Skip to content

Latest commit

 

History

History
113 lines (61 loc) · 3.4 KB

DEPLOYMENT.md

File metadata and controls

113 lines (61 loc) · 3.4 KB
Raw

Glazier deployment Overview

Deployment of Glazier is broken into distinct parts:

  1. Deploying the rails project to Heroku
  2. Uploading javascript and css assets to S3
  3. Updating the production database with the contents of index.html
  4. Ingesting card manifests

Dependencies

Notes:

  • heroku commands are run from within glazier-server
  • grunt commands are run from within glazier

Deploying the rails project to Heroku

One-time setup:

In glazier/glazier-server/:

git remote add prod git@heroku.com:glazier.git

Deploying changes

Once you have created the prod remote for heroku, you can deploy with:

git push prod master

This will push the latest server code to heroku. Run migrations with:

heroku run rake db:migrate

heroku-surrogate can be used to run migrations before deploying code, useful when updates aren't compatible with the unmigrated database:

# check current status of prod DB against local migrations folder
heroku surrogate rake db:migrate:status

# run pending local migrations against prod
heroku surrogate rake db:migrate

# deploy new server code that depends on latest migrations
git push prod master

Uploading javascript and css assets to S3

Glazier's approach for deploying the client-side application is to create fingerprinted javascript and md5 files, and upload them to s3.

Setup

Set the following environment variables with the s3 key and secret for the bucket you want to deploy to:

GLAZIER_S3_KEY
GLAZIER_S3_SECRET

In the package.json file, ensure that the correct assetHost value is specified. For glazier we are using "https://glazier.s3.amazonaws.com" as our asset host.

Deployment

Running grunt deploy will result in fingerprinted asset files being uploaded to s3.

It will also create a tmp/public/index.html file that uses the urls of those assets.

Updating the production database with the contents of index.html

Part of what grunt deploy does is upload the index.html that uses the assets that it uploads to s3.

It does this by running:

heroku surrogate rake 'glazier:ingest_as_current[../tmp/public/index.html]' --app glazier

Uploading assets of cards in glazier/cards to s3

before you deploy cards, you'll want to make sure you have the latest code for each card you are going to deploy

the runInCards grunt command lets you quickly run a command in every symlinked card folder

in glazier/

grunt runInCards --cmd="git status"
grunt runInCards --cmd="git pull --ff-only"

# once you are satisfied that all code is up to date

grunt deployCards

# you can also use the CARDS env variable to control which cards are deployed

CARDS=github-issues,github-stars grunt deployCards

will run grunt deploy in each folder within glazier/cards/, and then a heroku surrogate rails run command that ingests the manifests that were just uploaded.

Ingesting card manifests

This guide assumes you have deployed card assets, including a manifest.json file for each card you want to use.

Start the production console with heroku run console.

Then, with each manifest.json url, run:

PaneType.ingest(url)

You can also reingest all previously-ingested manifests with

PaneType.reingest_all