Use non-optimized DelayedJob reserve sql strategy

Fixes sql deadlocks when running many quick jobs with multiple workers

Deadlocks happened when DJ was trying to delete finished jobs from the database,
as a result those jobs where run a second time which created duplicate event.

Adresses huginn#977

Gemfile

@@ -59,7 +59,7 @@ gem 'bundler', '>= 1.5.0'
 gem 'coffee-rails', '~> 4.1.0'
 gem 'daemons', '~> 1.1.9'
 gem 'delayed_job', '~> 4.0.0'
-gem 'delayed_job_active_record', '~> 4.0.0'
+gem 'delayed_job_active_record', :git => 'https://github.com/cantino/delayed_job_active_record', :branch => 'configurable-reserve-sql-strategy'
 gem 'devise', '~> 3.4.0'
 gem 'dotenv-rails', '~> 2.0.1'
 gem 'em-http-request', '~> 1.1.2'

Gemfile.lock

@@ -28,6 +28,15 @@ GIT
       omniauth (~> 1.0)
       omniauth-oauth2 (~> 1.1)
 
+GIT
+  remote: https://github.com/cantino/delayed_job_active_record
+  revision: 7ede4ea81561a66427f600decb87f6fcdb5b8aa4
+  branch: configurable-reserve-sql-strategy
+  specs:
+    delayed_job_active_record (4.0.3)
+      activerecord (>= 3.0, < 5.0)
+      delayed_job (>= 3.0, < 4.1)
+
 GEM
   remote: https://rubygems.org/
   specs:
@@ -119,9 +128,6 @@ GEM
     debug_inspector (0.0.2)
     delayed_job (4.0.6)
       activesupport (>= 3.0, < 5.0)
-    delayed_job_active_record (4.0.3)
-      activerecord (>= 3.0, < 5.0)
-      delayed_job (>= 3.0, < 4.1)
     delorean (2.1.0)
       chronic
     devise (3.4.1)
@@ -525,7 +531,7 @@ DEPENDENCIES
   coveralls
   daemons (~> 1.1.9)
   delayed_job (~> 4.0.0)
-  delayed_job_active_record (~> 4.0.0)
+  delayed_job_active_record!
   delorean
   devise (~> 3.4.0)
   dotenv-rails (~> 2.0.1)

config/initializers/delayed_job.rb

@@ -14,3 +14,7 @@ class Delayed::Job
   scope :awaiting_retry, ->{ where("failed_at IS NULL AND attempts > 0") }
   scope :failed, -> { where("failed_at IS NOT NULL") }
 end
+
+Delayed::Backend::ActiveRecord.configure do |config|
+  config.reserve_sql_strategy = :default_sql
+end