SEARCH_KEY - A Mapzen Search API key, obtained from mapzen.com/developers.
  • STATIC_PATH - Path to write static files to. Must be HTTP-accessible. Defaults to ./public.
  • STATIC_URI_PREFIX - Prefix to apply to static paths (e.g. http://example.org/path) to allow them to resolve. Defaults to BASE_URL.
  • PERSIST - File persistence. Can be local or s3. Defaults to s3.
  • DEFAULT_CENTER - Default center for atlas composition (when a geocoder is unavailable). Expected to be in the form <zoom>/<latitude>/<longitude>. Optional.
  • ATLAS_COMPLETE_WEBHOOKS - A comma separated string of URLs. Optional. When an atlas moves to the state 'complete', fp-web will POST the JSON representation of the atlas to each URL.
  • ATLAS_INDEX_HEADER_TILELAYER - A Leaflet TileLayer urlTemplate. Optional. Providing this url will override the header basemaps for the atlas index pages. Defaults to http://tile.stamen.com/toner/{Z}/{X}/{Y}.png
  • DISABLE_LOGIN_CONFIRMATIONS - A value of true will not require users to confirm their accounts after registration and will not send confirmation emails. Optional. Defaults to false -- registration confirmations are required

Running Tests

rake

Alternately, you can use Guard to automatically run tests when related files change:

guard

Translation and Localization

To mark a string as one that should be localized, wrap it in _().

E.g. in an ERB template,

<% content_for :title, "Atlas - Field Papers" %>

becomes

<% content_for :title, _("Atlas - Field Papers") %>.

E.g. in Javascript within an ERB template,

window.alert("Hello Field Papers!")

becomes

window.alert(_('<%=escape_javascript _("Hello Field Papers!") %>')).

Install the Transifex client (tx):

# optionally create a virtualenv
virtualenv venv
source venv/bin/activate

# install Python dependencies
pip install -r requirements.txt

To extract strings from the app (and update pending translations):

rake gettext:find

To see the current translation status:

tx status

To push updated strings:

tx push -s

While it's possible to push updated translations, don't; Transifex is the source of truth for non-English strings.

To pull pending translations:

tx pull -af

To initialize a new language:

locale=es
mkdir -p locale/${locale}
cp locale/en/* locale/${locale}/
tx set -r fieldpapers.www -l ${locale} locale/${locale}/app.po

You'll also need to add the new locales to config/initializers/fast_gettext.rb and to the footer (app/views/shared/_footer.html.erb).

Heroku Deployment

Due to the presence of both Gemfile and requirements.txt, Heroku reports the ability to build this app using both the Ruby and Python buildpacks. The current buildpack detection order puts Ruby first, but explicit is better than implicit, so you can force the Ruby buildpack to be used:

heroku buildpack:set https://github.com/heroku/heroku-buildpack-ruby

Data

To bootstrap a database for development or on a new instance, run:

rake db:create db:schema:load

By default, it will create a fieldpapers_development (and fieldpapers_test) database on a local MySQL instance. To override this, set DATABASE_URL (in your environment, either directly or via .env), e.g.:

DATABASE_URL=mysql2://vagrant@somewhere/fieldpapers_development

To migrate an existing Field Papers database, first back it up. Then, set DATABASE_URL to point to it and run (with an appropriate RAILS_ENV if needed):

rake db:migrate

This will produce a database schema that is no longer compatible with the PHP version. Part of the migration involves cleaning up encoding errors (UTF-8 text stored as latin1 in UTF-8 columns)--your database may include some invalid characters, causing the migration to fail. To work-around that, identify the affected rows and clear their values before retrying the migration.

AWS Deployment

The Rails production environment is set up to allow a "quick start" deployment on Amazon Web Services using the aws-quick-start.py script in the fieldpapers/fieldpapers repository. See the documentation here details.

A couple of things to note about this production environment:

  • The database configuration (in config/database.yaml) is taken from a set of RDS_* environment variables which are set up automatically by AWS within the Docker container where the Rails web app runs. The AWS Relational Database Service (RDS) database is set up automatically by the aws-quick-start.py script.

  • Access to AWS resources (the S3 bucket used to store atlas pages and snapshots, the database, the SES mail service) is managed using AWS Identity and Access Management (IAM) roles, policies and instance profiles.

  • Obviously the relevant IAM roles, policies and instance profiles have to exist with the appropriate permissions. The easiest (and only recommended) way to do this is to use the aws-quick-start.py script to set everything up. It's kind of complicated and there are no guarantees that it will work if you try to do it by hand...

  • No AWS credentials appear anywhere in the code and no credentials are loaded from environment variables (such as AWS_ACCESS_KEY_ID or AWS_SECRET_ACCESS_KEY) when running on an EC2 instance; instead, temporary AWS credentials are made available by the infrastructure on the EC2 instance and are accessed via the instance metadata (the Ruby AWS SDK deals with this transparently).

  • There are some cases where extra authentication information is required to perform AWS actions from within an EC2 instance. In particular, a session token is needed to validate temporary AWS credentials (a case handled transparently by the AWS Ruby SDK), and a "source ARN" is required for sending email (which needs to be handled explicitly). This source ARN is needed to associate the EC2 instance with a mail identity policy so that the web app can send email using the AWS SES email service.

Extra environment variables

These are all set up by the aws-quick-start.py script, but are documented here for reference. They should not need to be set explicitly!

  • MAIL_ORIGIN - the originating email address used for sending account confirmation, password reset, etc. emails.
  • MAIL_SOURCE_ARN - AWS resource identifier used to associate EC2 instance with a mail identity policy, allowing email to be sent from within an EC2 instance using the AWS Simple Email Service (SES).