I have following code in my deploy.rb
namespace :app do
desc "copies the configuration frile from ~/shared/config/*.yml to ~/config"
task :copy_config_files,:roles => :app do
run "cp -fv #{deploy_to}/shared/config/hoptoad.rb #{release_path}/config/initializers"
run "cp -fv #{deploy_to}/shared/config/app_config.yml #{release_path}/config/app_config.yml"
end
end
I thought it would be a good idea to keep my deploy.rb file clean and I attempted to move above code to capistrano_utilities.rb under config. I am using Rails application. And I added following line of code to deploy.rb
require File.expand_path(File.dirname(__FILE__) + "/../lib/capistrano_utilities")
Now I am getting following error.
undefined method `namespace' for main:Object (NoMethodError)
The value of self in the deploy.rb is Capistrano::Configuration . While the value of self in capistrano_utilities is Main. So I understand why I am getting namespace method error. What is the fix for this problem?
In your config/deploy.rb, try load instead of require. Also, capistrano already runs as if you're at the RAILS_ROOT, so there's no need to use __FILE__:
load "lib/capistrano_utilities"
In a capistrano config file, load is redefined to load another configuration file into the current configuration. When passing a path to it, it actually calls load_from_file (a private method defined by capistrano) that just reads the file from disk and instance_eval's it.
Check your Capfile on Rails.root.
if you use capistrano 3, you see this line;
Dir.glob('lib/capistrano/tasks/*.cap').each { |r| import r }
Now, put your file on "lib/capistrano/tasks/capistrano_utilities.cap" and it will be loaded.
Related
I've just updated my Rails app to 5.2, and configured it to use the new config/credentials.yml.enc file.
When I try to deploy, I get this error:
NameError: uninitialized constant Rails
/Users/me/Documents/project/config/deploy.rb:27:in `<top (required)>'
That's pointing to this line in my config/deploy.rb file:
set :rollbar_token, Rails.application.credentials[:rollbar_token]
So it appears that while capistrano is running, it doesn't have access to Rails.application.credentials.
How are you all handling this? I've got some ideas...
Set this one variable as an ENV variable
I don't love how this separates/customizes this one setting
Somehow make it so capistrano has access to Rails.application.credentials
I don't know if this is a good idea or if there are other things I need to be aware of if I go this route
Remove deploy tracking in rollbar
🤷♂️
Put the following line(s) on top of your config/deploy.rb
# config/deploy.rb
require File.expand_path("./environment", __dir__)
This include make constants like Rails.application accessible in files like config/deploy/production.rb. Now things like the following are possible:
# config/deploy/staging.rb
server "production.lan", user: "production", roles: %w{app db web}
set :stage, :production
set :branch, "development"
set :pg_password, Rails.application.credentials[:staging][:postgres][:password]
I solved the problem as follows:
set :rollbar_token, YAML.load(`rails credentials:show`)['rollbar_token']
1. Upload master.key the file on the server (user read-only) like so:
namespace :setup do
desc "setup: copy config/master.key to shared/config"
task :copy_linked_master_key do
on roles(fetch(:setup_roles)) do
sudo :mkdir, "-pv", shared_path
upload! "config/master.key", "#{shared_path}/config/master.key"
sudo :chmod, "600", "#{shared_path}/config/master.key"
end
end
before "deploy:symlink:linked_files", "setup:copy_linked_master_key"
end
Put it in your lib/capistrano/tasks/setup.rake
2. Ensure file is linked
In deploy.rb:
set :linked_files, fetch(:linked_files, []).push("config/master.key")
3. Ensure Capfile loads the task:
Make sure your Capfile has the line
# Load custom tasks from `lib/capistrano/tasks` if you have any defined
Dir.glob("lib/capistrano/tasks/*.rake").each { |r| import r }
require File.expand_path("./environment", __dir__)
puts App::Application.credentials.rollbar_token
The way I solve this is to declare a $ROLLBAR_ACCESS_TOKEN environment variable on the server. I place it at the top of ~deployer/.bashrc like this:
export ROLLBAR_ACCESS_TOKEN=...
Then I integrate with Capistrano by defining this task:
task :set_rollbar_token do
on release_roles(:all).first do
set :rollbar_token, capture("echo $ROLLBAR_ACCESS_TOKEN").chomp
end
end
before "rollbar:deploy", "set_rollbar_token"
It seemed to me non-ideal to load the whole of Rails here, or to have to read command line output, so here's an alternative solution:
require "active_support/encrypted_configuration"
require "active_support/core_ext/hash/keys"
module CredentialLoader
def read_credentials(environment:)
YAML.load(
ActiveSupport::EncryptedConfiguration.new(
config_path: "config/credentials/#{environment}.yml.enc",
key_path: "config/credentials/#{environment}.key",
env_key: environment.to_s,
raise_if_missing_key: true
).read
)
end
end
Then you can do this in your deploy.rb:
include CredentialLoader
set :rollbar_token, read_credentials(environment: fetch(:stage))["rollbar_access_token"]
I have a capistrano task that is executed like this in deploy.rb:
after 'deploy:update_code', 'deploy:create_symlink'
The task:
Capistrano::Configuration.instance.load do
namespace :deploy do
task :create_symlink do
run "touch #{shared_path}/somefile.yml"
run "ln -nfs #{shared_path}/somefile.yml #{release_path}/config/somefile.yml"
end
end
end
configs are loaded from somefile.yml like this:
customconf = OpenStruct.new(YAML::load_file(File.join(Rails.root, 'config', 'somefile.yml'))[Rails.env]||{})
The issue that I'm having is that the configs are loaded in config/initializers/customconfig.rb. but the symlink seems to be created after the code in customconfig.rb is created.
This is the error I'm getting when trying to cap deploy:
Errno::ENOENT: No such file or directory - /var/www/vhosts/mysite/rails/releases/20170705083649/config/somefile.yml
Basically how can I load the configs from somefile.yml after the symlink is created. Or how can I run the cap task before the initializer is executed?
I solved my issue by changing after 'deploy:update_code' to after after 'deploy:finalize_update'. But I want to ask if this is the correct way of doing it before I accept this as accepted answer.
I'm trying to add the sandbox to my rails spree application and have run into this error
(using windows 8/powershell with Rails 4.1.6). I'm going by this manual: https://github.com/spree/spree/issues/411
This link Use older version of Rake
seems to have a similar issue but I am not sure how to take the necessary steps to achieve it.
When I try:
C:\Ruby193\openUpShop> bundle exec rake sandbox
I get:
rake aborted!
Don't know how to build task 'sandbox'
I'm am new to rails and am still not sure how everything works so a throughout explanation
with step by step instructions would be greatly appreciated! Thanks.
you can use a file sandbox.rb
# use example: rake task:sub_task -- --sandbox
if ARGV.any? {|arg| arg == '--sandbox' }
puts "** << USING SANDBOX!! >> **"
# beginning
$sandbox = -> do
ActiveRecord::Base.connection.begin_transaction
end
# end
at_exit do
ActiveRecord::Base.connection.rollback_transaction
end
end
then only you need add at the beginning of your task.rake file
require_relative 'path_to_your/sandbox.rb'
..and add at the beggining of your task code
desc "description task"
task example_task: :environment do
$sandbox.call if $sandbox.present?
...
I have no experience with Ruby or rake or anything, but I am using slate for API documentation, and it uses Ruby and rake and stuff to build the file. I know nothing at all about these things, but what I do know is this: when I do a rake build it updates a folder (slate/build). I then have to manually copy slate/build to ../app/docs after every single rake build. Is there something I can do that will copy that folder on every rake build automatically for me?
Add to your Rakefile:
ROOT = File.expand_path('..', __FILE__)
task :build_and_move => [:build] do
cp_r(File.join(ROOT, 'slate/build'), File.join(ROOT, '../app/docs'))
# or
# mv(File.join(ROOT, 'slate/build'), File.join(ROOT, '../app/docs'))
end
and then run rake build_and_move.
You can use FileUtils for this.
Docs: http://ruby-doc.org/stdlib-1.9.3/libdoc/fileutils/rdoc/FileUtils.html#method-c-copy
Example from the docs:
Copies src to dest. If src is a directory, this method copies all its contents recursively. If dest is a directory, copies src to dest/src.
FileUtils.cp 'eval.c', 'eval.c.org'
FileUtils.cp %w(cgi.rb complex.rb date.rb), '/usr/lib/ruby/1.6'
FileUtils.cp %w(cgi.rb complex.rb date.rb), '/usr/lib/ruby/1.6', :verbose => true
FileUtils.cp 'symlink', 'dest' # copy content, "dest" is not a symlink
I've got a plugin that is a bit heavy-weight. (Bullet, configured with Growl notifications.) I'd like to not enable it if I'm just running a rake task or a generator, since it's not useful in those situations. Is there any way to tell if that's the case?
It's as simple as that:
if $rails_rake_task
puts 'Guess what, I`m running from Rake'
else
puts 'No; this is not a Rake task'
end
Rails 4+
Instead of $rails_rake_task, use:
File.basename($0) == 'rake'
I like NickMervin's answer better, because it does not depend on the internal implementation of Rake (e.g. on Rake's global variable).
This is even better - no regexp needed
File.split($0).last == 'rake'
File.split() is needed, because somebody could start rake with it's full path, e.g.:
/usr/local/bin/rake taskname
$0 holds the current ruby program being run, so this should work:
$0 =~ /rake$/
It appears that running rake will define a global variable $rakefile, but in my case it gets set to nil; so you're better off just checking if $rakefile has been defined... seeing as __FILE__ and $FILENAME don't get defined to anything special.
$ cat test.rb
puts(global_variables.include? "$rakefile")
puts __FILE__
puts $FILENAME
$ cat Rakefile
task :default do
load 'test.rb'
end
$ ruby test.rb
false
test.rb
-
$ rake
(in /tmp)
true
./test.rb
-
Not sure about script/generator, though.
The most stable option is to add $is_rake = true at the beginning of Rakefile and use it from your code.
Use of $0 or $PROGRAM_NAME sometimes will fail, for example when using spring and checking variables from config/initializers
You can disable the plugin using environment variable:
$ DISABLE_BULLET= 1 rake some:task
And then in your code:
unless ENV['DISABLE_BULLET']
end
We could ask this
Rake.application.top_level_tasks
In a rails application, this is an empty array, whereas in a Rake task, the array has the task name in it.
top_level_tasks probably isn't a public API, so it's subject to changes. But this is the only thing I have found.