Searchkick + ElasticSearch in Production: Faraday::ConnectionFailed: Connection refused - connect(2) - ruby

Question

I have an application in production with ElasticSearch. (It works) but now I can add searchkick gem instead use only elasticsearch.
For this I added the following:
In my Gemfile added:
gem 'elasticsearch-model'
gem 'elasticsearch-rails'
gem 'searchkick'
In my config/initializers/elasticsearch.rb (I'm using Amazon Elasticsearch Service for production. Ok I know this is bad practices but I'll go change this for Environment variables)
if Rails.env == "production"
Elasticsearch::Model.client = Elasticsearch::Client.new url: 'https://xxxxxxxx.xxxxxxx.amazonaws.com/'
else
Elasticsearch::Model.client = Elasticsearch::Client.new url: 'http://localhost:9200/'
end
in models/product.rb
require 'elasticsearch/model'
class Product < ActiveRecord::Base
include Elasticsearch::Model
include Elasticsearch::Model::Callbacks
searchkick
#currently use this function for search. (It works)
scope :searching, ->(query) { __elasticsearch__.search(query).records }
end
but now when I try to use in the server:
Product.search("foobar") #before run: rails c production
Show me the next message: Faraday::ConnectionFailed: Connection refused - connect(2)
After that I try this
rake searchkick:reindex:all RAILS_ENV=production
And I got the same message. Any idea for this ?

Answer 1

After I finished write this post I read this: Searchkick with ElasticSearch returns "Faraday::ConnectionFailed: execution expired"
in config/initializers/elasticsearch.rb
if Rails.env == "production"
url = 'http://myelasticsearch-xyz-foobar.amazonaws.com'
Elasticsearch::Model.client = Elasticsearch::Client.new url: url
Searchkick.client = Elasticsearch::Client.new(hosts: url, retry_on_failure: true, transport_options: {request: {timeout: 250}})
else
url = 'http://localhost:9200/'
Elasticsearch::Model.client = Elasticsearch::Client.new url: url
Searchkick.client = Elasticsearch::Client.new(hosts: url, retry_on_failure: true, transport_options: {request: {timeout: 250}})
end
After refactoring and clean this code I got this:
Elasticsearch::Model.client = Elasticsearch::Client.new url: env['RAILS_ENV']
Searchkick.client = Elasticsearch::Client.new(hosts: env['RAILS_ENV'], retry_on_failure: true, transport_options: {request: {timeout: 250} })
I hope this explanation works for others

Answer 2

I found the above answer to get me 95% of the way there. I followed Nelson's advice and created an Elasticsearch initializer which eliminated the Faraday::ConnectionFailed (Connection refused - connect(2) for "localhost" port 9200):
However I still had to reindex and migrate my Heroku DB.
heroku run rake searchkick:reindex CLASS=YOUR CLASS NAME
then heroku run rake db:migrate
Hope that clarifies for any Heroku newbies such as myself who got stuck on the last part.

Answer 3

If you are using gem 'searchkick' just add the below line in /config/initializers/elasticsearch.rb
ENV["ELASTICSEARCH_URL"] = "http://myelasticsearch-xyz-foobar.amazonaws.com:port"