I've added an extra step to explicitly add the production.rb file.
Added S3 CDN instructions with asset sync gem. Users assets will auto sync with S3 on deploy, using Gziped files where available. This should significantly help with Heroku stability under load.
Added add-on instructions for some load testing services available through Heroku. Hopefully help us hack on performance and stability issues.
Some general houskeeping, fixed h1 h2 muddle and removed bundle exec from foreman commands, as it uses bundler by default, and added some commands for running tests.
I added a line to activate the user from the rails console, for the case when the mail server hasn't been configured yet and discourse can't send activation mails.
* ubuntu should be Ubuntu
* Ubuntu 12.04 should be Ubuntu Server 12.04 or Ubuntu Server 12.04 LTS (preferred by Ubuntu Community)
* Consistent capitalization of "Discourse"
* add screenshots from Ubuntu config process
* add instructions for updating predeployed VMs
* add details on configuring mail server
* don't install test gems into prod
* no need for bundle exec
* small cleanups
* added RVM option for single-user install
* clarified Web Server options
* bluepill sample config: create socket directory if it doesn't exist
* add multisite.yml.production-sample
* rename production.sample.rb (OCD)
* make two database.yml samples (one for dev, one for prod)
* fixed travis build (database.yml.sample renamed)
* fix other references to database.yml.sample
* clean up 'rvm requirements' instructions as per @mpapis
This helps if somebody would fork a GitHub repository, as the links
now lead to files in fork, not files in original repository. Also, it
makes more sense when reading Markdown files locally.
