rajeshkp

namespace :log do
  desc "Downloads application logs to ./log/deploy/<RAILS_ENV>/<TIMESTAMP>/"
  task :fetch, :roles => :app do
      source = "#{shared_path}/log/#{rails_env}.log*"
      files_per_host = {}
      run "ls #{source}" do |channel, stream, data|
        files_per_host[channel[:host]] = data.split("\n")
      end

      destination = File.join('log', 'deploy')

rajeshkp

Ruby on Rails development setup on Ubuntu 12.04

System update

# change mirror to ubuntu.osuosl.org first
sudo apt-get update

Install common libraries

sudo apt-get install build-essential libreadline-dev libssl-dev zlib1g-dev libxml2-dev libxslt-dev

rajeshkp

Setting up Flume NG, listening to syslog over UDP, with an S3 Sink

My goal was to set up Flume on my web instances, and write all events into s3, so I could easily use other tools like Amazon Elastic Map Reduce, and Amazon Red Shift.

I didn't want to have to deal with log rotation myself, so I setup Flume to read from a syslog UDP source. In this case, Flume NG acts as a syslog server, so as long as Flume is running, my web application can simply write to it in syslog format on the specified port. Most languages have plugins for this.

At the time of this writing, I've been able to get Flume NG up and running on 3 ec2 instances, and all writing to the same bucket.

Install Flume NG on instances

rajeshkp

set :application, "appname"
set :deploy_to, "/var/www"

set :scm, :git
set :repository, "[email protected]:user/app.git"

default_run_options[:pty] = true
set :user, "www-data"
set :domain, "foo.tld"
set :normalize_asset_timestamps, false

rajeshkp

# config/deploy.rb

# probably a lot of ways to improve this...
set :application, 'my_app'
set :repo_url, '[email protected]:USERNAME/my_app.git'
# should set up a deploy user
set :user, 'deploy'

set :deploy_to, '/var/www/my_app'
set :scm, :git